PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow,...

A Survey on Low Latency Towards 5G RAN CoreNetwork and Caching Solutions

Imtiaz Parvez Student Member IEEE Ali Rahmati Student Member IEEE Ismail Guvenc Senior Member IEEEArif I Sarwat Senior Member IEEE and Huaiyu Dai Fellow IEEE

AbstractmdashThe fifth generation (5G) wireless network technol-ogy is to be standardized by 2020 where main goals are toimprove capacity reliability and energy efficiency while reducinglatency and massively increasing connection density An integralpart of 5G is the capability to transmit touch perception typereal-time communication empowered by applicable robotics andhaptics equipment at the network edge In this regard weneed drastic changes in network architecture including core andradio access network (RAN) for achieving end-to-end latencyon the order of 1 ms In this paper we present a detailedsurvey on the emerging technologies to achieve low latencycommunications considering three different solution domainsRAN core network and caching We also present a generaloverview of 5G cellular networks composed of software definednetwork (SDN) network function virtualization (NFV) cachingand mobile edge computing (MEC) capable of meeting latencyand other 5G requirements

Index Termsmdash5G cloud caching haptic communicationslatency massive connectivity real-time communication SDNtactile Internet ultra-high reliability ultra-low latency

I INTRODUCTION

The focus of next generation mobile communication is toprovide seamless communication for machines and devicesbuilding the Internet-of-Things (IoT) along with personal com-munication New applications such as tactile Internet1 high-resolution video streaming tele-medicine tele-surgery smarttransportation and real-time control dictate new specificationsfor throughput reliability end-to-end (E2E) latency and net-work robustness [2] Additionally intermittent or always-ontype connectivity is required for machine-type communication(MTC) serving diverse applications including sensing andmonitoring autonomous cars smart homes moving robots andmanufacturing industries

Several emerging technologies including wearable devicesvirtualaugmented reality and full immersive experience (3D)are shaping the demeanor of human end users and they havespecial requirements for user satisfaction Therefore these usecases of the next generation network push the specificationsof 5G in multiple aspects such as data rate latency reliabil-ity devicenetwork energy efficiency traffic volume densitymobility and connection density Current fourth generation(4G) networks are not capable of fulfilling all the technicalrequirements for these services

Fifth generation (5G) cellular network is the wireless ac-cess solution to fulfill the wireless broadband communication

1A network or network of networks for remotely accessing perceivingmanipulating or controlling real or virtual objects or processes in perceivedreal time by humans or machines [1]

specifications of 2020 and beyond [3] [4] In ITU 5G ITU-R working group is working for development of 5G underthe term IMT-2020 [5] The vision of this work is to achieveone thousand times throughput improvement 100 billion con-nections and close to zero latency [2] [3] In particular 5Gwill support enhanced mobile broadband (MBB) with end-user data rates of 100 Mbps in the uniform spatial distribu-tion with peak data rates of 10-20 Gbps [3] [4] Based onconsensus 5G will not only provide personal mobile servicebut also massive machine type communications (MTC) andlatencyreliability critical services In mission critical com-munication (MCC)ultra reliable low latency communication(uRLLC2) both the latency and reliability issues need to beaddressed [6] [7] In many cases the corresponding E2Elatency as low as 1 ms needs to be met with reliability ashigh as 9999 [8]

To achieve low latency for MCC drastic changes in thenetwork architecture need to be performed Since the delayis contributed by radio access network (RAN) and core net-work along with backhaul between RAN and core networknew network topology involving software define network(SDN) network virtualized function (NFV) and mobile edgecomputing (MEC)caching can be employed to reduce thelatency significantly This can happen due to the capabilityof seamless operation and independence from hardware func-tionality provided by these entities Moreover new physicalair interface with small time interval transmission small sizepackets new waveforms new modulation and coding schemesare the areas of investigation for attaining low latency Inaddition optimization of radio resource allocation massiveMIMO carrier aggregation in millimeter wave and priorityof data transmission need to be addressed All in all a robustintegration with existing LTE is necessary for 5G networksthat will enable industries to deploy 5G quickly and efficientlywhen it is standardized and available In summary 5G wirelessaccess should be an evolution of LTE complemented withrevolutionary architecture designs and radio technologies

Even though the goals of 5G are ambitious based on 4Gpoint of view researchers from industry and academia areworking to bring 5G key performance indicator (KPI) goals(including low latency) into reality The 5G road map is fixed5G standardization is set up by 2018 5G first commerciallaunch is to be by 2020 and 5G worldwide launch willbe by 2022 and onwards [9] Along with the ITU various

2uRLLC allows E2E latency of less than 1 ms on all layers with packeterror rates of 10minus5 to 10minus9

arX

iv1

708

0256

2v2

[cs

NI]

29

May

201

8

European research projects such as METIS-II MiWaveS andmmMAGIC are working to address diverse aspects of 5Gsuch as RF block and algorithm for mmWave communicationOn the other hand projects such as ADEL FANTASTIC5G SPEED 5G and Flex5GWare address hardware andfundamental building blocks for 5G while 5G CHAMPIONEuropeanKorean research project is working to implementproof of concept (PoC) of 5G network encompassing allcutting edge radio core and satellite technologies [10] Itaims to showcase the 5G PoC with latency of 1 ms on 2018Winter Olympics in PyeongChang Korea Telecommunicationvendors such as Ericsson Huawei and Nokia Siemens areworking to bring network infrastructure and UE for 5G roll outby 2020 Besides researchers from academia are working ondifferent aspectsgoals of 5G including low latency Howeverthe real field PoC and benchmarking of performance is to bedone

In the literature surveys on 5G network including archi-tecture [3] [4] SDNNFVMEC based core network [11][12] caching [13] [14] backhaul [15] resource management[16] and data centric network [17] [18] are available Apartfrom that surveys on latency reduction approaches in Internet[19] cloud computing [20] [21] and distributed networkapplications [22] [23] are also presented however to thebest of our knowledge a comprehensive survey on latencyreduction approaches in cellular networks towards 5G is notavailable yet

In this paper we present a comprehensive survey of latencyreduction solutions particularly in the context of 5G wirelesstechnology We first present the sources and fundamentalconstraints for achieving low latency in a cellular networkWe also overview an exemplary 5G network architecturewith compliance to IMT-2020 vision Finally we providean extensive review of proposed solutions for achieving lowlatency towards 5G The goal of our study is to bring allexisting solutions on the same page along with future researchdirections We divide the existing solutions into three parts(1) RAN solutions (2) Core network solutions (3) Cachingsolutions However detailed comparison of these solutions arebeyond the scope of this work

The rest of the paper is organized as follows Section IIpresents the latency critical services in 5G The sources oflatency in a cellular network are discussed in Section III Sec-tion IV reviews the fundamental constraints and approachesfor achieving low latency Three key low latency solutions inRAN core network and caching are presented in SectionsV VI and VII respectively Section VIII presents the fieldtests trials and experiments of low latency approaches Openissues challenges and future research directions are discussedin Section IX Finally concluding remarks are provided inSection X Some of the acronyms used in this paper arepresented in Table I

II LOW LATENCY SERVICES IN 5GLatency is highly critical in some applications such as

automated industrial production controlrobotics transporta-tion health-care entertainment virtual realty education andculture In particular IoT is quickly becoming a reality which

TABLE I LIST OF ACRONYMS

AS Access StratumAR Augmented RealityBC Broadcast ChannelBLER Block Error RateCCP Communication Control PortCCSE Control Channel Sparse EncodingCSIT Channel State Information at the TransmitterDTB Delivery Time per BitD2D Device to DeviceDRB Data Radio BearereMBB Enhanced Mobile BroadbandEPC Evolved Packet CoreFFT Fast Fourier TransformFDT Fractional Delivery TimeFBMC Filter Bank Multi CarrierGFDM Generalized Frequency Division MultiplexingGP Guard PeriodGTP GPRS Tunnel ProtocolGGSN Gateway GPRS (General Packet Radio Service) Ser-

vice NodeHARQ Hybrid Automatic-Repeat-RequestICI Inter Carrier InterferenceIoT Internet of ThingsITS Intelligent Transportation SystemIFFT Inverse Fast Fourier TransformITU International Telecommunications UnionITU-R International Telecommunication Union-Radio Com-

munication SectorIMT-2020 International Mobile Telecommunication System

with a target date set for 2020ISI Inter Symbol InterferenceMEC Mobile Edge ComputingMTP Machine Type CommunicationMETIS Mobile and Wireless Communications Enablers for

Twenty-Twenty (2020) Information SocietyNGMN Next Generation Mobile NetworksMCC Mission Critical CommunicationMRC-ZF Maximum Ratio Combining Zero ForcingMAC Medium Access ControlmMTc Massive Machine Type CommunicationMME Mobility Management EntityMIMO Multiple Input Multiple OutputMDP Markov Decision ProcessNFV Network Function VirtualizationNAS Non-Access StratumNDT Normalized Delivery TimeOFDM Orthogonal Frequency Division MultiplexingOOB Out of BandOW Optical WindowOLLA Outer Loop Link AdaptationPUCCH Physical Uplink Control ChannelP2P Peer-to-PeerRAN Radio Access NetworkSC-FDMA Single Carrier Frequency Division Multiple AccessSRB Signaling Radio BearerSCMA Sparse Code Multiple AccessSGW Serving GPRS GatewayTDD Time Division DuplexRAT Radio Access TechnologyuRLLC ultra Reliable Low Latency CommunicationUFMC Universal Filtered Multi-CarrierUDN Ultra Dense NetworkVR Virtual RealityVLC Visible Light CommunicationZF Zero Forcing5G Fifth Generation Mobile Network5GETLA 5G Flexible TDD based Local Area

connects anything to any other thing anytime and anywhereSmart wearable devices (smart watches glasses bracelets

and fit bit) smart home appliances (smart meters fridgestelevisions thermostat) sensors autonomous cars cognitivemobile devices (drones robots etc) are connected to always-on hyper-connected world to enhance our life style [24]ndash[26]Even though operators are supporting these IoT applicationsthrough existing 3GLTE some applications require muchmore stringent requirements from underlying networks suchas low latency high reliability [27] [28] and security [29][30] In some cases we need latency as low as 1 ms withpacket loss rate no larger than 10minus2 Several latency criticalservices which need to be supported by 5G are described asfollows

bull Factory Automation Factory automation includes real-time control of machine and system for quick productionlines and limited human involvement In these cases theproduction lines might be numerous and contiguous Thisis highly challenging in terms of latency and reliabilityTherefore the E2E latency requirement for factory au-tomation applications is between 025 ms to 10 ms with apacket loss rate of 10minus9 [31] [32] In factory automationthe latency is measured as E2E in which the sensorsmeasuring data are at one end and transmit the datafor processing to the other end for programmable logiccontroller (PLC) The proposed values for the latency arebased on the KoI (Koordinierte Industriekommunikation)project in which a detailed questionnaire-based survey isconducted to collect the information from an extensiverange of factory automation processes [33]

bull Intelligent Transportation Systems Autonomous driv-ing and optimization of road traffic requires ultra re-liable low latency communication According to intel-ligent transportation systems (ITS) different cases in-cluding autonomous driving road safety and traffic ef-ficiency services have different requirements [31] [34]Autonomous vehicles require coordination among them-selves for actions such as platooning and overtaking [35]For automated vehicle overtaking maximum E2E latencyof 10 ms is allowed for each message exchange Forvideo integrated applications such as see-through-vehicleapplication described in [36] requires to transmit rawvideo which allows maximum delay of 50 ms [37] Roadsafety includes warnings about collisions or dangeroussituations Traffic efficiency services control traffic flowusing the information of the status of traffic lights andlocal traffic situations For these purposes latency of10 ms to 100 ms with packet loss rate of 10minus3 to 10minus5

is requiredbull Robotics and Telepresence In the near future remote

controlled robots will have applications in diverse sec-tors such as construction and maintenance in dangerousareas A prerequisite for the utilization of robots andtelepresence applications is remote-control with real-timesynchronous visual-haptic feedback In this case systemresponse times should be less than a few millisecondsincluding network delays [31] [38] [39] Communicationinfrastructure capable of proving this level of real-timecapacity high reliabilityavailability and mobility support

is to be addressed in 5G networksbull Virtual Reality (VR) Several applications such as

micro-assembly and tele-surgery require very high levelsof sensitivity and precision for object manipulations VRtechnology accommodates such services where severalusers interact via physically coupled VR simulations ina shared haptic environment Current networked commu-nication does not allow sufficient low latency for stableseamless coordination of users Typical update rates ofdisplay for haptic information and physical simulation arein the order of 1000 Hz which allows round trip latencyof 1 ms Consistent local view of VR can be maintainedfor all users if and only if the latency of around 1 ms isachieved [38] [39] [45]

bull Augmented Reality (AR) In AR technology the aug-mentation of information into the userrsquos field of view en-ables applications such as driver-assistance systems im-proved maintenance citymuseum guides telemedicineremote education and assistive technologies for policeand firefighters [38] However insufficient computationalcapability of mobile devices and latency of current cellu-lar network hinder the applications In this case latencyas low as a few milliseconds is required

bull Health care Tele-diagnosis tele-surgery and tele-rehabilitation are a few notable healthcare applicationsof low latency tactile Internet These allow for remotephysical examination even by palpation remote surgeryby robots and checking of patientsrsquo status remotelyFor these purposes sophisticated control approaches withround trip latency of 1-10 ms and high reliability datatransmission is mandatory [38] [39]

bull Serious Gaming The purpose of serious gaming is notlimited to entertainment Such games include problem-solving challenges and goal-oriented motivation whichcan have applications in different areas such as educationtraining simulation and health Network latency of morethan 30-50 ms results in a significant degrade in game-quality and game experience ratings Ideally a roundtrip time (RTT) on the order of 1 ms is recommendedfor perceivable humanrsquos interaction with the high-qualityvisualization [38]

bull Smart Grid The smart grid has strict requirements ofreliability and latency [46]ndash[49] The dynamic controlallows only 100 ms of E2E latency for switching sup-pliers (PV windmill etc) on or off However in caseof a synchronous co-phasing of power suppliers (iegenerators) an E2E delay of not more than 1 ms is needed[3] [38] Latency more that 1 ms which is equivalent to aphase shift of about 18 (50 Hertz AC network) or 216

(60 Hertz AC network) may have serious consequencein smart grid and devices

bull Education and Culture Low latency tactile Internetwill facilitate remote learningeducation by haptic overlayof teacher and students For these identical multi-modalhuman-machine interfaces round trip latency of 5-10 msis allowed for perceivable visual auditory and hapticinteraction [38] [39] Besides that tactile Internet will

TABLE II TYPICAL LATENCY AND DATA RATE REQUIREMENTS FOR DIFFERENT MISSION CRITICAL SERVICES

Use case Latency Data rate RemarksFactory Automation 025-10 ms [31] 1 Mbps [40] ndash Generally factory automation applications require small

data rates for motion and remote controlndash Applications such as machine tools operation may allow

latency as low as 025 msIntelligent Transport Sys-tems (ITS)

10-100 ms [31] 10-700 Mbps [41] ndash Road safety of ITS requires latency on the order of10 ms

ndash Applications such as virtual mirrors require data rateson the order of 700 Mbps

Robotics and Telepresence 1 ms [42] 100 Mbps [43] ndash Touching an object by a palm may require latency downto 1 ms

ndash VR haptic feedback requires data rates on the order of100 Mbps

Virtual Reality (VR) 1 ms [38] 1 Gbps [43]ndash Hi-resolution 360

VR requires high rates on the order

of 1 Gbps while allowing latency of 1 msHealth care 1-10 ms [39] 100 Mbps [43] ndash Tele-diagnosis tele-surgery and tele-rehabilitation may

require latency on the order of 1 ms with data rate of100 Mbps

Serious Gaming 1 ms [38] 1 Gbps [43] ndash Immersive entertainment and humanrsquos interaction withthe high-quality visualization may require latency of1 ms and data rates of 1 Gbps for high performance

Smart Grid 1-20 ms [31] [38] 10-1500 Kbps [44] ndash Dynamic activation and deactivation in smart grid re-quires latency on the order of 1 ms

ndash Cases such as wide area situational awareness requiredate rates on the order of 1500 Kbps

Education and Culture 5-10 ms [38] 1 Gbps [43] ndash Tactile Internet enabled multi modal human-machineinterface may require latency as low as 5 ms

ndash Hi-resolution 360

and haptic VR may require datarates as high as 1 Gbps

allow to play musical instruments from remote locationsIn such scenarios supporting network latency lower thanfew milliseconds becomes crucial [38]

Based on the applications and use case scenarios abovelatency critical services in 5G networks demand an E2E delayof 1 ms to 100 ms The latency requirements along withestimated data rates for various 5G services are summarizedin Table II Some use cases such as VR and online gamingmay require round trip latency on the order of 1 ms withdata rates as high as 1 Gbps On the other hand use casessuch as factory automation and smart grid require lower datarates on order of 1 Mbps with demanding latency of 1 msFor required data rates on the order of 1 Gbps [43] reportsthat bandwidth of 40 MHz is sufficient at 20 node densityper square kilometer For data rates of few Mbps bandwidthof 20 MHz and lower can be sufficient for most scenariosThis means spectral efficiency supported by 5G is 50 bpsHzwhile LTE-A can support upto 30 bpsHz [50] For lowerbandwidth spectrum below 6 GHz can be utilized while forhigh bandwidth requirement mmWave can be an attractivechoice [43]

In the next section the major sources of latency in a cellularnetwork are discussed

III SOURCES OF LATENCY IN A CELLULAR NETWORK

In the LTE system the latency can be divided into twomajor parts (1) user plane (U-plane) latency and (2) controlplane (C-plane) latency The U-plane latency is measured by

one directional transmit time of a packet to become availablein the IP layer between evolved UMTS terrestrial radio accessnetwork (E-UTRAN) edgeUE and UEE-UTRAN node [51]On the other hand C-plane latency can be defined as thetransition time of a UE to switch from idle state to activestate At the idle state an UE is not connected with radioresource control (RRC) After the RRC connection is beingsetup the UE switches from idle state into connected stateand then enters into active state after moving into dedicatedmode Since the application performance is dependent mainlyon the U-plane latency U-plane is the main focus of interestfor low latency communication

In the U-plane the delay of a packet transmission in acellular network can be contributed by the RAN backhaulcore network and data centerInternet As referred in Fig 1

InternetCloud

SDNVirtualizedserver

TRadio TBackhaul Tcore TTransport

UsereNB

Evolved packetcore (EPC)

SGSNMME

Fig 1 Latency contribution in E2E delay of a packettransmission

the total one way transmission time [52] of current LTE systemcan be written as

T = TRadio + TBackhaul + TCore + TTransport (1)

wherebull TRadio is the packet transmission time between eNB and

UEs and is mainly due to physical layer communicationIt is contributed by eNBs UEs and environment Itconsists of time to transmit processing time at eNBUEretransmissions and propagation delay Processing de-lay at the eNB involves channel coding rate matchingscrambling cyclic redundancy check (CRC) attachmentprecoding modulation mapper layer mapper resourceelement mapper and OFDM signal generation On theother hand uplink processing at UE involves CRC at-tachment code block segmentation code block concate-nation channel coding rate matching data and controlmultiplexing and channel interleaver Propagation delaydepends on obstacles (ie building trees hills etc) onthe way of propagation and the total distance traveled bythe RF signal

bull TBackhaul is the time for building connections betweeneNB and the core network (ie EPC) Generally the corenetwork and eNB are connected by copper wires or mi-crowave or optical fibers In general microwave involveslower latency while optic fibers come with comparativelyhigher latency However spectrum limitation may curbthe capacity of microwave [53]

bull TCore is the processing time taken by the core networkIt is contributed by various core network entities suchas mobility management entity (MME) serving GPRSsupport node (SGSN) and SDNNFV The processingsteps of core network includes NAS security EPS bearercontrol idle state mobility handling mobility anchoringUE IP address allocation and packet filtering

bull TTransport is the delay to data communication betweenthe core network and Internetcloud Generally distancebetween the core network and the server bandwidth andcommunication protocol affect this latency

The E2E delay TE2E is then approximately given by 2timesT The TRadio is the sum of transmit time propagation latencyprocessing time (channel estimation encoding and decodingtime for first time) and retransmission time (due to packetloss) In particular the TRadio for a scheduled user [54] [55]can be expressed as

TRadio = tQ + tFA + ttx + tbsp + tmpt (2)

wherebull tQ is the queuing delay which depends on the number of

users that will be multiplexed on same resourcesbull tFA is the delay due to frame alignment which depends on

the frame structure and duplexing modes (ie frequencydivision duplexing (FDD) and time division duplexing(TDD))

bull ttx is the time for transmission processing and payloadtransmission which uses at least one TTI depending on

radio channel condition payload size available resourcestransmission errors and retransmission

bull tbsp is the processing delay at the base stationbull tmpt is the processing delay of user terminal Both

the base station and user terminal delay depend on thecapabilities of base station and user terminal (ie UE)respectively

In compliance with ITU TRadio should not be more than05 ms for low latency communication [56] In this regardradio transmission time should be designed to be on the orderof hundreds of microseconds while the current configurationin 4G is 1 ms For this enhancement in various areas ofRAN such as packetframe structure modulation and codingschemes new waveform designs transmission techniques andsymbol detection need to be carried out In order to reducethe delay in TBackhaul approaches such as advanced backhaultechniques cachingfog enabled networks and intelligent in-tegration of AS and NAS can provide potential solutions ForTCore new core network consists of SDN NFV and variousintelligent approaches can reduce the delay significantly ForTTransport MECfog enabled Internetcloudcaching can providereduced latency

In the following section we discuss the constraints andapproaches for achieving low latency

IV CONSTRAINTS AND APPROACHES FOR ACHIEVINGLOW LATENCY

There are major fundamental trade-offs between capacitycoverage latency reliability and spectral efficiency in a wire-less network Due to these fundamental limits if one metric isoptimized for improvement this may results in degradation ofanother metric In the LTE system the radio frame is 10 mswith the smallest TTI being 1 ms This fixed frame structuredepends on the modulation and coding schemes for adaptationof the transmission rate with constant control overhead Sincelatency is associated with control overhead (cyclic prefixtransmission mode and pilot symbols) which occupies a majorportion of transmission time of a packet (approximately 03-04 ms per packet transmission) it is not wise to considera packet with radio transmission time less than 1 ms Ifwe design a packet with time to transmit of 05 ms morethan 60 of the resources will be used by control overhead[52] Moreover retransmission per packet transmission takesaround 8 ms and removal of retransmission will affect packeterror significantly As a result we need radical modificationsand enhancements in packetframe structure and transmissionstrategy In this regardbull First a novel radio frame reinforced by limited control

overhead and smaller transmission time is necessary to bedesigned For reduction of control overhead proceduresfor user scheduling resource allocation and channeltraining can be eliminated or merged

bull Second packet error probability for first transmissionshould be reduced with new waveforms and transmissiontechniques reducing the retransmission delay

bull Third since latency critical data needs to be dispatchedimmediately techniques for priority of data over normaldata need to be identified

bull Fourth synchronization and orthogonality are the indis-pensable aspects of OFDM that are major barriers forachieving low latency Even though asynchronous modeof communication is more favorable over synchronizedoperation in terms of latency it requires additional spec-trum and power resources [57]

bull Fifth since the latency for data transmission also dependson the delay between the core network and the BScaching networks can be used to reduce latency by storingthe popular data at the network edge

Researchers proposed various techniquesapproaches forachieving low latency in 5G As summarized in Fig 2 wedivided the existing solutions into three major categories(1) RAN solutions (2) core network solutions and (3)caching solutions The RAN solutions include newmodifiedframe or packet structure waveform designs multiple ac-cess techniques modulation and coding schemes transmissionschemes control channels enhancements low latency symboldetection mmWave aggregation cloud RAN reinforcing QoSand QoE energy-aware latency minimization and locationaware communication techniques On the other hand newentities such as SDN NFV MEC and fog network alongwith new backhaul based solutions have been proposed for thecore network The solutions of caching can be subdivided intocaching placement content delivery centralized caching anddistributed caching while backhaul solutions can be dividedinto general and mmWave backhaul In the following sectionsthese solutions are described in further details

V RAN SOLUTIONS FOR LOW LATENCY

To achieve low latency various enhancements in the RANhave been proposed Referring to Table III RAN solu-tionsenhancements include framepacket structure advancedmultiple access techniqueswaveform designs modulation and

coding scheme diversity and antenna gain control channelsymbol detection energy-aware latency minimization carrieraggregation in mmWave reinforcing QoS and QoE cloudRAN and location aware communication In what follows thedetailed overview for each of these solutions is presented

A Framepacket structure

In the RAN solutions modification in the physical air inter-face has been considered as an attractive choice In particularmost of the proposed solutions are on the physical (PHY) andmedium access control (MAC) layers

In LTE cellular network the duration of a radio frame is10 ms Each frame is partitioned into 10 subframes of size1 ms which is further divided into 05 ms units that arereferred as a resource block (RB) Each RB spans 05 ms(6 or 7 OFDM symbols) in time domain and 180 KHz (12consecutive subcarriers each of which 15 KHz) in frequencydomain Based on this the subcarrier spacing ∆f is 15 KHzthe OFDM symbol duration TOFDM is 1

∆f = 6667micros the FFTsize is 2048 the sampling rate fs is ∆ftimesNFFT = 3372 MHzand the sampling interval Ts is 1fs

To reduce TTI for achieving low latency the subcarrierspacing ∆f can be changed to 30 KHz [60] This results thecorresponding OFDM symbol duration TOFDM to be 3333 microsand the FFT size NFFT to become 1024 while sampling ratefs is kept 3072 MHz similar to LTE systems The frameduration Ts=10 ms can be divided into 40 subframes in whicheach subframe duration Tsf is 025 ms and contains 6 or 7symbols Two types of cyclix prefixs (CPs) can be employedin this configuration with durations

Tcp1 = 564timesNIFFT times Ts asymp 2604 micros (3)

5G low latencycommunication

Core network CachingRAN

FramePacketstructure

WaveformMultiple Access

Modulation andcoding

Transmitteradaptation

Control signaling

SDN

NFV

MEC

Fog network

Caching placement

Content delivery

Generalbackhauling

mmWavebackhauling

Symbol detection

mmWave

Location awarecommunication

QoSQoEdifferentiation

CRAN and others

Centralized caching

Distributed caching

Fig 2 Categories of different solutions for achieving low latency in 5G

TABLE III OVERVIEW OF TECHNIQUES IN RAN FOR LOW LATENCY

CaseArea Reference Approach Summary[58] [59] Small packetsshort TTI Transmission of small scale data is investigated for packet loss rate of 10minus9 and

latency as low as 1 ms[60] Subcarrier spacing Subcarrier spacing is enlarged to shorten the OFDM symbol duration and the

number of OFDM symbols is proposed to keep unchanged in each subframeFramePacketstructure

[8] [61]ndash[63] Flexible OFDMA based TDDsubframe

TDD numerology is optimized for dense deployment with smaller cell sizes andlarger bandwidth in the higher carrier frequencies

[64] Modification of physical sub-frame

Different control and data part patterns for consecutive subframes TX and RXcontrol parts are proposed to be separated from each other and from the datasymbols with a GP leading to total number of 3 GPs per subframe

[54] [65]ndash[67] Numerology flexible sub frameand resource allocation

Numerology and subframe structure are defined considering diverse carrier fre-quencies and bandwidths to envision 5G including low latency Cyclic prefix FFTsize subcarrier spacing and sampling frequency were expressed as the function ofcarrier frequency

Advancedmultiple ac-cessWaveform

[68] [69] [70] Filtered CP-OFDM UFMC andFBMC

UFMC outperforms over OFDM by about 10 in case of both large and smallpackets FBMC demonstrates better performance in case of transmitting longsequences however it suffers during the transmission of short burstsframes

[71] [72] Polar coding Based on simulation and field test polar coding has been proposed for 5Goutperforming over turbo coding in case of small packet transmission

[73] Turbo decoding with combinedsliding window algorithm andcross parallel window (CPW) al-gorithm

A highly-parallel architecture for the latency sensitive turbo decoding is proposedcombining two parallel algorithms the traditional sliding window algorithm andcross parallel window (CPW) algorithm

[74] New IFFT design with butterflyoperation

Input signal of IFFT processor corresponding to guard band are assigned as nullrevealing the existence of numerous zeros (ie 0) If the sequence of OFDM symboldata which enter the IFFT is adjusted the memory depth can be reduced from 1024to 176

Modulationand coding

[75] Sparse code multiple access(SCMA)

A dynamic shrunk square searching (DSSS) algorithm is proposed which cuts offunnecessary communication control port (CCP) calculation along with utilizationof both the noise characteristic and state space structure

[76] Priority to latency critical data A latency reduction approach by introducing TDM of higher priority ultra-lowlatency data over other less time critical services is proposed which maps higherpriority user data during the beginning of a subframe followed by the normal data

[77] Balanced truncation Balanced truncation is applied for the model reduction in the linear systems thatare being coupled over arbitrary graphs under communication latency constraints

[78] Finite block length bounds andcoding

Recent advances in finite-block length information theory are utilized in order todemonstrate optimal design for wireless systems under strict constraints such aslow latency and high reliability

[79] Asymmetric window Asymmetric window is proposed instead of well-known symmetric windows forreduction of cyclic prefix by 30 This technique suppresses OOB emission butmakes the system more susceptible to channel induced ISI and ICI

[80] Transmission power optimiza-tion

Transmission power is optimized by steepest descent algorithm considering trans-mission delay error probability and queuing delay

[81] Path-switching method and apacket-recovery method

Low latency packet transport system with a quick path-switching method and apacket-recovery method are introduced for a multi-radio-access technology (multi-RAT) environment

[82] Diversity Diversity could be employed through various approaches such as spatial diversitytime diversity and frequency diversity

[83] Control channel sparse encoding(CCSE)

CCSE is introduced in order to provide the control information using non-orthogonal spreading sequences

[84] Scaled control channel design A scaled-LTE frame structure is proposed assuming the scaling factor to be 5 with adedicated UL CCHs for all sporadic-traffic users in each transmission time intervalwith possible smallest SR size

[85] Symbol-level frequency hoppingand sequence-based sPUCCH

A sequence-based sPUCCH (SS-PUCCH) incorporating two SC-FDMA symbolsis introduced in order to meet a strict latency requirement Symbol-level frequencyhopping technique is employed to achieve frequency diversity gain and reliabilityenhancement

Controlsignaling

[86] Radio bearer and S1 bearer man-agement

Establishment of radio bearer and S1 bearer in parallel are proposed where eNBand mobility management element (MME) manages and controls radio bearer andS1 bearer respectively The eNB sends only single control signal in order toconfigure radio bearers such as SRB1 SRB2 and DRBs that decreases the signalinginteraction rounds between the UE and the eNBs

[87] Outer-loop link adaptation(OLLA) scheme

The proposed scheme controls the size of the compensation in the estimated SINRbased on the time elapsed after a UE transits from an idle state to an active statewhich helps to reduce latency for small packet applications

TABLE III OVERVIEW OF TECHNIQUES IN RAN FOR LOW LATENCY (CONTINUED)

CaseArea Reference Approach Summary[88] SM-MIMO detection scheme

with ZF and MRC-ZFA low-complexity and low latency massive SM-MIMO detection scheme is intro-duced and validated using SDR platforms The low complexity detection schemeis proposed with a combination of ZF and MRC-ZF

[89] Linear MMSE A linear MMSE receiver is presented for low latency wireless communicationsusing ultra-small packets

Symboldetection

[90] Space-time encoding and widelylinear estimator

Space-time encoding is introduced within a GFDM block for maintaining overalllow latency in the system On the other hand a widely linear estimator is used todecode the GFDM block at the receiver end which yields significant improvementsin gain over earlier works

[91]ndash[93] Compressed sensing Compressed sensing has been proposed to be effective in reducing latency ofnetworked control systems if the state vector can be assumed to be sparse in somerepresentation

[94] Low complexity receiver design A low complexity receiver is designed and using this the performance of an SCMAsystem is verified via simulations and real-time prototyping This approach triplesthe whole system throughput while maintaining low latency similar to flexibleorthogonal transmissions

mmWave [95]ndash[98] mmWave based air interface Physical layer air interface is proposed using mmWave aggregation Large band-width along with various approaches such as small frame structure mmWavebackhaul and beamforming can help to achieve low latency

Locationaware com-munication

[99]ndash[101] Location information Issues and research challenges of 5G are discussed followed by the conclusionthat 5G networks can exploit the location information and accomplish performancegains in terms of throughput and latency

QoSQoEDifferentia-tion

[102]ndash[111] Parameter manipulation Differentiation of constraints on QoS and QoE can maintain low latency in 5Gservices including ultra high definition and 3D video content real time gamingand neurosurgery

Cloud RAN(CRAN)

[112]ndash[116] Cloud architecture based RAN CRANs combine baseband processing units of a group of base stations into acentral server while retaining radio front end at the cell sides Proper optimizationof resources can ensure low latency along with capital expenditure reduction

GB GB GB GB GB GB GB

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

n times 180 kHzmtimes1ms

m = Fraction of 1 msn = Multiplier of 180 kHz

(a)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(b)

Fig 3 Physical air interface (a) Conventional LTE radio frame (b) Exemplary 5G radio frame with flexible time andfrequency division for low latency [99] (GB guard band LLC low latency communication)

A conventional LTE radio frame with equal sized RB and anexemplary 5G physical air frame are illustrated in Fig 3(a)and Fig 3(b) respectively

In [58] an extensive analysis of the theoretical principlesthat regulates the transmission of small-scale packets with lowlatency and high reliability is presented with metrics to assesstheir performance The authors emphasize control overheadoptimization for short packet transmission In [54] a flexible5G radio frame structure is introduced in which the TTI sizeis configurable in accordance with the requirement of specificservices At low offered load 025 ms TTI is an attractivechoice for achieving low latency due to low control overheadHowever for more load control overhead increases whichaffects reliability and packet recovery mechanism resulting inincreased latency This study argues to employ user schedulingwith different TTI sizes in the future 5G networks In [59]the authors try to improve the outage capacity of URLLC andsatisfy the low latency requirement of 5G using an efficientHARQ implementation with shortened transmission TTI andRTT Moreover some simulations are conducted in order toprovide insights on the fundamental trade-off between the out-age capacity system bandwidth and the latency requirementfor URLLC

In [65] the numerology and subframe structure are definedconsidering diverse carrier frequencies and bandwidths forlow latency 5G networks Cyclic prefix FFT size subcarrierspacing and sampling frequency were expressed as a functionof the carrier frequency In [63] software defined radio (SDR)platform based 5G system implementation with strict latencyrequirement is presented The scalability of the proposed radioframe structure is validated with E2E latency less than 1 ms In[60] the proposed subcarrier spacing is enlarged to shorten theOFDM symbol duration and the number of OFDM symbolsin each subframe is kept unchanged in the new frame structurefor TDD downlink The subcarrier spacing is changed to 30KHz resulting the corresponding OFDM symbol duration T= 3333 micros The fast Fourier transform (FFT) size N is 1024while the sampling rate fs is kept same as 3072 MHz Theframe duration Ts is still 10 ms with 40 subframes

In [64] in order to have fully flexible allocations of differentcontrol and data RB in the consecutive subframes TX and RXcontrol RBs are proposed to be separated from each other andalso from the data RB by guard periods (GPs) This leads tototal number of 3 GPs per subframe which separates themAssuming symmetrical TX and RX control parts with Nctrl s

symbols in each and defining that same subcarrier spacingis used for control and data planes with Ndata s being thenumber of data symbols and Tsymbol being the length of anOFDM symbol the subframe length Tsf can be determined as

Tsf = (2Nctrl s +Ndata s(Tsymbol + TCP)) + 2TGP (5)

In [61] the fundamental limits and enablers for low airinterface latency are discussed with a proposed flexible OFDMbased TDD physical subframe structure optimized for 5G localarea (LA) environment Furthermore dense deployment withsmaller cell sizes and larger bandwidth in the higher carrierfrequencies are argued as notable enablers for air interface

latency reduction In [8] a new configurable 5G TDD framedesign is presented which allows flexible scheduling (resourceallocation) for wide area scenarios The radical trade-offsbetween capacity coverage and latency are discussed furtherwith the goal of deriving a 5G air interface solution capableof providing low latency high reliability massive connectivityand enhanced throughput Since achieving low latency comesat cost of lower spectral efficiency the proposed solution ofthe study includes control mechanisms for user requirementie whether the link should be optimized for low latency orhigh throughput

A 5G flexible frame structure in order to facilitate users withhighly diversified service requirements is proposed in [62] Al-though in-resource physical layer control signaling is the basisof this proposed radio frame it allows the corresponding datatransmission based on individual user requirements For this itincorporates adaptable multiplexing of users on a shared chan-nel with dynamic adjustment of the TTI in accordance with theservice requirements per link This facilitates optimization ofthe fundamental trade-offs between latency spectral efficiencyand reliability for each link and service flow In [66] a schemethat reserves resources for re-transmission for a group of ultrareliable low latency communication UEs is presented Theoptimum dimensioning of groups and block error rate (BLER)target can reduce the probability of contention for the sharedretransmission resources Moreover the unused resources canbe utilized for non-grouped UEs resulting in overall efficiencyenhancement

In [67] fundamental trade-offs among three KPIs (reliabil-ity latency and throughput) in a 4G network is characterizedand an analytical framework is derived In cases where the

TABLE IV PHY AND MAC LAYER BASED RADIOINTERFACE SOLUTIONS FOR LOW LATENCY

References ApproachArea PHY layer MAC layer

[54] Short TTI

[58] [65] Numerology and subframe structure

[60] Subcarrier

[61] Flexible subframe

[63] Flexible subframeimplementation withSDR platform

[64] Allocation of controland data RB

[8] Radio frame andscheduling

[62] Flexible TTI and mul-tiplexing

[59] Efficient HARQimplementationwith shortenedtransmission TTI andRTT

[66] Reservation ofresources

[67] Calculating thefundamental trade-offs among threeKPIs

Nsubcarrier

K time samplesM sub-symbols

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Fig 4 Slot placment in GFDM OFDM and SCFDM

theory can not be extended via mathematical formulationsdue to complexity of scenario in hand some guidelines areprovided to make the problem tractable In order to improvethe aforementioned trade-offs between these KPIs in future 5Gsystems different candidate techniques are proposed

The above approaches of framepacket structure to achievelow latency at the RAN level are tabulated in Table IV

B Advanced Multiple Access TechniquesWaveform

Different kinds of candidate multiple access (MA) tech-niques and waveforms including orthogonal non orthogonaland asynchronous access have been proposed for low latencycommunication [57] [68]ndash[70] Since synchronization andorthogonality (integral to OFDM) is a hindrance for achievinglow latency asynchronized non orthogonal multiple accesstechniques have been discussed in [57] Reduction of symbolduration to 67 micros is not a promising solution in critical timebudgeting In this regard interleave division multiple access(IDMA) has been introduced in [117] [118] for generatingsignal layers The IDMA is a variant of the CDMA techniquewhich uses specific interleaving for user segregation in lieu ofusing a spread sequence to the individual user Here channelcoding forward error correction coding and spreading are

combined into a single block by a low rate encoder Thespreading can not be considered as a distinct and specialtask Interleaving usually utilizes a simpler iterative multiuseridentification approach However this approach needs furtherrigorous investigation

In order to supply synchronization and orthogonality sparsecode multiple access (SCMA) and non orthogonal multipleaccess (NOMA) have been presented in [119] for 5G scenar-ios In SCMA symbol mapping and spreading are combinedtogether and the mapping of multi dimensional codeword overincoming bits is performed directly from SCMA codebookSCMA is comparatively simpler and has superior performanceover low density version of CDMA Another modulation tech-nique that aims to reduce latency is referred as the generalizedfrequency division multiplexing (GFDM) is introduced in [90][120] The flexibility of covering both the cyclic prefix OFDM(CP-OFDM) and single carrier frequency domain equalization(SC-FDE) and block structure of GFDM help to achievelow latency A typical mapping structure of GFDM OFDMand SC-FDM is illustrated in Fig 4 The overall comparisonamong IDMA SCMA and GFDM is presented in Table V

Filter bank multi carrier (FBMC) has been a strong candi-date waveform for 5G [3] [121] FBMC demonstrates betterperformance in case of transmitting long sequences howeverit suffers during the transmission of short burstsframes Forusage of cyclic prefix wide frequency guards and morerequired coordination OFDM may be inefficient in case of lowlatency communication [120] Universal filtered multi-carrier(UFMC) [120] [122] is upgraded version of FBMC whichoffsets the disadvantage of FBMC It outperforms OFDM byabout 10 in cases of time frequency efficiency inter carrierinterference (ICI) and transmissions of long or short packetsAdditionally UFMC preforms better than FBMC in the caseof very short packets while demonstrating similar performancefor long sequences These make UFMC as the one of the bestchoices for next generation low latency communication

In case of UFMC the time domain transmit vector [122] fora user is superposition of sub-band wise filtered componentsThe time domain transmit vector for a particular multi-carriersymbol of user k with filter length L and FFT length N is

Xk(N+Lminus1)times1

=

Bsumi=1

Fik(N+Lminus1)timesN

VikNtimesni

Siknitimes1

(6)

TABLE V PROPOSED MULTIPLE ACCESS TECHNIQUES FOR 5G

Cases IDMA [117] [118] SCMA [119] GFDM [90] [120]

Fundamentalconceptfeatures

bull Specific interleavingbull User segregationbull Iterative multiuser identification

bull Multiple dimensional codeword

bull QAM spreading combination

bull Block frame consists of timeslots and subcarriers

bull Non-orthogonalbull FFTIFFT implementation

Low complexity

Flexibility (in case of cov-ering CP-OFDM and SC-FDE)

Low latency

TABLE VI WAVEFORM CONTENDER FOR 5G

Cases OFDM FBMC [3] [121] UFMC [120] [122]

FilteringGeneralized filtering to all subcar-riers of entire band

Filtering to each subcarriers Generalized filtering to a group ofconsecutive subcarriers

Requirement of coordination Higher Lower LowerTime-frequency efficiency(due to CP and guard band) 084 1 1ICI (in case of lower degree ofsynchronization with UEs andeNBs)

Higher Lower Lower

Performance Performs well for large packetswith well coordination

Performs well for large packetswith less coordination

Performs well for short packetswith less coordination

wherebull S is the complex QAM symbol vectorbull V is the transformed time domain vector by IDFT matrix

In this case the relevant columns of the inverse Fouriermatrix are incorporated in accordance with the respectivesubband position within entire available band

bull i is the index of each subband of Bbull F is a Toeplitz matrix It is comprised of filter impulse

response and performs the linear convolutionThe symbol duration of (N +Lminus1) samples is determined

by the filter length and FFT size Filtering per block persubcarrier allows spectrally broad filters in pass band andshorter in time domain compared to FBMC The reduced timeyields shortened OFDM CP The filter ramp up and rampdown in shorten time domain ensures symbol shaping in a waythat allows protection against ISI and robustness for multipleaccess users Furthermore being orthogonal with respect tocomplex plain complex modulation symbol can be transmittedwithout further complication

Another advantage of UMFC is the ability of using differ-ent subcarrier spacings or filter times for users in differentsubbands If a user uses FFT size N1 and filter length L1and another user uses filter length and FFT size of N2

and L2 respectively then UFMC symbol duration can bedesigned such that N1 + L1 minus 1 = N2 + L2 minus 1 Thismakes UFMC a remarkable adaptive modulation scheme withcapability to be tailored easily under various characteristicsof communications including delayDoppler spread variationsin the radio channel and user QoS needs The comparativediscussion among OFDM FBMC and UMFC is presented inTable VI

C Modulation and Channel Coding

Although use of small packets is a potential approach forachieving low latency appropriate modulation and coding isrequired for small packet transmission for acceptable reliabil-ity In the literature mainly three types of coding schemes areproposed for 5G As presented in [71] low-density parity-check (LDPC) and polar codes outperform turbo codes interms of small packets while for medium and large packetsthe opposite is true While small packet is a requirementfor low latency other aspects such as implementation com-plexity performance in practical test and flexibility need tobe investigated In [72] polar code has been tested in fieldfor 5G considering various scenarios air interface frame

structure settings for large and small packets OFDM andfiltered OFDM (f-OFDM) waveforms In all cases polar codeperformed better than turbo codes which makes it a candidatechannel coding scheme for 5G The comparison among theschemes are illustrated in Table VII

In [73] a highly-parallel architecture for the latency sen-sitive turbo decoding is proposed by combining two parallelalgorithms the traditional sliding window algorithm and crossparallel window (CPW) algorithm New IFFT design withbutterfly operation is proposed in [74] which reduces IFFToutput data delay through the reduction of IFFT memory sizeand butterfly operation (eg additionsubtraction) Input signalof the IFFT processor corresponding to guard band is assignedas zero (ie lsquo0rsquo) revealing the existence of numerous zerosIf the sequence of OFDM symbol data which enter the IFFTis adjusted the memory depth can be reduced from 1024 to176

A dynamic shrunk square searching (DSSS) algorithm isproposed in [75] which cuts off unnecessary communicationcontrol port (CCP) calculation by utilizing both the noisecharacteristic and state space structure In this way it canmaintain close to optimal decoding performance in terms ofthe block error rate (BLER) This results in reduction ofdelay in communication In [76] a latency reduction approachby introducing time division multiplexing (TDM) of higherpriority ultra-low latency data over other less time criticalservices is proposed which maps higher priority user data

TABLE VII COMPARISON AMONG CHANNEL CODINGSCHEMES FOR LOW LATENCY [71]

Cases Turbocoding[73]

LDPC-PEG[71]

Convolutionalcoding [71]

Polarcodes[72]

Algorithm com-plexity for cod-ing 13 of 40 bitswith respect toturbo codes

100 98 667 15

Algorithm com-plexity for cod-ing 13 of 200bits with respectto turbo codes

100 98 667 1107

Performance inshort packets

Performance inmedium packets

during the beginning of a subframe followed by the normaldata In [77] balanced truncation is applied for the modelreduction in the linear systems that are being coupled overarbitrary graphs under communication latency constraints In[78] recent advances in finite-block length information theoryare utilized in order to demonstrate optimal design for wirelesssystems under strict constraints such as low latency and highreliability For a given set of constraints such as bandwidthlatency and reliability the bounds for the number of the bitsthat can be transmitted for an OFDM system is derived

D Transmitter Adaptation

A representative set of approaches for reducing latencyusing transmission side processing are tabulated in Table VIIIwhich will be overviewed in the rest of this subsection

In [79] an asymmetric window is proposed instead of well-known symmetric windows for reduction of cyclic prefix by30 and hence reducing latency due to reduced overheadThis technique suppresses out of bound (OOB) emissionbut makes the system more susceptible to channel inducedinter symbol interference (ISI) and inter carrier interference(ICI) Transmission power optimization by the steepest descentalgorithm considering transmission delay error probabilityand queuing delay is proposed in [80] In [81] low latencypacket transport system with a quick path-switching and apacket-recovery method is introduced for a multi-radio-accesstechnology (multi-RAT) environment In [82] use of diversitygain is proposed as a solution for capacity enhancementand latency reduction Diversity could be achieved throughvarious approaches such as spatial diversity time diversityand frequency diversity

In [123] a mmWave based switched architecture systemis proposed where control signals use low-resolution digitalbeamforming (to enable multiplexing of small control packets)with analog beamforming in the data plane (to enable higherorder modulation) This reduces the overhead significantly dueto the control signaling which results in more resources fordata transmission This technique leads to reduction of roundtrip latency in the physical layer

Recent advancements in full duplex (FD) communicationcomes forward with feature of doubling the capacity im-proving the feedback and latency mechanism meanwhileupholding steady physical layer security [124]ndash[129] Variousproposed techniques of 5G networks such as massive MIMOand beamforming technology providing reduced spatial do-main interference can be contributive for FD realization [125]Besides that intelligent scheduling of throughputdelay criticalpackets along with proper rate adaption and power assignmentcan results in capacity gain and reduction of latency Howeverthis field needs to be extensively investigated for studyingcapacity and latency trade offs

E Control SignalingWhen the packet size is reduced as envisioned in 5G sys-

tems control overhead takes the major portion of the packetAddressing this various approaches are proposed in order toreduce the control channel overhead The potential solutionstargeting the control channel enhancements to achieve lowlatency are illustrated in Table IX

In [83] control channel sparse encoding (CCSE) is in-troduced with vision to transmit the control information bymeans of non-orthogonal spreading sequences A scaled-LTEframe structure is proposed in [84] assuming the scaling factorto be 5 with dedicated UL control channels (CCHs) forall sporadic-traffic users in each TTI with possible smallestscheduling request (SR) size In [130] short TTI based uplinkframe has been proposed for achieving E2E latency no longerthan 1 ms In the proposed scheme subslot consisting of2 symbols has been proposed for uplink data and controlchannel A sequence-based sPUCCH (SS-PUCCH) incorpo-rating two single carrier-frequency division multiple access(SC-FDMA) symbols is introduced in [85] in order to meeta strict latency requirement Symbol-level frequency hoppingtechnique is employed to achieve frequency diversity gain andreliability enhancement

In the proposed procedure of [86] establishment of radiobearer and S1 bearer in parallel are proposed where eNB andmobility management element (MME) manages and controlsradio bearer and S1 bearer respectively The eNB sends only

TABLE VIII OVERVIEW OF SOLUTIONS IN TRANSMITTER ADAPTATION FOR LOW LATENCY

Reference Techniques Merits Demerits[79] Asymmetric window Reduces cyclic prefix by 30 and maintains

good OOB suppression along with latencyreduction

It assumes that spectral mask will be stricterin 5G networks

[80] Transmission power opti-mization

Queuing delay is considered in optimizationalong with transmission delay and packet er-ror

No uniform cross layer information exchangeformat is provided Besides that cross layersignaling may result extra overhead in thenodes

[81] Path-switching and packet-recovery method

Provides fast switching and recovery methodin multi-RAT environment

It depends on availability of good channel forpath switch and packet recovery may affectthe resiliency

[82] Diversity gain Option of various diversity gains such asspace time and spatial gain for low latencytransmission

Gain depends on various aspects such as beamforming beam training and antenna array

[123] Beam forming usingmmWave

Design and analysis of MAC layer underrealistic conditions

Proper channel model in mmWave is underdevelopment

[124]ndash[129] Full duplex communicationin same channel

Improves throughput reduces latency and up-holds PHY layer security

Crosstalk between the transmitter (Tx) and thereceiver (Rx) internal interference fadingand path loss

TABLE IX OVERVIEW OF SOLUTIONS IN CONTROL SIGNALING FOR LOW LATENCY

References Techniques Merits Demerits[83] Control channel sparse encod-

ing (CCSE)Uses non-orthogonal spreading sequences Needs field test for further validation

[84] Dedicated UL CCHs Provides CCH for sporadic packets with smallsize scheduling request (SR)

Requires dedicated CCH in each TTI and welldesigned scheduling request (SR) detector atBS Also considered scenario with UL andDL signal space of 10 and 40 bits spatialdiversity of 16 and bandwidth 10 MHz maynot be always feasible

[130] Sub slotted data and controlchannel

Two symbols are used in each subslot whichis compatible with current LTE

Reliability issue is not addressed

[85] SS-PUCCH consists of SC-FDMA symbol

More robust to channel fading compared toreference signal based PUCCH Symbol levelfrequency hoping harnesses frequency diver-sity gain with enhanced reliability

Need to be validated by field test

[86] Radio bearer and S1 bearermanagement

Control overhead and latency are decreasedfor both light and heavy traffic It ensures100 accessibility of all UEs

The technique is more suitable for very largetraffic networks such as vehicle networks

[87] Outer-loop link adaptation Besides reduction of latency for small pack-ets it can boost throughput just after changingfrom idle state to connected state

Needs field test for further validation

[131] Slotted TTI based radio re-source management

It can be implemented as an extension ofLTE-A

Validation through simulation and field test isnot presented

[132] Adaptive radio link control(RLC)

Besides latency reduction it improvesthroughput and reduces processing power

Control and data plane need to be separated

[133] SDN based control plane opti-mization

Using bandwidth rebating strategy balancebetween cost and performance is maintained

For large number of players (vehicles) thegame can be complicated Also real worldfield test is required for performance evalu-ation

[134] SDN based X2 signaling man-agement

It reduces signaling overhead and handoverlatency

The approach has been investigated for onlyfemtocells

[135] Inter BS data forwarding andmake-over-handover

This technique reduces X2 communicationprocessing and reconfiguration delays

Cases such as packet loss handover failurescenario with poor communication link arenot considered for performance evaluation

[136] Optical connected splitterswith dynamic bandwidthallocation allocation andtailored MAC protocol

It ensures X2 latency less than 1 ms Needs field test for further validation

single control signal in order to configure radio bearers suchas SRB1 SRB2 and DRBs that decreases the signaling inter-action rounds between the UE and the eNBs In [87] a newouter-loop link adaptation (OLLA) scheme is proposed Thescheme controls the size of the compensation in the estimatedSINR based on the time elapsed after a UE transits from anidle state to an active state which helps to reduce latencyfor small packet applications The study [131] proposed aslotted TTI based radio resource management for LTE-A and5G in order to achieve low latency The approach can servelow latency services utilizing short TTI and enhance downloadcontrol channel (ePDCCH)

The study [132] proposed a novel mechanism that intro-duces an adaptive radio link control (RLC) mode which dy-namically alternates between unacknowledgment mode (UM)and acknowledgment mode (AM) according to the real-timeanalysis of radio conditions This technique reduces systemlatency and processing power and improves throughput usingUM On the other hand it improves data reliability by ac-tivating AM during the degraded radio conditions In [133]SDN based control plane optimizing strategy is presented tobalance the latency requirement of vehicular ad hoc network(VANET) and the cost on radio networks The interactionbetween vehicles and controller is formulated and analyzed asa two-stage Stackelberg game followed by optimal rebating

strategy which provides reduced latency compared to othercontrol plane structures

In [134] SDN-based local mobility management with X2forwarding is proposed where total handover signaling isminimized by reduction of inter node signaling exchangesand X2 signaling forwarded to centralized SDN system Thisapproach can reduce handover latency while reducing of signaloverhead In [135] QoS CQI and other parameter based datautilization is proposed among eNBs to reduce X2 latencyprocessing and reconfiguration delays Additionally make-before-handover is proposed for low latency 5G servicesfor no data interruption In order to meet stringent latencyin X2 interface enhanced passive optical network (PON)based radio network is proposed in [136] where the splittersare connected thorough optical connections Following thisdynamic bandwidth allocation algorithm and tailored MACprotocol are introduced for achieving less than 1 ms latencyover X2 interface

F Symbol Detection

As illustrated in Fig 5 symbol detection encompassesvarious processes such as channel estimation and decodingwhich can all contribute into the overall latency The relatedliterature in the symbol detection side for latency reductionare tabulated in Table X

TABLE X OVERVIEW OF SOLUTIONS IN SYMBOL DETECTION FOR LOW LATENCY

Reference TechniqueApproach Merits Demerits[88] SM-MIMO detection with ZF

and MRC-ZFSignificant improvement of SINR and latencyis achieved compared to other schemes Alsothe method is validated in live environmentdesigned by SDR platform

Availability of large number of antennas isassumed

[89] Linear MMSE receiver Reduces latency along with throughput gainimprovement

It is not clear how much latency can bereduced in this scheme

[90] Space-time encoding andwidely linear estimator

Significant improvements in terms of symbolerror rate and latency over earlier works

Complexity at receiver side is increased

[91]ndash[93] Compressed sensing CS algorithm exhibits reduced complexityand increases reliability It is compatible withthe current LTE systems as well and requiresless measurement (resource) to decode controlinformation It provides sub-Nyquist samplingmethod for reconstruction of sparse signalefficiently in a linear system

It is challenging to design CS and sparserecovery system considering diverse wirelessconditions and input conditions

[94] Low complexity receiver inSCMA system

The prototype triples capacity while maintain-ing low latency

More suitable for MTC Needs field test forfurther validation

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Fig 5 Transmission and symbol detection in cellularnetwork

In [88] a low-complexity and low-latency massive SM-MIMO detection scheme is introduced and validated us-ing SDR platforms The low complexity detection schemeis proposed with a combination of zero forcing (ZF) andmaximum-ratio-combining-zero-forcing (MRC-ZF) In [89]a linear minimum mean square error (MMSE) receiver ispresented for low latency wireless communications using ultra-small packets The estimation of receiver filter using thereceived samples is proposed during the data transmissionperiod in lieu of interference training period Additionallysoft decision-directed channel estimation is argued using thedata symbols for re-estimation of the channels In [90] space-time encoding is introduced within a GFDM block in order to

achieve transmit diversity for overall low latency in the systemOn the other hand a widely linear estimator is used to decodethe GFDM block at the receiver end which yields significantimprovements in terms of symbol error rate and latency overearlier works

In [91]ndash[93] compressed sensing is proposed for latencyreduction in networked control systems if the state vectorcan be modeled as sparse in some representation domain In[94] a low complexity receiver design is proposed and thesuperiority of an SCMA system is verified via simulations Inaddition it is demonstrated with a real-time prototype that thewhole system throughput triples while maintaining low latencysimilar to flexible orthogonal transmissions

G mmWave Communications

Carrier aggregation using the mmWave spectrum is widelyconsidered to be a promising candidate technology for 5Gcapable of providing massive bandwidth and ultra low latencyThe mmWave technology is especially critical for VRAR typeof applications which require high throughput and low latencyThe works in mmWave spectrum for achieving low latency issummarized in Table XI

In [95] a new frame design for mmWave MAC layer isintroduced which provides several improvements includingadaptable and smaller transmission intervals dynamic loca-tions for control signals and the capability of directionalmultiplexing for control signals (dynamic HARQ placement)It addresses ultra low latency along with the multiple usersshort bursty traffic and beam forming architecture constraintsThe study [96] focuses on three critical higher-layer designareas low latency core network architecture flexible MAC

TABLE XI OVERVIEW OF SOLUTIONS IN MMWAVE COMMUNICATIONS FOR LOW LATENCY

References Techniques PHY layer MAC layer NET layer

[95] mmWave based MAC layer frame structure

[96] Low latency core network architecture flexible MAC layer and congestioncontrol

[97] mmWave based physical layer air interface with basic numerology and logicalchannel arrangement

[98] Low latency frame structure with beam tracking

TABLE XII LOCATION-AWARE COMMUNICATIONS FOR LOW LATENCY

References Techniques Merits Demerits[101] Location information uti-

lization in protocol stackLatency scalability and robustness can beimproved

Location accuracy spatial channel model-ing balancing trade-off between locationinformation and channel quality metric arechallenging

[99] Physical layer parametersdesign using FFT frameduration and local area(LA) physical channel

Improves spectral and energy efficiencyalong latency reduction

[100] Utilization of channelquality and trafficstatistics from smallcell

Coexistence capability with overlay LTE-Anetwork sleeping modes contention baseddata channel channel quality indicator andinterference statistics

The technique is more feasible for smallcells

layer and congestion control Possible solutions to achieveimprovements in these critical design areas are short symbolperiods flexible TTI low-power digital beam forming forcontrol and low latency mmWave MAC which can all beconsidered for data channel downlink control channel anduplink control channel

In [97] in order to decrease the latency of the systemtwo different physical layer numerologies are proposed Thefirst approach is applicable for indoor or line of sight (LOS)communications and the second one is suitable for non lineof sight (NLOS) communications This is justified by somechannel measurements experiments in 28 minus 73 GHz rangeIn [98] a 5G mmWPoC system is employed to evaluate thethroughput functionality in field tests at up to 20 kmh mobilespeed in an outdoor LOS environment Additionally someimprovements for a frame design is obtained which decreasethe latency in the field tests In the experiments it is observedthat the new slotted frame design can decrease the RTT to3 ms for 70 minus 80 of the cases in experiments alongsidethe observed throughput up to 1 Gbps

H Location-Aware Communications for 5G Networks

Location knowledge (in particular the communication linkdistance) can be considered as a criterion of received powerinterference level and link quality in a wireless networkTherefore overhead and delays can be reduced with location-aware resource allocation techniques because of the possibilityof channel quality prediction beyond traditional time scalesThe literature on location-aware communications regardinglow latency are tabulated in Table XII

In [101] several approaches are presented for monolithiclocation aware 5G devices in order to identify correspondingsignal processing challenges and describe how location datashould be employed across the protocol stack from a bigpicture perspective Moreover this work also presents sev-eral open challenges and research directions that should besolved before 5G technologies employ mmWave to achievethe performance gains in terms of latency connectivity andthroughput In [99] 5G flexible TDD is proposed for localarea (5GETLA) radio interface with FFT size of 256 and512 and short frame structure to achieve latency lower than1 ms The packets of size of less than 50 kbits can betransmitted with E2E latency of 025 ms The main focus in

designing physical layer parameters is on FFT frame durationand physical channel (LA)

In [100] a novel numerology and radio interface architec-ture is presented for local area system by flexible TDD andframe design The proposed framework ensures coexistencewith overlay LTE-A network sleeping modes contentionbased data channel and channel quality indicator and interfer-ence statistics Here the channel quality and traffic statisticsare accumulated from the small cells which can help togain high throughput and low latency Especially in order toreduce the latency the delay due to packets containing criticaldata for the higher layer protocols for instance transmissioncontrol protocol (TCP) acknowledgment (ACK) packets mustbe optimized To do so one possible approach is to carryout the retransmissions as quick as possible compared to thehigher layer timers Moreover capability of data transmissionto a contention based data channel (CBDCH) can play a keyrole here As a result by introducing CBDCH in small cellsthat are not highly loaded the average latency of small packetstransmission can be decreased considerably

I QoSQoE Differentiation

Differentiation of constraints on QoS and QoE can maintainlow latency in 5G services including ultra high definition and3D video content real time gaming and neurosurgery The

TABLE XIII LITERATURE OVERVIEW RELATED TOQOSQOE DIFFERENTIATION

Reference TechniquesApproaches QoS QoE

[102] mmWave utilization with beam tracking

[103] SDN and cloud technology

[104] QoS-aware multimedia scheduling

[105] Client based QoS monitoring architec-ture

[106] Colored conflict graph

[107] QoS architecture with heterogeneousstatistical delay bound

[108] Dynamic energy efficient bandwidth al-location scheme

[109] Predictive model based on Internetvideo download

[110] Routing using proximity information

[111] Predictive model based on empiricalobservations

related literature on QoS and QoE control for low latencyservices are tabulated in Table XIII

Abundance of mmWave bandwidth and extensive use ofbeamforming techniques in 5G will allow high QoS and QoEovercoming the resource and sharing constraints [102] How-ever current transmission protocols and technologies cannotbe employed simply for addressing technical issues in 5GThe mapping of diverse services including latency criticalservice to the optimal frequency SDN and cloud technologiescan ensure to achieve the best QoS and QoE as discussedin [103] In [104] a QoS-aware multimedia scheduling ap-proach is proposed using propagation analysis and propercountermeasure methods to meet the QoS requirements in themmWave communications Mean opinion score (MOS) whichis a criterion for user satisfaction can be employed for func-tionality evaluation of the newly presented QoS approach andwell-known distortion driven scheduling in different frequencyranges

Client based QoS monitoring architecture is proposed in[105] to address the issue of QoS monitoring from serverpoint of view Different criteria such as bandwidth error rateand signal strength are proposed with the well-known RTTdelay for maintaining desirable QoS A colored conflict graphis introduced in [106] to capture multiple interference andQoS aware approaches in order to take the advantage ofbeamforming antennas In this case reduction in call blockingand handoff failure helps to have a better QoS for multi classtraffic Each device can be sensitive to time based on itsapplication This can be considered as an issue for QoS provi-sioning To address this a novel QoS architecture is presentedin [107] with heterogeneous statistical delay bound over awireless coupling channel The authors presented the dynamicenergy efficient bandwidth allocation schemes in [108] whichimprove system quality significantly and maintain QoS

Previous QoS criteria which consist of packet loss ratenetwork latency peak signal-to-noise ratio (SNR) and RTT arenot sufficient for streaming media on Internet and thereforeusersrsquo perceived satisfaction (ie QoE) needs to be addressed[109] [137] Higher QoS may not ensure the satisfactoryQoE Different routing approaches of video streams in themobile network operatorsrsquo scenario is discussed in [110] forsubstantial refinement in QoE considering bit rate streamslow jitter reduced startup delay and smoother playback Apredictive model from empirical observations is presented in[111] to address interdependency formulated as a machinelearning problem Apart from that a predictive model of userQoE for Internet video is proposed in [109]

J CRAN and Other Aspects

Cloud radio access network (CRAN) (as illustrated in Fig 6)is introduced for 5G in order to reduce the capital expendi-ture (CAPEX) and simplify the network management [138]CRANs combine baseband processing units of a group of basestations into a central server retaining radio front end at thecell sides However this requires connection links with delayof 250 micros to support 5G low latency services In order tomeet strict latency requirements in CRAN two optimization

Virtual BTSpool

Backhaul

Corenetwork

Cloud-RANOpticalWireless

Fig 6 Cloud-RAN architecture in 5G networks

techniques including (i) fine-tuned real-time kernel for pro-cessing latency and (ii) docker with data plane development kit(DPDK) for networking latency have been proposed in [112]The experimental results clearly demonstrate the effectivenessof the approaches for latency optimization In [113] split ofPHY and MAC layer in a CRAN with Ethernet fronthaul isproposed and verified through experimental test followed bylatency interpretation It is found that latency for packets ofsize 70 and 982 bytes is 10732 micros and 12818 micros confirming20 latency increase from small to large packet The promis-ing results affirm that latency critical services in 5G can besupported by CRAN In [114] it is demonstrated that basedon experimental results from Wi-Fi and 4G LTE networksoffloading the traffic to cloudlets outperforms the responsetimes by 51 in comparison to cloud offloading

The CRAN can utilize backhaul information to redistributeusers for QoE maximization and adaptation of temporalbackhaul constraints Corresponding to this authors in [115]proposed a centralized optimization scheme to control thecell range extension offset so as to minimize the averagenetwork packet delay In [116] a CRAN on the basis ofthe optical network (PON) architecture is presented whichis called virtualized-CRAN (V-CRAN) The proposed schemecan dynamically affiliate each radio unit (RU) to a digital unit(DU) which results in coordination of multiple RUs with theircorresponing DU Moreover definition of virtualized BS (V-BS) is brought up which is able to mutually send shared signalsfrom several RUs to a user V-CRAN can reduce latency forjoint transmission due to the following reasons First it canprovide more than enough bandwidth for data transmissionbetween RUs and DU Furthermore for each DU a dedicatedhardwaresoftware is assigned that can be utilized by joint

transmission controller in order to provide data and signalingfor RUs The last but not least in order to handle the loaddistribution between DUs the virtualized PON can connect aDU directly via a linecard (LC)

Though mmWave will be the major contributor in attaining5G goals spectrum below 6 GHz is always the primary choicedue to less attenuation supporting long distance and antennacompatibility Moreover conventional cellular networks areusually deployed within the expensive licensed bands andthey use reliable core networks that are optimized to providelow-volume delay-sensitive services such as voice Howeverwith appearance of high-volume delay-insensitive resource-hungry applications including multimedia downloads suchconventional networks may not be cost-effective anymore[139] [140] Such concerns can be tackled or at least partiallyaddressed by spectrum sharing in 5G network improvingspectrum and energy efficiency along with QoSQoE control[] [141]ndash[145] In [146] in order to exploit the TV whitespace for D2D communications underlying existing cellularinfrastructure a framework is proposed A location-specificTV white space database is proposed in which D2D servicecan be provided using a look-up table for the D2D linkso that it can determine its maximum permitted emissionpower in the unlicensed digital TV band to avoid interferenceIn [147] a QoE driven dynamic and intelligent spectrumassignment scheme is proposed which can support both celland device level spectrum allocation This technique enhancesnot only the spectrum utilization but also can maintain desiredQoE including latency aspect The optimization problem ofnetwork sum rate and access rate with resource allocationand QoS constraints in D2D communication is presented in[148] To solve this a fast heuristic algorithm is proposedto reduce computational complexity resulting desired QoSsuch as latency In [149] a game theory and interferencegraph based optimization problem considering user schedul-ing power allocation and spectrum access is presented withan aim to maximize user satisfaction across the network Twoalgorithms including spatial adaptive play iterative (SAPI)learning are proposed to achieve Nash equilibrium In [150]a multi agents based cognitive radio framework is proposedfor effective utilization of spectrum and meet the goals of 5Gincluding latency Here sensing capability of secondary users(SU) is replaced by spectrum agents (SA) where the user canswitch between SU mode and SA mode based on availablespectrum information

Before starting the packet transmission in data applications

a tolerable initial delay can be considered Thus this shortdelay can be employed to decrease the energy required tooperate the small cell base stations (SBSs) With the goalof average power consumption reduction of SBSs when theload is low under-utilized SBSs can be are switched off As aresult the users and the network can be able to save energy bypostponing transmissions By doing so the users have time towait for an SBS with better link quality It can be observed thatsleeping mode operation for energy efficiency improvementsin SBSs will also introduce a source of latency In [173]the energy-efficiency versus delay trade-off is investigated andoptimality conditions for UErsquos transmit power is derived Bypostponing the access of the users energy efficiency of thesystem can be improved The optimal threshold distance isderived in order to minimize the average distance betweena user and an SBS Simulation results demonstrate that theenergy consumption of SBS can be reduced by about 35 ifsome of the SBSs are switched off

In [174] authors introduced an energy efficient low-complexity technique for load-based sleep mode optimizationin densely deployed 5G small cell networks By defining a newanalytic model the distribution of the traffic load of a smallcell is characterized using Gamma distribution It is shownthat the network throughput can be improved significantlywhile some amount of energy is saved by taking the benefitof the initial delay In [175] impact of average sleeping timeof BS and association radius on the mean delay in an UDNis investigated using a MG1N queuing model An explicitequation of delay is derived and the effect of average sleepingtime and association radius on the mean delay is analyzed

In [176] the remote PHY (R-PHY) and the remote PHY-MAC (R-PHYMAC) based modular broadcast cable networkis proposed for the access network In this architecture R-PHYMAC can achieve lower mean upstream packet delaycompared to R-PHY for bursty traffic and long distanceover 100 miles In [177] a virtual converged cable accessplatform (CCAP) system and procedure is proposed for hybridfiber coaxial (HFC) cable network In this method a newdigital optical configuration is introduced to receive datapackets with capability to convert them into RF waveformsThis method improves space and power requirements whileenhancing operational flexibility In [178] a novel remoteFFT (R-FFT) module is proposed which can perform physicallayer processing of FFT module towards RF transmissionThis module reduces fronthaul bit-rate requirement for CRANwhile providing solution for unified data over cable service

TABLE XIV OVERVIEW OF TECHNIQUES IN CORE NETWORK FOR LOW LATENCY

CaseArea Reference Approach Summary[52] [153]ndash[156] [157]ndash

[166]SDN-based architecture The architecture of 5G is proposed based on SDN with 5G

vision to meet large throughput massive connectivity and lowlatency

Core Network Architecture [157] [159]ndash[161][166]ndash[172]

NFV-based architecture NFV invalidates the dependency on hardware platform andmakes easy deployment of EPC functions as well as thesharing of resources in the RAN This can reduce the E2Elatency with improved throughput performance

[52] [161] [167] [169][170] [183]

MECfog-based network MECfog provides computation and storage near user end andalso separates the data plan from control plan These reducesthe latency

eg OpenFlow ForCESSouthbound Interface (SBI)

eg RESTful JSONNorthbound Interface (NBI)

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

SDN Controllers (eg ODL ONOS)

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

Virtualization Layer

Compute Storage Network

(b)

Fig 7 Simplified example architectures for core network (a) Architecture of SDN [151] and (b) Architecture of NFV [152](APP application ODL opendaylight platform ONOS open network operating system DFE dyna forming engineering

OSS operations support systems BSS base station subsystem EMS element management system NFVO NFVorchestrator VNFM virtual network function manager VIM virtualized infrastructure manager)

interface specification (DOCSIS) and LTE service over HFCcable network

VI CORE NETWORK SOLUTIONS FOR LOW LATENCY

To meet the vision of 5G encompassing ultra low latencyin addition to enhancements in the RAN drastic changes arealso proposed in the core network The new core networkincludes some new entities such as SDN MEC and NFVas well as new backhaul techniques [11] [12] [179] Theseenhancements aim to reduce the processing time bypassseveral protocol layers and ensure seamless operation Thecore network solutions for low latency are reviewed in furtherdetail in the rest of this section

A Core Network EntitiesThe SDN and NFV are assumed to be the main candidates

for the design of 5G core network [180] Based on thisin this section we mainly focus on the role of SDN andNFV technologies in latency reduction in 5G core networkExemplary architectures of SDN and NFV of the 5G corenetwork are illustrated in Fig 7 The existing literature on thecore network entities that can facilitate to achieve low latencyare summarized in Table XIV

The EPC which is developed by 3GPP for the LTE cellularnetwork has some limitations which affect the latency of theoverall system One concern is that the control plane and thedata plane in the EPC are not fully separated There is a levelof coupling between Serving Gateway (SGW) and PacketData Network Gateway (PGW) Decoupling of control planeand data plane seems necessary because they have differentnetwork QoS criteria to be met In particular the controlplane needs low latency to process signaling messages whilethe data plane requires high throughput to process the dataThus in order to design such planes efficiently it is preferableto decouple them completely Based on the literature SDN

and NFV can be employed in EPC architecture in order todecouple data plane and control plane and have a seamlessoperation of core network functions [181] [182]

After modification of NFV based EPC in which the wholenetwork elements are implemented using softwares running onVirtual Machines (VM) control plane and user plane can beseparated by employing SDN in EPC An SDN controller canact as an interface between the decoupled planes In addition toseveral advantages of SDNNFV-based user plane and controlplane separation including independent scalability flexibilityof flow distribution and better user mobility managementsuch a decoupling can have considerable effect on reducingthe latency as well This plane decoupling can facilitatethe mobile edge computing technology which decreases thelatency However adding an SDN controller to the networkcan be another source of the latency for the system On theother hand the scalability of SDN controller can be addressedby deploying several controllers Thus there is a trade offhere between the scalability of controllers and latency increasewhich should be considered in the design process for specificapplications [158]

Another limitation is that the data plane of the LTE EPCis implemented in a centralized manner Even the users thatneed to communicate locally have to transmit their traffic ina hierarchal system ending with few number of centralizedPGWs which increases the E2E latency Although centralizedimplementation of network can facilitate the management andmonitoring the network by operators it increases the E2Elatency which can not meet the applications requiring lowlatency including autonomous driving smart-grid or automatedfactory Thus this kind of implementation leads to inefficientsystem performance and high latency which can not meet the5G vision [183]

Recently by emergence of the new technologies such ascloud computing fog networks mobile edge computing NFVand SDN the implementation of the network can be moredistributed [159] [160] By employing such technologies theCAPEX and OPEX of the network can be reduced consid-erably Moreover by bringing the elements of core networkcloser to the users the E2E delay can be decreased sig-nificantly In [161] authors proposed SDNNFV-based MECnetworks algorithms that can enable the data plane to create adistributed MEC by placement of network functions at a dis-tributed manner They demonstrated that the proposed schemecan reduce the redundant data center capacity around 75and meet the 5G latency requirement along with considerablebackhaul link bandwidth reduction

Mobility management in core network based on SDNcan potentially introduce some delays In [162] the maincontributers for processing delays in an SDN-based mobilitymanagement system is discussed By implementing twoproactive and reactive solutions for mobility managementusing Mininet and OpenFlow it is observed that withhigh probability (almost 95) in the proactive mobilitymanagement system the overall processing latency is aroundthe median value By visualizing all of EPC entities asdecentralized VMs in different locations in [171] a carriercloud architecture is introduced To improve the E2E latencyof the users the concept of Follow-Me-Cloud is presentedThe main point of this concept is that all parts of the networkcan keep track of the movement of the user which results inseamless connectivity and lower E2E latency In [163] [164]the authors proposed a decentralized scheme for controlplane called SoftMow which is a hierarchical reconfigurablenetwork-wide control plane The proposed control planeincludes geographically distributed controllers where eachcontroller is responsible to serve the network in its particularlocation The number of the levels in this hierarchical schemecan be designed based on the available latency budgets

In [52] authors presented an LTE compliant architectureto decrease delay for combination of fog networks MECand SDN in which the architecture is supposed to take theadvantage of NFV in the evolved packet core (EPC) func-tions Following this optimization of general packet radioservice (GPRS) tunneling protocol (GTP) is introduced forsupporting low latency services GTP tunnels management isaccomplished by a novel element between the eNB and themobile network interface with the Internet In [153] [154]SDN is proposed along with some changes in the existing4G architecture for moving forward towards 5G The changesinclude reduction of number of serving gateways (S-GW) andelimination of some protocol layers In SDN based systemvirtualization is possible and routes can be optimized Thiswill allow handling of QoS by setting specific rules in theswitches along the data path Network coding integrated withSDN is proposed in [155] for low latency and reduced packet-retransmission Network coding can work as network routerand can be integrated with SDN which provides seamlessnetwork operation and reduction in latency

The NFV is proposed as a major entity of 5G core networkin [157] [167]ndash[170] NFV removes the dependency on thehardware platform and makes flexible deployment of EPCfunctions as well as sharing of resources in RAN This canreduce the E2E latency with improved throughput perfor-mance SDN and NFV based 5G architecture with enhancedprogrammability of the network fabric decoupled networkfunctionalities from hardware separated control plane fromdata plane and centralized network intelligence in the networkcontroller is presented in [157] In [167] an informationcentric scheme is presented in order to integrate the wirelessnetwork virtualization with information centric network (ICN)In this architecture key components such as wireless networkinfrastructure radio spectrum resource virtual resources (in-cluding content-level slicing network-level slicing and flow-level slicing) and information centric wireless virtualizationcontroller have been introduced which can support low latencyservices

An NFV-based EPC is introduced in [184] which is an EPCas a service to ease mobile core network (EASE) In thisscheme the elements of the EPC are visualized using VMsIn [166] a simple implementation for EPCaaS is proposed inwhich one of the drawbacks is the increment in the latencybetween the EPC and virtual network function components Toaddress such an issue in [172] the main idea is to partition thevirtual network functions into several subsetsgroups based ontheir interaction and workload in order to reduce the networklatency It should be noted that employing the decentralizedcontrol plane and having it closer to the users at the networkedge can be helpful for applications with high mobility andlow latency However it can introduce some issues related topolicies and charging enforcements which should be consid-ered for network design based on the requirements In [169]authors presented the optimization problem of composingcomputing and networking virtual functions to select thosenodes along the path that minimizes the overall latency (ienetwork and processing latency) The optimization problem isformulated as a resource constrained shortest path problem onan auxiliary layered graph followed by initial evaluation

In [185] the authors proposed a detailed approach to

TABLE XV PROPOSED CORE NETWORK ENTITIES FOR5G VISIONS

Reference SDN NFV MEC Fog net-works

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

implement the big data empowered self organizing network(SON) in 5G The procedure to employ the data based onmachine learning and data analytics is demonstrated in orderto create E2E visibility of the network for implementation of amore efficient SON This approach can meet the stringent 5Grequirements such as low latency In [165] the smart gateway(Sm-GW) is employed for scheduling the uplink transmissionsof the eNBs Based on simulations it is demonstrated thatthe Sm-GW scheduling can allocate the data rate in uplinktransmission to the eNBs in a fair manner along with reducingpacket delays Traffic of heavily loaded eNBs can make thebuffer of the queue of a Sm-GW full which results extralatency This situation happens due to the massive numberof connected eNBs to a single Sm-GW However by usingeffective scheduling connection with Sm-GW can be dis-tributed among eNBs while maintaining QoS The proposedcore network entities for 5G vision are summarized in TableXV

B Backhaul Solutions

Backhaul between base stations and the core network carriesthe signaling and data from the core and the Internet Dueto the enormous number of small cells and macro cell basestations supporting 1000x capacity massive connectivity andlatency critical services in 5G the capacity of backhaul isa bottleneck for achieving low latency At current scenariomicrowave copper and optical fiber links are used for backhaulconnections based on availability and requirements 5G back-haul requires higher capacity lower latency synchronizationsecurity and resiliency [15] Referring to Table XVI we divideexisting backhaul solutions into 2 parts (1) General backhauland (2) mmWave backhaul The solutions are described asfollows

1) General backhaul General backhaul includes a dynamicGPRS tunneling protocol (GTP) termination mechanism thatcombines cloud based GTP with a quick GTP tunnel proposedin [156] Based on the user request or other factors thesystem can change its mode from a cloud-based GTP tunnelto the quick GTP tunnel In [52] a 5G vision compliantarchitecture is presented to reduce latency combining withthe fog network MEC and SDN The optimization of GTP

tunnels is accomplished by a novel element acting as anintermediate node between eNB and the mobile networkinterface accompanied with the Internet In [186] modifiedVLC technology is used to set up an optical window (OW) linkfor low-cost backhaul of small cells to achieve a latency of 10ms Moreover using a next generation baseband chipset E2Elatency below 2 ms can be achieved An efficient PON-basedarchitecture is proposed in [187] that offers ultra-short latencyfor handovers by enhancing connectivity between neighboringcells Additionally the authors propose a tailored dynamicbandwidth allocation algorithm for a fast handover betweeneNBs which are associated to the same or diverse opticalnetwork units

In transport networks latency requirement plays a keyrole For instance the main requirement for machine typecommunications is the latency that should be kept as low aspossible Therefore efficient design of 5G transport networksis critical In [188] a detailed perspective of the 5G crosshauldesign is proposed in order to introduce the key goal oftransporting the backhaul and fronthaul traffic in a unifiedpacket-based transport network based on MAC-in-MACEthernet Moreover the SDNNFV-based 5G-crosshaulcontrol plane architecture is investigated which decouples thelogically centralized control plane and data plane This cancontribute in latency reduction aspect In [189] two candidatetechnologies for 5G transport networks are presented One ofthem is based on the over-provisioning of transport resourceswhile the second architecture is based on dynamic resourcesharing and NFVSDN-based controller to handle the latencyrequirements

In [190] SDN and cache enabled heterogeneous networkis proposed where C-plane and U-plane are split The cachesof macro and small cells are overlayed and cooperated ina limited backhaul scenario while ensuring seamless userexperiences for coverage low latency energy efficiencyand throughput In [189] two candidate technologies for5G transport networks is presented to handle the latencyrequirements in which one of them is based on the over-provisioning of transport resources while the secondarchitecture is based on dynamic resource sharing and NFV

TABLE XVI OVERVIEW OF LITERATURE IN BACKHAUL SOLUTIONS TO ACHIEVE LOW LATENCY

Category Reference ApproachTechniquesGeneralbackhaul

[156] A dynamic GTP termination scheme combining cloud based GTP with a quick GTP tunnel with a dedicated hardware

[52] GTP tunnel optimization by a new component in 5G complaint network consists of fog networks MEC and SDN[186] Modified VLC technology to set up an OW link for low-cost back hauling of small cells[187] PON-based architecture with a tailored dynamic bandwidth allocation algorithm[188] MAC-in-MAC Ethernet based unified packet-based transport network[189] The first architecture is based on over provision transport network whereas the second one is based on dynamic

sharing SDN and NFV controllers[190] SDN and cache enable architecture for limited backhaul secererio

mmWavebackhaul

[191] mmWave for fronthaul and backhaul and split of control and user plane

[192] A digitally-controlled phase-shifter network based hybrid precodingcombining scheme for mmWave massive MIMO[193] A framework supporting of in-band point-to-multipoint non-Line-of-sight and mmWave backhaul[194] A mmWave based backhaul frame structure in 3 - 10 GHz carrier frequencies[179] [195] Ultra dense wavelength division multiplexing (UDWDM) passive optical networks (PONs) based backhaul solution

for mmWave networks

Local caching D2D caching SBS caching MBS caching

User 1 User 2 User 1 User 2 User 1 User 2 User 1 User 2

SBS

MBS MBS

SBS

MBS

SBS SBS

Fig 8 Different types of caching in 5G

and SDN-based controller In [190] SDN and cache enablederogenous network is proposed where C-plane and U-planeare spitted The caches of macro and small cells are overlayedand cooperated in a limited backhaul scenario while ensuringseam user experiences such as coverage energy efficiencyand throughput

2) mmWave Backhaul In addition to the presented solu-tions for backhaul mmWave employment in backhaul can beconsidered as a promising solution for latency reduction Inorder to have the enhanced user experience the BSs should bein touch with core network and all other BSs via a low latencybackhaul [221] In [191] the authors proposed a scheme thatemployed mmWave links as backhaul fronthaul and access inwhich a new separation method between control and user planeis proposed for 5G cellular network A reasonable split amongcontrol and user plane improve the user QoS by providingubiquitous high data rates in mmWave SBS coverage

In [192] to implement an ultra-dense network (UDN) forthe future 5G network and providing high data rates the needof a reliable gigahertz bandwidth and economical backhaulis emphasized Since mmWave can be easily integrated withmassive MIMO to improve link reliability and can providesufficient data rate for wireless backhaul it is a promisingcandidate for such a scenario Considering a massive MIMOscenario a hybrid precoding approach is considered in whicheach BS can cover several SBSs with multiple streams for eachSBS at the same time In [193] the authors proposed a solutionframework for supporting an in-band point to multi-pointNLOS mmWave backhaul in order to provide a cost-effective

and low latency solution for wireless backhaul It is shownthat an in-band wireless backhaul for inter BS coordinationis feasible while the cell access capacities are not affectedconsiderably

In [194] a frame design for mmWave communications isproposed for 5G SBS network radio interface in 3-10 GHzFor both of LOS and NLOS scenarios different frame designsare proposed which have a frame duration of 01 ms and005 ms respectively to achieve low latency The proposedLOS structure can be assumed as a suitable solution for shortdistance indoor wireless access or in-band backhaul In orderto obtain high capacity and low latency backhauling the EUresearch project 5G STEP-FWD introduced a novel designthat deploys ultra dense wavelength division multiplexing(UDWDM) passive optical networks (PONs) as the backhaulof mmWave networks The proposed scheme is based on theultra-narrow wavelength spacing of the UDWDM technologyto provide seamless connectivity for dense small-cell networks[195] [222]

VII CACHING SOLUTIONS FOR LOW LATENCY

In addition to the shortage of the radio spectrum theinsufficient capacity of backhaul links can be considered as abottleneck for low latency communication The long delay canbe due to the requests of too many users in peak-traffic hoursThus latency reduction is crucial for usersrsquo QoS and QoE inthe 5G networks Caching and in a more general categoryinformation centric networking can be assumed as one ofthe promising candidate technologies to design a paradigm

TABLE XVII OVERVIEW OF LITERATURE IN CACHING

Aspect Reference SummaryContent caching [196]ndash[215] Filling of appropriate data is investigated by diverse techniques employing time intervals in which

the network is not congestedContent delivery [197] [200]ndash[218] Content delivery to requested users is presented by different approaches for reduction of latencyCentralized caching [196]ndash[198] [201] [203]

[205]ndash[211] [213]ndash[217]Various centralized caching is investigated with assumption that a coordinator with access to almostall the information about the storage capacities of different BSs the connectivity of the users andBSs and etc

Distributed caching [199] [200] [202] [204][212] [215] [218]

Various aspects of distributed caching has been investigated in order to minimize the communi-cation overhead among SBSs and the central scheduler

Latency-Storage trade-off [205]ndash[210] [219] [220] Fundamental trade off between storage and latency is investigated in radio networks complementedwith cache-enabled nodes

TABLE XVIII DIFFERENT TYPES OF CACHING SCHEMESFOR THE CELLULAR NETWORK

Cases Reference Description

Localcaching

[227]

When a UE wants to access a content it firstchecks in itself Once the content is confirmedin the local caching storage it is accessed bythe UE without any delay

D2Dcaching

[223]

If the requested content is not found locallyuser will seek it within the range of itrsquosD2D communication If it is found in nearbydevices it is delivered to the requester UE byD2D communication

SBScaching [228]

If the requested content is available in thelocal SBS the content is delivered to the UEby the local SBS

MBScaching

[223]If the content is not found in local cachingstorage nearby devices or SBS caching thecontent is delivered by MBS caching

shift for latency reduction in next generation communicationsystems [13] [14]

In this section referring to Table XVII we present a detailedoverview of caching concepts for cellular network followed byfundamental limits and existing solutions

A Caching for cellular networkLet us consider a scenario that a user requests content

from a content library F = f1 f2 fk where k is thenumber of files The files are sorted with popularity where f1

and fk are the most and least popular files respectively Thepopularity of a requested file l can be written [223] as

φl =lminusγsumki=1 l

minusγ (7)

where l isin 1 2 k and γ is the parameter for unevendistribution of popularity in F which follows Zipf distributionFor N eNBs B = BS1BS2 BSN with each eNBhaving capacity C the probability of caching of file fl byan eNB can be obtained as

PBl = 1minus eminusρσπR2

(8)

where ρ is the spatial density of eNBs following a Poissonpoint process [224]ndash[226] and σ is the probability that filefl is cached within B Then the total probability of gettingcontent from the eNB can be written as

PB =

Nsumi=1

φiPBi (9)

The probability of getting the content as in (9) is directlyassociated with the latency of downloading it and henceeffective caching strategies can help in significantly reducinglatency in 5G networks

The proposed caching schemes for mobile networks can bedivided into 4 categories (1) Local caching (2) Device todevice (D2D) caching (3) SBS caching and (4) Macro basestation (MBS) caching Each of these caching types can reducethe latency by providing the requested content for the users

using a way other than bringing it from the core network usingbackhaul links In fact each user starts from the nearest sourceto look for its desiered content and proceed until finding it inany of the proposed sources The different types of cachingfor cellular network are illustrated in Fig 8 followed by thesummarized descriptions presented in Table XVIII

B Fundamental Latency-storage trade-off in Caching

There are several fundamental limits for caching in mobilenetworks including latency versus storage memory versus rate[211] memory versus CSIT [212] storage versus maximumlink load [213] and caching capacity versus delivery rate[214] As defined in Table XIX from an information theoreticpoint of view authors employed the metrics such as normal-ized delivery time (NDT) fractional delivery time (FDT) anddelivery time per bit (DTB) for investigation of the latencystorage trade-off in caching networks In most of these works[199] [200] [202] [204] [212] [215] [218] for a givenscenario an upper bound or lower bound for the consideredmetric is derived in order to get useful insights of this trade-offThe summary of latency storage trade-off works is presentedin Table XX

The authors in [205] investigated the storage latency trade-off using a new metric called NDT This metric measures theworst-case latency that can happen in a cache-aided wirelessnetwork divided by that of an ideal system with unlimitedcaching capability Considering a general cache-aided wirelessnetwork the lower bound for NDT is presented in terms ofthe ratio of the existing file memory at the edge node and thetotal size of files for both perfect channel state information(CSI) and imperfect CSI

Authors in [206] employed NDT as well in order tocharacterize the trade-off between NDT and fronthaulcachingresources Using this information-theoretic analysis of fogradio access networks optimal caching front-haul transmissionis obtained In [207] the latency storage trade-off in a 3 times 3wireless interference network is investigated while all trans-mitters and receivers are equipped with caches Another metriccalled (FDT) is proposed in order to characterize the trade-off between latency and storage This information theoreticperformance metric is actually a refined version of the metricoriginally proposed in [205] The FDT can reflect the loadreduction as well In a similar work [208] the well-knownDoF metric is used which does not reflect the load reductionMoreover the proposed approach in [208] just considers the

TABLE XIX DEFINITION OF THE METRICS USED FORLATENCY EVALUATION IN CACHING SCHEMES

Cases Reference Defenition

NDT [205]Defined as asymptotic delivery delay per bit in thehigh-power long-blocklength case

DTB [210]Defined as the ratio between the duration of trans-mission in channel to the file size in bits for thevery large file size regime

FDT [209]Defined as the worst-case delivery latency for thereal load at a rate described by the DoF of thechannel

TABLE XX SUMMARY OF THE WORKS ONLATENCY-STORAGE TRADE-OFF IN CACHING

Ref Usedmetric

Description

[205] NDT Lower bound for NDT is derived for a general cache-enabled network for both perfect and imperfect CSI

[206] NDT The trade-off between NDT and front-haul andcaching resources is characterized and optimalcaching front-haul transmission is obtained

[207] FDT For a 3times3 wireless interference network the storage-latency trade-off is investigated while all transmittersand receivers are equipped with caches

[209] FDT For a scenario with a 3times 3 MIMO system in whichthe nodes are enabled with several antennas the tradeoff between storage and latency is investigated

[210] DTB DTB is used to characterize the system performanceas a function of cache storage and capacity ofbackhaul links connected to SBS

[219] NDT The trade-off between storage and latency for adistributed caching scenario in fog radio access net-works is characterized

[220] NDT Cloud-based compressed precoding and edge-basedinterference management introduced as two majortechniques for optimal performance of cloud andcaching resources in different cases

[229] NDT Considering heterogeneous timeliness requests de-pending on application the fundamental trade-offbetween the delivery latencies of different usersrsquorequests is characterized using NDT

[230] NDT Upper and lower bounds for minimum delivery la-tency as a function of cache and fronthaul resourcesis obtained over fronthaul and wireless link in a F-RAN with a wireless multicast fronthaul

[231] NDT Assuming an F-RAN when pipelined fronthaul edgetransmission is used lower and upper bounds on theNDT is presented

[232] NDT Assuming both content placement and deliveryphases in one time slot it is shown that the proposedapproach outperforms the offline caching

one-shot linear processing but interference alignment schemein [205] may require infinite symbol extension

In [208] an optimization problem is designed to minimizethe number of required communication blocks for contentdelivery Then a lower bound is proposed on the value of theobjective function Using the same metric the authors in [209]investigated the fundamental trade-off for a cache-enabledMIMO system Considering a scenario with a 3 times 3 MIMOsystem in which the nodes are enabled with several antennasthe trade off between storage and latency is investigated Inaddition to FDT and showing its optimality for some rangesof cache size the model can consider the effect of real trafficload at a rate specified by the DoF of the channel

In [210] a cellular network is considered with multipleSBSs with limited cache capacity in which there is interferenceamong them Here another information theoretic metric basedon delivery latency is defined as well in order to characterizethe system performance as a function of SBS cache memoryand capacity of backhaul links connected to SBS Using thismetric which is called DTB the trade off between latency andsystem resources is investigated In [219] using NDT trade-off between storage and latency a distributed caching scenarioin fog radio access networks is characterized In the presentedapproach a coded delivery scheme is proposed to minimize

the latency for delivering user demands for two edge-nodesand arbitrary number of users It is shown that using decen-tralized placement the presented delivery approach can obtaina considerable performance improvement in comparison to thederived lower bound

In [220] again NDT is employed to characterize the funda-mental trade off between delivery latency and system architec-ture Considering NDT as the criterion for latency evaluationof the system some bounds on its value are proposed In thelight of such bounds useful insights on the latency and storagetrade off are obtained It is demonstrated that in order to obtainthe lowest delivery latency cloud-based compressed precodingand edge-based interference management should be consideredas two major techniques for optimal performance of cloudand caching resources in different cases In [229] for a fogradio access network the heterogeneous timeliness requestsdepending on application is considered while in the existingworks the assumption is that all requests have identical latencyfor all files in the content library The fundamental trade-offbetween the delivery latencies of different usersrsquo requests ischaracterized using NDT The minimization of the averagedelivery latency as a function of the content popularity profileremains as a future research direction

In [230] the total delivery latency over fronthaul andwireless link in a fog radio access network with a wirelessmulticast fronthaul is investigated Again using NDT theoptimal delivery latency based on cache storage and fronthaulresources is formulated and upper and lower bounds areobtained It is shown that in contrast to the receiver-sidecaching coded multi-casting can not help in decreasing theNDT when two users and two edge nodes (ENs) are availableIn [231] for a F-RAN the NDT is used to characterize theperformance of a fog radio access network when pipelinedfronthaul edge transmission is used

In [232] a F-RAN system is considered in which ENsare cache-enabled with limited storage On the contrary tothe existing works that focused on offline caching whereboth caching phases are considered separately the proposedmethod can integrate both of them in each time slot Theperformance is characterized using NDT and compared tothat of optimal offline caching schemes It is shown that theproposed approach outperforms the offline caching

C Existing Caching Solutions for 5GIn general the file delivery service in mobile networks can

be classified into two parts cache placement and contentdelivery [233] In cache placement the cached content on theBSs is determined which is usually based on the amounts ofrequests from users Cache placement can be done using acentralized or distributed approach In centralized approacha coordinator is assumed with access to almost all the infor-mation about the memory size of BSs the connectivity ofthe users and the BSs However in some scenarios that thereis no central controller these schemes can not be applicableand a distributed cache placement is required [199] In eachof the caching phases whether centralized or distributed thedesign can reduce the delivery time of the requested contentand latency of the system

There are some efforts in the literature on investigationof different challenges raised up in centralized cache place-ment problems In [196] the cache placement problem wasinvestigated in a scenario including SBSs called helpers withweak backhaul links but large memory size In experimentalevaluation it is shown that the proposed scheme can achievea considerable performance improvement for the users atreasonable QoS levels In [198] authors aim at minimizingthe average download delay of wireless caching networks withrespect to caching placement matrix It is demonstrated that thebackhaul propagation delay can affect the caching placement

In some scenarios it is more desirable to design the cachingproblems in a distributed approach In [199] a distributedcache placement approach is proposed in order to minimizethe average download delay while some constraint for BSsstorage capacities are met The formulated optimization prob-lem which is NP-hard is solved using a belief propagationbased distributed algorithm with low complexity This opti-mization problem makes sense because there is a trade-offbetween latency and storage capacity in caching networksThe comparison between the performance of the proposeddistributed scheme with that of the centralized algorithm in[196] is presented as well In [200] a decentralized contentplacement caching scheme is presented Although there is nocoordination the proposed scheme can attain a rate as goodas the optimal centralized scheme proposed in [196] In [234]two caching and delivery schemes are considered The firstone operates in a centralized manner while the second one isbased on decentralized caching For both cases the trade-offbetween coded multi-casting and spatial reuse is reflected bythe code length

In addition to the aforementioned literature in [242] con-tent caching and content delivery schemes are proposed con-sidering cooperation to address the explosive enhancementsof demand for mobile network applications Defining theobjective as minimizing the average downloading latency it

is demonstrated that the proposed content assignment anddelivery policy scheme has a better performance in comparisonto the previous known content caching schemes in termsof average downloading latency A weighted optimizationproblem is formulated in [201] to minimize the traffic ofbackhaul and downlink while the constraints for cache memorysize and bandwidth limitation for D2D communication isconsidered It is shown in [218] that if latency awareness isconsidered in caching management it is an effective approachto reduce the delivery time of latency sensitive applicationsand the global delivery time of users In the proposed modeltwo main advantages is claimed First it not only has a betterperformance in term of delivery time at the end-user but alsoaffects the link load reduction Second a faster convergencewith respect to probabilistic caching is achieved In [202] theeffect of joint latency awareness and forwarding is investigatedin a cache-enabled network Authors proposed a scheme whichis based on caching and forwarding strategies in order toimprove E2E experienced latency by the UEs while there isno coordination among them

In [203] a cooperative content caching approach betweenBSs in cache-enabled multi-cell network is considered Due tothe trade-off between storage and latency cooperative cachingoptimization problem is designed in order to minimize theaverage delay while a constraint on the finite cache sizeat BSs is met It is shown that cooperation among cellscan considerably reduce delay in comparison to that of non-cooperative case Moreover the gains of the proposed schemewill be increased in more diverse and heavier load traffics In[204] the aim of the work is to minimize the data transmissiondelay for the P2P caching system while considering the effectof cache size all mobiles in a cell are considered as severalP2P caching groups Then the problem is formulated as astochastic optimization problem and solved using Markovdecision process (MDP) to obtain the optimal solution

In [243] a cooperative multicast-aware caching strategy is

TABLE XXI PLATFORM AND PERFORMANCE ON FIELD TESTS TRIALS AND EXPERIMENTS

Reference Evaluationmethodology

SDR DSP mmWave Conventional Proprietary LTE Remarks

[235] Trials of 5G concepts along with a novel air interface[236] 4 test cases and 15 KPIs is proposed[237] mmWave aggregation[63] RTT latency of 1 or 2 ms is achieved Moreover to

achieve latency on the order of couple of hundredsmicroseconds over air interface cross later approachis recommended

[238] Low latency VANET[239] DSP round-trip latency less than 2micros is achieved

for channel aggregation and de-aggregation for 4820 MHz LTE signals

[88] 20 times latency reduction in comparison to existingworks

[98] Minimum latency 3 ms[63] Latency le 1 or 2 ms[94]

[240] Latency le 17 ms[60] HARQ RTT le 15 ms[241] RTT latency le 1 ms

proposed for the BSs to decrease the average latency of contentdelivery in 5G cellular networks The proposed scheme iscarefully designed in order to take into account the benefitof multicast and cooperation while in the existing cachingschemes the popular content simply is brought close to theusers The optimization problem is formulated in order tominimize average latency for all the content requests It isdemonstrated that via various trace-driven simulations thatthe proposed cooperative multicast-aware caching scheme canprovide up to 13 decrease in the average content-accesslatency in comparison to multicast-aware caching scheme withthe same total cache capacity

In [244] the authors presented a cooperative caching ar-chitecture in which multiple locally cache-enabled nodes ofcloudlets interact cooperatively in a decentralized cloud ser-vice networks By proposing a content distribution strategy theproblem is formulated so that the mean total content deliverydelay for all users in the proposed scheme is minimized It isconfirmed that the approach can enhance the cache hit rateand also reduce the the content delivery latency in comparisonto existing solutions In [245] the E2E packet transmissionin a cache-enabled network is modeled in which both thewired backhaul and the RAN are jointly considered Theperformances of both the on-peak and the off-peak networkare investigated while both the wired backhaul and the RANare considered The E2E average packet latency is elaboratedwith the change of the request rate It is shown that the averagepacket latency reduces in comparison to that of the systemwithout caching ability due to the traffic offloading of thewired backhaul via caching

VIII FIELD TESTS TRIALS AND EXPERIMENTS

In this section we present some representative field teststrials and experiments for 5G low latency The related litera-ture is summarized in Table XXI where each of the individualreferences will be described in further detail below

The study [235] presents SDR based hardware platformto verify the concept of 5G This facilitates initial proof-of-concepts (PoC) of novel 5G air interface and other conceptsby extending hardware-in-the-loop (HIL) experiments to smalllaboratory experiments and finally trials of outdoor testsSuch an SDR based hardware can demonstrate high-capacitylow latency and coverage capabilities of LTE-A solutionsIn [236] evaluation methodology including some novel testenvironments and certain new key performance indicators arediscussed in order to evaluate 5G network Here four candi-date test environments such as indoor isolated environment andhigh speed train environment and fifteen key performance in-dicators such as latency throughput network energy efficiencyand device connection density are emphasized for performanceevaluation

In [237] 5G system operating at 15 GHz is presentedfollowed by some experimental results Here 02 ms subframe(14 OFDM symbols) is used for throughput latency and otherperformance evaluation The hardware implementation resultsof digital signal processing (DSP) and SDR based 5G systemfor low latency is presented in [63] In this study both the shortTTI (sTTI) frame structures and wider subcarrier spacings are

implemented in DSP platform Based on the configurations ofthe system RTT latency as low as 1 ms can be achievedHowever for achieving latency on the order of a few microsoptimization in between controllers and processing machinesneeds to be performed by cross-layer fashion Additionallythe tail latency is argued to be considered in strict latency re-quirements assessment while maintaining required reliability

An SDR based test bed is presented in [238] for co-operative automated driving with some experimental resultsfrom lab measurements It implements flexible air inter-face consisting of re-configurable frame structure with fast-feedback new pulse shaped OFDM (P-OFDM) waveformlow latency multiple-access scheme and robust hybrid syn-chronization which ensure low latency high reliable commu-nication Results of the experimental trials are presented in[239] which utilizes DSP techniques for channel aggregationand de-aggregation adjacent channel leakage ratio reductionfrequency-domain windowing and synchronous transmissionof IQ waveforms and code words used in control and man-agement function In the proposed experiment transmissionof 48 chunks of 20 MHz LTE signals using a common publicradio interface of capacity 59 Gbs can achieve RTT DSPlatency of less than 2 micros and mean error-vector magnitude ofabout 25 after fronthaul fiber communication This mobilefronthaul technique shows the path towards ultra low latencyintegrated fiberwireless access networks

In [88] a multi-terminal massive SM-MIMO system is eval-uated considering realistic scenarios The authors developed amassive SM-MIMO OFDM system prototype utilizing multi-ple off-the-shelf SDR modules which serve as IoT terminalsTwo linear detection schemes with diverse complexity levelswere tested for detection instead of maximum likelihood de-tection (MLD) schemes It demonstrates the similar real-timeSINR performance of the MLD techniques along with 20 timeslatency reduction over existing works The promising resultsurge the utilization of massive SM-MIMO systems for latencyreduction and reliability enhancement in IoT transmissions In[98] the performance of a lower latency frame structure wasevaluated in field tests using a 5G mmWave proof-of-concept(PoC) system It was found that the slot interleaving framestructure can achieve RTT latency of 3 ms in the 70minus80 ofthe trial course Additionally beam tracking and throughputperformance were evaluated in field tests at a speed up to20 kmh on LOS outdoor environment It was confirmed thatmmWave system can obtain throughput of 1 Gbps in the 38of the trial time at 20 kmh speed

In [94] a low complexity receiver design was introducedfollowed by verification of superiority of an SCMA systemvia simulations and real-time prototyping It can provide upto 300 overloading that triples the whole system throughputwhile still enjoying the link performance close to orthogonaltransmissions In [240] the concept of MEC was introducedfirst time for 5G followed by promising field tests The MECwas tested and analyzed on various cases including localbreakout and network E2E latency It was concluded thatMEC can support low latency services of not lower than 17ms It also urged that stricter requirement of latency needs

to be investigated from the new radio technologies or D2Dcommunication In [60] a lab trial is presented to study thefeasibility of ultra-low latency for 5G It is shown that 15ms HARQ RTT for TDD downlink in a lab trial is achievablewhen using the available test equipment in the literature whilea novel frame structure and the associated signaling procedureis employed The proposed scheme has 5 times better latencyperformance in comparison to the existing LTE-Advancedstandard

In [241] a novel frame structure is tested using a propri-etary quasi-static system simulator for ultra-dense 5G outdoorRANs In this regard a frame structure is designed in orderto facilitate low latency and multiuser spatial multiplexingon radio interface along with small-scale packet transmissionand mobility support It is found that performance of theintroduced 5G network is better than that of LTE in case ofair interface latency In particular considering UL schedulingrequests in the RTT latency the proposed frame structure canachieve latency as low as 08365 ms which is reduced by afactor of 5 in comparison to that of LTE This satisfies 5Glatency requirement (ie1 ms latency)

IX OPEN ISSUES CHALLENGES AND FUTURE RESEARCHDIRECTIONS

While there are some existing proposals to reduce latencyto 1 ms there are several open issues and challenges for futureresearch The area of exploration includes RAN core networkbackhauling caching and resource management Also theexisting techniques need to be validated in field tests andshould evolve from current LTE systems In the followingsection we discuss some of the open issues and challengeswhich needed to be explored and addressed by researchersfrom both academia and industry

A RAN IssuesAs discussed in Section IV most of the fundamental con-

straints for achieving low latency requires modification in PHYand MAC layer which are at RAN level Even though severalpromising solutions are proposed to date we believe that thefollowing issues at RAN level need to be investigated furtherbull For achieving low latency in 5G networks mmWave

is a promising technology which brings massive newspectrum for communications in the 3-300 GHz bandHowever mmWave is dependent on diverse aspects suchas transmitterreceiver location and environmental topol-ogy [3] [246] Moreover channel modeling with de-lay spread path loss NLOS beam forming and angu-lar spread need to be investigated in indoor and out-door environment which are still evolving [247] Ad-ditionally more in-depth knowledge of physics behindmmWave regarding aspects such as Doppler propagationatmospheric absorption reflection refraction attenuationand multi-path should be developed for utilization ofmmWave

bull In conventional packet transmission distortion and ther-mal noise induced by propagation channel get averageddue to large size of packet [58] However in case ofsmall size packet such averaging is not possible Thus

proper channel modeling followed by simulations andfield tests for small packet in diverse carrier bands needbe investigated

bull The challenge of admission control in RAN for spectraland energy efficiency with latency constraint is not wellexplored [248] CRANHRAN provides spectral and en-ergy efficiency while aspects such as caching can ensurelow latency [138] Researchers can work for performancebounds regarding this issue

bull Orthogonality and synchronization is a major drawbackof OFDM modulation for achieving low latency On theother hand orthogonality and synchronization are veryimportant for data readability Recently different non-orthogonal and asynchronous multiple access schemessuch as SCMA IDMA and GFDM FBMC and UFMChave been proposed However more effective access tech-niques and waveforms which require less coordinationensure robustness in disperse channel and provide highspectral efficiency is a potential research area [3] [249]

bull Low complexity antenna beam steering large antennaarray and efficient symbol detection such as compressedsensing is conducive for low latency communication Forthis heuristic beam forming design at BS level beamtraining protocols and weight calculation and reliableerror correction technique need to be studied [3] [250]On the other hand at the receiver level low complexitysensing techniques and receiver design should be thefocus of the future research

bull One of the major challenges is that latency critical packetsare to be multiplexed with other packets There aresolutions such as instant access for latency critical packetsceasing transmission of other packets and reservation ofresources for latency critical services [7] We believe thatthese issues are not well-explored and calls for furtherstudy

bull Even though the main objective of CRAN is to reducecosts and to enhance energy and spectral efficiency itmight be combined with heterogeneous networks termedas H-RAN However it is very challenging to design5G network with large CRANHRAN [138] In thiscase various trade-offs including energy efficiency versuslatency can be investigated

B Core Network Issues

In the core network several new entities such as SDNand NFV have been introduced for supporting large capacitymassive connectivity and low latency with seamless operationSince these entities are not part of the legacy LTE systemextensive works need to be carried out for the standardizationand development in the context of 5G Some of the issuesnecessitating research in the core network level are as followsbull The main challenge of SDNNFV-based core network

design is the management and orchestration of these het-erogeneous resources [251] Effective resource allocationand implementation of functions in this heterogeneousenvironment while also maintaining low latency is an areaof emerging research need

bull Most of the surveyed works are based on Open Flowprotocols and their extension integrating control planeand user plane without a detailed implementation planeMoreover the scalability of the user plane is not con-sidered while taking mainly control plane into considera-tion [11] Researchers have a great opportunity to exploreregarding the standardization and scalability of these corenetwork entities

bull To boost spectralenergy efficiency and reduce latencyutilizing CRAN and coordinated multi point (CoMP)mmWave is an attractive choice for frontback haulingbecause of its low implementation cost and unavailabilityof fiber everywhere [15] However research in mmWaveregarding frontback hauling is a popular area of in-vestigation Dynamic intelligent and adaptive techniquesneed to be developed with optimized utilization of theheterogeneous back hauling networks while catering lowlatency

bull Even though MEC is envisioned to reduce individualcomputation inclusion of the caching in MEC can furtherboost usersrsquo QoE The caching enabled MEC will providecontent delivery and memory support for memory hungryapplications such as VR and online gaming along withBS level caching [183] Researchers are encouraged tostudy various trade-offs such as capacity versus latencystorage versus link load and memory versus rate

C Caching Issues

Edge caching can be a critical tool for latency reductionalong spectral and energy efficiency improvement Recentlythis issue attracted huge attention from researchers in bothacademia and industry resulting in many different approachesHowever we believe there are still extensive research problemsopen and need further explorationbull Even though several exciting works have been carried out

for content placement and content delivery time furtherstudies could be done regarding the latency aspects suchas how latency is impacted by caching size and locationand wireless channel parameters [252]

bull Assuming that content delivery and content placementare the main two phases of wireless caching networkarchitecture including caching storage size placementand cooperation for caching are potential areas for furtherstudy [253] [254] Besides that the protocol designfor caching redundancy and intra cache communicationcan be investigated with latency constraint [254] In thisregard performance limits and bounds of caching can bestudied for getting insights on optimum performance

bull The BS makes a tunnel between UE and EPC for contentrequest However the contents are packeted through GTPtunnel which creates difficulty in content-aware or object-orientated caching [255] Proper protocol designs canaddress such problems

bull In low user density areas caching capacity may be insurplus for serving UE while in urban areas the situationmay be opposite Intelligent and co-operative resourceallocation and caching strategies can ensure proper hit

ratio along with low latency content delivery [254] In thisregard relatively few works are available in the literatureand can be further investigated

bull Mobility is an important issue in latency critical applica-tions such as AR The movement and trajectory boost thelocation and performance information for local cachingwhich handle the current userrsquos experience The move-ment of users among cells will incur interference and pilotcontamination along with complication in system config-urations and user-server association policies Moreoverfrequent handover will introduce latency degrading userrsquosexperience [256] Thus handover in diverse cachingscenarios with focus on low latency can be a potentialarea of research

X CONCLUSION

Along with very large capacity massive connection densityand ultra high reliability 5G networks will need to supportultra low latency The low latency will enable new servicessuch as VRAR tele-medicine and tele-surgery in somecases latency not more than 1 ms is critical To achievethis low latency drastic changes in multiple network domainsneed to be addressed In this paper an extensive survey ondifferent approaches in order to achieve low latency in 5Gnetworks is presented Different approaches are reviewed inthe domain of RAN core network and caching for achievinglow latency In the domain of RAN techniques we havestudied short framepackets new waveform designs multipleaccess techniques modulation and coding schemes controlchannel approaches symbol detection methods transmissiontechniques mmWave aggregation cloud RAN reinforcingQoS and QoE and location aware communication as differentaspects of facilitating low latency

On the other hand SDN NFV and MECfog networkarchitectures along with high speed backhaul are reviewed inthe literature for core network with vision to meet the low la-tency requirements of 5G The new core network will providediverse advantages such as distributed network functionalityindependence of software platform from hardware platformand separation of data plane from software plane whichwill all help in latency reduction In caching distributed andcentralized caching with various trade-offs cache placementand content delivery have been proposed for latency reductionin content download Following this promising results fromfield tests trials and experiments have also been presentedhere However more practical and efficient techniques in thepresence of existing solutions need to be investigated beforethe standardization of 5G In this regard we discussed theopen issues challenges and future research directions forresearchers The authors believe that this survey will serveas a valuable resource for latency reduction for the emerging5G cellular networks and beyond

ACKNOWLEDGMENT

The work was supported by the National Science Founda-tion entitled ldquoTowards Secure Networked Cyber-Physical Sys-tems A Theoretic Framework with Bounded Rationalityrdquo andldquoCyber Physical Solution for High Penetration Renewables in

Smart Gridrdquo under the grant number 1446570 and 1553494respectively

REFERENCES

[1] ldquoTactile Internetrdquo Tactile Internet ad-hoc definition group IEEEP19181

[2] S Zhang X Xu Y Wu and L Lu ldquo5G Towards energy-efficient low-latency and high-reliable communications networksrdquo in Proc IEEE IntConf on Commun Syst (ICCS) Nov 2014 pp 197ndash201

[3] M Agiwal A Roy and N Saxena ldquoNext Generation 5G WirelessNetworks A Comprehensive Surveyrdquo IEEE Commun Surv Tutorvol 18 no 3 pp 1617ndash1655 thirdquarter 2016

[4] A Gupta and R K Jha ldquoA Survey of 5G Network Architecture andEmerging Technologiesrdquo IEEE Access vol 3 pp 1206ndash1232 2015

[5] ITU-R ldquoFramework and Overall Objectives of the Future Developmentof IMT for 2020 and Beyondrdquo Feb 2015

[6] A Kumbhar F Koohifar I Guvenc and B Mueller ldquoA Surveyon Legacy and Emerging Technologies for Public Safety Commu-nicationsrdquo IEEE Commun Surv Tutor vol 19 no 1 pp 97ndash124Firstquarter 2017

[7] H Ji S Park J Yeo Y Kim J Lee and B Shim ldquoIntroductionto Ultra Reliable and Low Latency Communications in 5Grdquo arXivpreprint arXiv170405565 2017

[8] K I Pedersen F Frederiksen G Berardinelli and P E MogensenldquoThe Coverage-Latency-Capacity Dilemma for TDD Wide Area Op-eration and Related 5G Solutionsrdquo in Proc IEEE Veh Technol Conf(VTC Spring) May 2016 pp 1ndash5

[9] F L Figueiredo ldquoA Survey on Key Technology Trendsfor 5G Networksrdquo httpshttpswwwslidesharenetcpqda-survey-on-key-technology-trends-for-5g-networks Feb 2014[Online accessed 19-Nov-2017]

[10] M Mueck E C Strinati I G Kim A Clemente J B Dore A DDomenico T Kim T Choi H K Chung G Destino A ParssinenA Pouttu M Latva-aho N Chuberre M Gineste B VautherinM Monnerat V Frascolla M Fresia W Keusgen T HausteinA Korvala M Pettissalo and O Liinamaa ldquo5G CHAMPION - Rollingout 5G in 2018rdquo in Proc IEEE Global Commun Conf WorkshopsDec 2016 pp 1ndash6

[11] V G Nguyen A Brunstrom K J Grinnemo and J TaherildquoSDNNFV-based Mobile Packet Core Network Architectures A Sur-veyrdquo IEEE Commun Surv Tutor vol PP no 99 pp 1ndash1 2017

[12] T Taleb K Samdanis B Mada H Flinck S Dutta and D SabellaldquoOn Multi-Access Edge Computing A Survey of the Emerging 5GNetwork Edge Architecture Orchestrationrdquo IEEE Commun Surv Tu-tor vol PP no 99 pp 1ndash1 2017

[13] A Ioannou and S Weber ldquoA Survey of Caching Policies and Forward-ing Mechanisms in Information-Centric Networkingrdquo IEEE CommunSurv Tutor vol 18 no 4 pp 2847ndash2886 Fourthquarter 2016

[14] M Zhang H Luo and H Zhang ldquoA Survey of Caching Mechanismsin Information-Centric Networkingrdquo IEEE Commun Surv Tutorvol 17 no 3 pp 1473ndash1499 thirdquarter 2015

[15] M Jaber M A Imran R Tafazolli and A Tukmanov ldquo5G Back-haul Challenges and Emerging Research Directions A Surveyrdquo IEEEAccess vol 4 pp 1743ndash1766 2016

[16] T O Olwal K Djouani and A M Kurien ldquoA Survey of ResourceManagement Toward 5G Radio Access Networksrdquo IEEE CommunSurv Tutor vol 18 no 3 pp 1656ndash1686 thirdquarter 2016

[17] W Xia P Zhao Y Wen and H Xie ldquoA Survey on Data CenterNetworking (DCN) Infrastructure and Operationsrdquo IEEE CommunSurv Tutor vol 19 no 1 pp 640ndash656 Firstquarter 2017

[18] M F Bari R Boutaba R Esteves L Z Granville M PodlesnyM G Rabbani Q Zhang and M F Zhani ldquoData Center NetworkVirtualization A Surveyrdquo IEEE Commun Surv Tutor vol 15 no 2pp 909ndash928 Second 2013

[19] B Briscoe A Brunstrom A Petlund D Hayes D Ros I J TsangS Gjessing G Fairhurst C Griwodz and M Welzl ldquoReducingInternet Latency A Survey of Techniques and Their Meritsrdquo IEEECommun Surv Tutor vol 18 no 3 pp 2149ndash2196 thirdquarter 2016

[20] S Srivastava and S P Singh ldquoA Survey on Latency ReductionApproaches for Performance Optimization in Cloud Computingrdquo inProc Int Conf Comput Intell Commun Technol (CICT) Feb 2016pp 111ndash115

[21] S Hosseinalipour and H Dai ldquoOptions-based sequential auctions fordynamic cloud resource allocationrdquo in Proc IEEE Int Conf Commun(ICC) May 2017 pp 1ndash6

[22] D Delaney T Ward and S McLoone ldquoOn Consistency and NetworkLatency in Distributed Interactive Applications A Survey Part IrdquoPresence vol 15 no 2 pp 218ndash234 April 2006

[23] mdashmdash ldquoOn Consistency and Network Latency in Distributed InteractiveApplications A Survey Part IIrdquo Presence vol 15 no 4 pp 465ndash482Aug 2006

[24] P Schulz M Matthe H Klessig M Simsek G Fettweis J AnsariS A Ashraf B Almeroth J Voigt I Riedel et al ldquoLatency criticalIoT applications in 5G Perspective on the design of radio interfaceand network architecturerdquo IEEE Commun Mag vol 55 no 2 pp70ndash78 2017

[25] M R Palattella M Dohler A Grieco G Rizzo J Torsner T Engeland L Ladid ldquoInternet of things in the 5G era Enablers architectureand business modelsrdquo Journ on Select Areas in Commun vol 34no 3 pp 510ndash527 2016

[26] A I Sarwat A Sundararajan I Parvez M Moghaddamiand A Moghadasi Toward a Smart City of InterdependentCritical Infrastructure Networks Cham Springer InternationalPublishing 2018 pp 21ndash45 [Online] Available httpsdoiorg101007978-3-319-74412-4 3

[27] M Tavana A Rahmati and V Shah-Mansouri ldquoCongestion controlwith adaptive access class barring for LTE M2M overload using kalmanfiltersrdquo Computer Networks 2018

[28] I Parvez F Abdul H Mohammed and A I Sarwat ldquoReliabilityassessment of access point of advanced metering infrastructure basedon Bellcore standards (Telecordia)rdquo in Proc IEEE SoutheastCon April2015 pp 1ndash7

[29] A Ferdowsi and W Saad ldquoDeep learning-based dynamic watermarkingfor secure signal authentication in the internet of thingsrdquo in arXivpreprint arXiv171101306v1

[30] I Parvez F Abdul and A I Sarwat ldquoA Location Based KeyManagement System for Advanced Metering Infrastructure of SmartGridrdquo in Proc IEEE Green Techn Conf (GreenTech) Apr 2016 pp62ndash67

[31] P Schulz M Matthe H Klessig M Simsek G Fettweis J AnsariS A Ashraf B Almeroth J Voigt I Riedel A PuschmannA Mitschele-Thiel M Muller T Elste and M Windisch ldquoLatencyCritical IoT Applications in 5G Perspective on the Design of RadioInterface and Network Architecturerdquo IEEE Commun Mag vol 55no 2 pp 70ndash78 Feb 2017

[32] O N C Yilmaz Y P E Wang N A Johansson N Brahmi S AAshraf and J Sachs ldquoAnalysis of ultra-reliable and low-latency 5Gcommunication for a factory automation use caserdquo in Proc IEEE IntConf Commun Workshop (ICCW) Jun 2015 pp 1190ndash1195

[33] B Holfeld D Wieruch T Wirth L Thiele S A Ashraf J HuschkeI Aktas and J Ansari ldquoWireless communication for factory automa-tion An opportunity for LTE and 5G systemsrdquo IEEE Commun Magvol 54 no 6 pp 36ndash43 2016

[34] A F Cattoni D Chandramouli C Sartori R Stademann andP Zanier ldquoMobile Low Latency Services in 5Grdquo in Proc IEEE VehTechnol Conf (VTC Spring) May 2015 pp 1ndash6

[35] C Campolo A Molinaro G Araniti and A O Berthet ldquoBetterPlatooning Control Toward Autonomous Driving An LTE Device-to-Device Communications Strategy That Meets Ultra-low LatencyRequirementsrdquo IEEE Veh Technol Mag vol 12 no 1 pp 30ndash38March 2017

[36] R Alieiev A Kwoczek and T Hehn ldquoAutomotive requirements forfuture mobile networksrdquo in Proc IEEE MTT-S Inter Conf on Microfor Intell Mobil (ICMIM) Apr 2015 pp 1ndash4

[37] M A Lema A Laya T Mahmoodi M Cuevas J Sachs J Mark-endahl and M Dohler ldquousiness Case and Technology Analysis for5G Low Latency Applicationsrdquo IEEE Access vol 5 pp 5917ndash59352017

[38] ITU-Report ldquoThe Tactile Internetrdquo ITU-T Technology WatchReport Aug 2014 [Online] Available httpshttpwwwituintothT2301000023en

[39] M Simsek A Aijaz M Dohler J Sachs and G Fettweis ldquoThe 5G-Enabled Tactile Internet Applications requirements and architecturerdquoin Proc IEEE Wireless Commun Netw Conf (WCNC) Apr 2016 pp1ndash6

[40] N A Johansson Y P E Wang E Eriksson and M Hessler ldquoRadioaccess for ultra-reliable and low-latency 5G communicationsrdquo in ProcIEEE Intern Conf on Comm Works (ICCW) Jun 2015 pp 1184ndash1189

[41] J Choi V Va N Gonzalez-Prelcic R Daniels C R Bhatand R W Heath ldquoMillimeter-wave vehicular communication to

support massive automotive sensingrdquo Comm Mag vol 54no 12 pp 160ndash167 Dec 2016 [Online] Available httpsdoiorg101109MCOM20161600071CM

[42] M A Lema K Antonakoglou F Sardis N Sornkarn M CondoluciT Mahmoodi and M Dohler ldquo5G case study of Internet of SkillsSlicing the human sensesrdquo in Porc Europ Conf on Netw andCommun(EuCNC) Jun 2017 pp 1ndash6

[43] A Ghosh ldquo5G mmWave Revolution and New Radiordquohttps5gieeeorgimagesfilespdf5GmmWave Webinar IEEENokia 09 20 2017 finalpdf Sep 2017 [Online accessed 19-Nov-2017]

[44] ldquoCommunications Requirements Of Smart Grid Technologiesrdquo USDepartment of Energy Tech Rep Oct 2010 [Online] Availablehttpwwwsmartgridgov

[45] A T Z Kasgari W Saad and M Debbah ldquoHuman-in-the-loopwireless communications Machine learning and brain-aware resourcemanagementrdquo arXiv preprint arXiv180400209 2018

[46] I Parvez M Jamei A Sundararajan and A I Sarwat ldquoRSS basedloop-free compass routing protocol for data communication in ad-vanced metering infrastructure (AMI) of Smart Gridrdquo in Proc IEEESymp on Comp Intell App in Smart Grid (CIASG) Dec 2014 pp1ndash6

[47] I Parvez A Islam and F Kaleem ldquoA key management-based two-level encryption method for AMIrdquo in Proc IEEE PES General Meetingmdash Conference Exposition July 2014 pp 1ndash5

[48] I Parvez A I Sarwat L Wei and A Sundararajan ldquoSecuring Meter-ing Infrastructure of Smart Grid A Machine Learning and LocalizationBased Key Management Approachrdquo Energies vol 9 no 9 2016

[49] I Parvez A I Sarwat J Pinto Z Parvez and M A KhandakerldquoA gossip algorithm based clock synchronization scheme for smartgrid applicationsrdquo in Proc North Ameri Power Sympo (NAPS) Sep2017 pp 1ndash6

[50] C Xu J Cosmas Y Zhang P Lazaridis G Araniti and Z D Zaharisldquo3D MIMO radio channel modeling of a weighted linear array systemof antennas for 5G cellular systemsrdquo in Proc Int Conf TelecommunMultimedia (TEMU) July 2016 pp 1ndash6

[51] Latency Analysis in LTE network [Online]Available httpwwwtechmahindracomDocumentsWhitePaperWhitePaperLatencyAnalysispdf

[52] C A Garcia-Perez and P Merino ldquoEnabling Low Latency Services onLTE Networksrdquo in Proc IEEE Int Workshop Found Appl Self Syst(FASW) Sep 2016 pp 248ndash255

[53] K Wilner and A P Van Den Heuvel ldquoFiber-optic delay lines formicrowave signal processingrdquo Proc IEEE vol 64 no 5 pp 805ndash807 1976

[54] G Pocovi K I Pedersen B Soret M Lauridsen and P MogensenldquoOn the impact of multi-user traffic dynamics on low latency commu-nicationsrdquo in Proc IEEE Int Symp Wireless Commun Syst (ISWCS)2016 pp 204ndash208

[55] ETSI ldquoUniversal Mobile Telecommunications System (UMTS) Fea-sibility study for evolved Universal Terrestrial Radio Access (UTRA)and Universal Terrestrial Radio Access Network (UTRAN)rdquo ETSI TR125 912 V710 Tech Rep 09 2006

[56] P K Agyapong M Iwamura D Staehle W Kiess and A BenjebbourldquoDesign considerations for a 5G network architecturerdquo IEEE CommunMag vol 52 no 11 pp 65ndash75 Nov 2014

[57] G Wunder P Jung M Kasparick T Wild F Schaich Y Chen S TBrink I Gaspar N Michailow A Festag L Mendes N CassiauD Ktenas M Dryjanski S Pietrzyk B Eged P Vago and F Wied-mann ldquo5GNOW non-orthogonal asynchronous waveforms for futuremobile applicationsrdquo IEEE Commun Mag vol 52 no 2 pp 97ndash105Feb 2014

[58] G Durisi T Koch and P Popovski ldquoToward Massive Ultrareliableand Low-Latency Wireless Communication With Short Packetsrdquo ProcIEEE vol 104 no 9 pp 1711ndash1726 Sep 2016

[59] C-P Li J Jiang W Chen T Ji and J Smee ldquo5G ultra-reliable andlow-latency systems designrdquo in Proc European Conf Netw Commun(EuCNC) June 2017 pp 1ndash5

[60] P Guan X Zhang G Ren T Tian A Benjebbour Y Saito andY Kishiyama ldquoUltra-Low Latency for 5G - A Lab Trialrdquo CoRR volabs161004362 2016 [Online] Available httparxivorgabs161004362

[61] E Lahetkangas K Pajukoski J Vihriala G Berardinelli M Laurid-sen E Tiirola and P Mogensen ldquoAchieving low latency and energyconsumption by 5G TDD mode optimizationrdquo in Proc IEEE Int ConfCommun Workshop (ICCW) Jun 2014 pp 1ndash6

[62] K I Pedersen G Berardinelli F Frederiksen P Mogensen andA Szufarska ldquoA flexible 5G frame structure design for frequency-division duplex casesrdquo IEEE Commun Mag vol 54 no 3 pp 53ndash59Mar 2016

[63] T Wirth M Mehlhose J Pilz B Holfeld and D Wieruch ldquo5Gnew radio and ultra low latency applications A PHY implementationperspectiverdquo in Proc Asilomar Conf Signals Syst and Comput Nov2016 pp 1409ndash1413

[64] E Lahetkangas K Pajukoski E Tiirola G Berardinelli I Harjulaand J Vihriala ldquoOn the TDD subframe structure for beyond 4G radioaccess networkrdquo in 2013 Future Network Mobile Summit Jul 2013pp 1ndash10

[65] J Vihriala A A Zaidi V Venkatasubramanian N He E TiirolaJ Medbo E Lahetkangas K Werner K Pajukoski A Cedergren andR Baldemair ldquoNumerology and frame structure for 5G radio accessrdquoin Proc IEEE Symp Pers Indoor Mobile Radio Commun (PIMRC)Sep 2016 pp 1ndash5

[66] R Abreu P Mogensen and K I Pedersen ldquoPre-Scheduled Resourcesfor Retransmissions in Ultra-Reliable and Low Latency Communica-tionsrdquo in Proc IEEE Wireless Commun and Netw Conf (WCNC)Mar 2017 pp 1ndash5

[67] B Soret P Mogensen K I Pedersen and M C Aguayo-TorresldquoFundamental tradeoffs among reliability latency and throughput incellular networksrdquo in Proc IEEE Globecom Workshop (GC Wkshps)2014 pp 1391ndash1396

[68] F Schaich T Wild and Y Chen ldquoWaveform Contenders for 5G -Suitability for Short Packet and Low Latency Transmissionsrdquo in ProcVeh Technol Conf (VTC Spring) May 2014 pp 1ndash5

[69] M Kasparick G Wunder Y Chen F Schaich and T Wild ldquo5GWaveform Candidate Selection D31rdquo in 5Gnow Nov 2013 [Online]

[70] B Farhang-Boroujeny ldquoOFDM Versus Filter Bank Multicarrierrdquo IEEESignal Process Mag vol 28 no 3 pp 92ndash112 May 2011

[71] M Sybis K Wesolowski K Jayasinghe V Venkatasubramanianand V Vukadinovic ldquoChannel Coding for Ultra-Reliable Low-LatencyCommunication in 5G Systemsrdquo in Proc IEEE Veh Technol Conf(VTC Fall) Sep 2016 pp 1ndash5

[72] B Zhang H Shen B Yin L Lu D Chen T Wang L Gu X WangX Hou H Jiang A Benjebbour and Y Kishiyama ldquoA 5G Trial ofPolar Coderdquo in Proc IEEE Globecom Workshops (GC Wkshps) Dec2016 pp 1ndash6

[73] D Liu C Zuo and Z Wu ldquoBenefit and cost of cross sliding windowscheduling for low latency 5G Turbo decodingrdquo in Proc IEEECICInt Conf Commun China (ICCC) Nov 2015 pp 1ndash4

[74] I-G Jang and G-D Jo ldquoStudy on the latency efficient IFFT designmethod for low latency communication systemsrdquo in Proc Int SympIntell Signal Process Commun Syst (ISPACS) Oct 2016 pp 1ndash4

[75] Y Zhou H Luo R Li and J Wang ldquoA dynamic states reductionmessage passing algorithm for sparse code multiple accessrdquo in ProcIEEE Wireless Telecommun Symp (WTS) Apr 2016 pp 1ndash5

[76] K Ganesan T Soni S Nunna and A R Ali ldquoPoster A TDMapproach for latency reduction of ultra-reliable low-latency data in 5Grdquoin Proc IEEE Veh Netw Conf (VNC) Dec 2016 pp 1ndash2

[77] D A Jaoude and M Farhood ldquoBalanced truncation of linear systemsinterconnected over arbitrary graphs with communication latencyrdquo inProc IEEE Conf Decision and Control (CDC) Dec 2015 pp 5346ndash5351

[78] J Ostman G Durisi E G Strom J Li H Sahlin and G Liva ldquoLow-latency Ultra Reliable 5G Communications Finite-Blocklength Boundsand Coding Schemesrdquo arXiv preprint arXiv161108714 2016

[79] T Taheri R Nilsson and J van de Beek ldquoAsymmetric Transmit-Windowing for Low-Latency and Robust OFDMrdquo in Proc IEEEGlobecom Workshop (GC Wkshps) Dec 2016 pp 1ndash6

[80] C She C Yang and T Q S Quek ldquoCross-layer Transmission Designfor Tactile Internetrdquo CoRR vol abs161002800 2016 [Online]Available httparxivorgabs161002800

[81] H Hirai T Tojo and M M N Takaya ldquoLow Latency packet transportmethods for remote-controlled devices in multi-RAT environmentsrdquoin Proc IEEE Int Symp on Local and Metropolitan Area Netw(LANMAN) Jun 2016 pp 1ndash2

[82] N A Johansson Y P E Wang E Eriksson and M Hessler ldquoRadioaccess for ultra-reliable and low-latency 5G communicationsrdquo in ProcIEEE Int Conf Commun Workshop (ICCW) Jun 2015 pp 1184ndash1189

[83] H Ji S Park and B Shim ldquoSparse coding for reliable control channeltransmissionrdquo to appear on IEEE Commun Mag 2017

[84] S A Ashraf F Lindqvist R Baldemair and B Lindoff ldquoCon-trol Channel Design Trade-Offs for Ultra-Reliable and Low-LatencyCommunication Systemrdquo in Proc IEEE Globecom Workshops (GCWkshps) Dec 2015 pp 1ndash6

[85] S Xia X Han X Yan Z Zuo and F Bi ldquoUplink control channeldesign for 5G ultra-low latency communicationrdquo in Proc IEEE PersonIndoor Mobile Radio Commun (PIMRC) Sep 2016 pp 1ndash6

[86] R Jin X Zhong and S Zhou ldquoThe Access Procedure Design for LowLatency in 5G Cellular Networkrdquo in Proc IEEE Globecom Workshop(GC Wkshps) Dec 2016 pp 1ndash6

[87] T Ohseki and Y Suegara ldquoFast outer-loop link adaptation scheme re-alizing low-latency transmission in LTE-Advanced and future wirelessnetworksrdquo in Proc IEEE Radio and Wireless Symp (RWS) Jan 2016pp 1ndash3

[88] H Tang W Zhang W Hardjawana and B Vucetic ldquoImprovinglatency and reliability in 5G Internet-of-Things networksrdquo in ProcIEEE Int Conf Smart Grid Commun (SmartGridComm) Nov 2016pp 509ndash513

[89] B Lee S Park D J Love H Ji and B Shim ldquoPacket Structure andReceiver Design for Low-Latency Communications with Ultra-SmallPacketsrdquo in Proc IEEE Global Commun Conf (GLOBECOM) Dec2016 pp 1ndash6

[90] M Matthe L L Mendes N Michailow D Zhang and G FettweisldquoWidely Linear Estimation for Space-Time-Coded GFDM in Low-Latency Applicationsrdquo IEEE Trans Commun vol 63 no 11 pp4501ndash4509 Nov 2015

[91] L D Gregorio ldquoA Note on Compressed Sensing for Low Latencyrdquoin Proc Mobile SystTechnol Workshop (MST) May 2015 pp 8ndash11

[92] J W Choi B Shim Y Ding B Rao and D I Kim ldquoCompressedSensing for Wireless Communications Useful Tips and Tricksrdquo IEEECommun Surveys Tuts vol PP no 99 pp 1ndash1 2017

[93] C Bockelmann N Pratas H Nikopour K Au T Svensson C Ste-fanovic P Popovski and A Dekorsy ldquoMassive machine-type commu-nications in 5G physical and MAC-layer solutionsrdquo IEEE CommunMag vol 54 no 9 pp 59ndash65 Sep 2016

[94] L Lu Y Chen W Guo H Yang Y Wu and S Xing ldquoPrototype for5G new air interface technology SCMA and performance evaluationrdquoChina Commun vol 12 no Supplement pp 38ndash48 Dec 2015

[95] S Dutta M Mezzavilla R Ford M Zhang S Rangan and M ZorzildquoFrame Structure Design and Analysis for Millimeter Wave CellularSystemsrdquo CoRR vol abs151205691 2015 [Online] Availablehttparxivorgabs151205691

[96] R Ford M Zhang M Mezzavilla S Dutta S Rangan and M ZorzildquoAchieving Ultra-Low Latency in 5G Millimeter Wave CellularNetworksrdquo CoRR vol abs160206925 2016 [Online] Availablehttparxivorgabs160206925

[97] T Levanen J Pirskanen and M Valkama ldquoRadio interface design forultra-low latency millimeter-wave communications in 5G Erardquo in ProcIEEE Globecom Workshop (GC Wkshps) Dec 2014 pp 1420ndash1426

[98] S Yoshioka Y Inoue S Suyama Y Kishiyama Y Okumura J Ke-pler and M Cudak ldquoField experimental evaluation of beamtrackingand latency performance for 5G mmWave radio access in outdoor mo-bile environmentrdquo in Proc IEEE Pers Indoor Mobile Radio Commun(PIMRC) Sep 2016 pp 1ndash6

[99] T Levanen J Pirskanen T Koskela J Talvitie and M ValkamaldquoLow latency radio interface for 5G flexible TDD local area commu-nicationsrdquo in Proc IEEE Int Conf Commun Workshop (ICCW) Jun2014 pp 7ndash13

[100] T A Levanen J Pirskanen T Koskela J Talvitie and M ValkamaldquoRadio Interface Evolution Towards 5G and Enhanced Local AreaCommunicationsrdquo IEEE Access vol 2 pp 1005ndash1029 2014

[101] R D Taranto S Muppirisetty R Raulefs D Slock T Svensson andH Wymeersch ldquoLocation-Aware Communications for 5G NetworksHow location information can improve scalability latency and robust-ness of 5Grdquo IEEE Signal Process Mag vol 31 no 6 pp 102ndash112Nov 2014

[102] O Bazan and M Jaseemuddin ldquoA Conflict Analysis Framework forQoS-Aware Routing in Contention-Based Wireless Mesh Networkswith Beamforming Antennasrdquo IEEE Trans Wireless Commun vol 10no 10 pp 3267ndash3277 Oct 2011

[103] Huawei ldquo5G A technology visionrdquo White paper 2013[104] D Wu J Wang Y Cai and M Guizani ldquoMillimeter-wave multimedia

communications challenges methodology and applicationsrdquo IEEECommun Mag vol 53 no 1 pp 232ndash238 Jan 2015

[105] A Kamel A Al-Fuqaha and M Guizani ldquoExploiting Client-SideCollected Measurements to Perform QoS Assessment of IaaSrdquo IEEETrans Mobile Comput vol 14 no 9 pp 1876ndash1887 Sep 2015

[107] X Zhang W Cheng and H Zhang ldquoHeterogeneous statistical QoSprovisioning over 5G mobile wireless networksrdquo IEEE Networkvol 28 no 6 pp 46ndash53 Nov 2014

[108] W Wang X Wang and A A Nilsson ldquoEnergy-efficient bandwidthallocation in wireless networks algorithms analysis and simulationsrdquoIEEE Trans Wireless Commun vol 5 no 5 pp 1103ndash1114 May2006

[109] A Balachandran V Sekar A Akella S Seshan I Stoicaand H Zhang ldquoDeveloping a Predictive Model of Quality ofExperience for Internet Videordquo SIGCOMM Comput Commun Revvol 43 no 4 pp 339ndash350 Aug 2013 [Online] Availablehttpdoiacmorg10114525341692486025

[110] A Roy P De and N Saxena ldquoLocation-based social video sharingover next generation cellular networksrdquo IEEE Commun Mag vol 53no 10 pp 136ndash143 Oct 2015

[111] A Balachandran V Sekar A Akella S Seshan I Stoicaand H Zhang ldquoA Quest for an Internet Video Quality-of-experience Metricrdquo in Proc ACM Workshop Hot Topics inNetw ser HotNets-XI 2012 pp 97ndash102 [Online] Availablehttpdoiacmorg10114523902312390248

[112] C N Mao M H Huang S Padhy S T Wang W C Chung Y CChung and C H Hsu ldquoMinimizing Latency of Real-Time ContainerCloud for Software Radio Access Networksrdquo in Proc IEEE Int ConfCloud Comput Technol and Sci (CloudCom) Nov 2015 pp 611ndash616

[113] G Mountaser M Lema Rosas T Mahmoodi and M Dohler On thefeasibility of MAC and PHY split in Cloud RAN IEEE 5 2017

[114] Y Gao W Hu K Ha B Amos P Pillai and M SatyanarayananldquoAre cloudlets necessaryrdquo 2015

[115] G Zhang T Q S Quek A Huang M Kountouris and H ShanldquoBackhaul-aware base station association in two-tier heterogeneouscellular networksrdquo in Proc IEEE Inter Works on Sig Proces Advin Wire Commun (SPAWC) June 2015 pp 390ndash394

[116] X Wang C Cavdar L Wang M Tornatore Y Zhao H S ChungH H Lee S Park and B Mukherjee ldquoJoint Allocation of Radio andOptical Resources in Virtualized Cloud RAN with CoMPrdquo in ProcIEEE Global Commun Conf (GLOBECOM) Dec 2016 pp 1ndash6

[117] P Niroopan and Y-h Chung ldquoA user-spread interleave divisionmultiple access systemrdquo Int J Adv Resear Comput and CommunEng pp 837ndash841 2012

[118] J C Fricke H Schoeneich and P A Hoeher ldquoAn interleave-divisionmultiple access based system proposal for the 4G uplinkrdquo Proc ISTMobil Wire Commun Summit 2005

[119] H Nikopour and H Baligh ldquoSparse code multiple accessrdquo in ProcIEEE Pers Indoor Mobile Radio Commun (PIMRC) Sep 2013 pp332ndash336

[120] N Michailow I Gaspar S Krone M Lentmaier and G FettweisldquoGeneralized frequency division multiplexing Analysis of an alterna-tive multi-carrier technique for next generation cellular systemsrdquo inProc Int Symp Wireless Commun Syst (ISWCS) Aug 2012 pp171ndash175

[121] A Sahin I Guvenc and H Arslan ldquoA Survey on Multicarrier Com-munications Prototype Filters Lattice Structures and ImplementationAspectsrdquo IEEE Commun Surv Tutor vol 16 no 3 pp 1312ndash1338Third 2014

[122] F Schaich and T Wild ldquoWaveform contenders for 5G OFDM vsFBMC vs UFMCrdquo in Proc Int Symp on Commun Control andSignal Process (ISCCSP) May 2014 pp 457ndash460

[123] S Dutta M Mezzavilla R Ford M Zhang S Rangan and M ZorzildquoMAC layer frame design for millimeter wave cellular systemrdquo in EuroConf Netw Commun (EuCNC) Jun 2016 pp 117ndash121

[124] S Goyal P Liu S S Panwar R A Difazio R Yang and E BalaldquoFull duplex cellular systems will doubling interference prevent dou-bling capacityrdquo IEEE Commun Mag vol 53 no 5 pp 121ndash127May 2015

[125] G Zheng ldquoJoint Beamforming Optimization and Power Control forFull-Duplex MIMO Two-Way Relay Channelrdquo IEEE Trans SignalProcess vol 63 no 3 pp 555ndash566 Feb 2015

[126] E Ahmed A M Eltawil and A Sabharwal ldquoRate Gain Region andDesign Tradeoffs for Full-Duplex Wireless Communicationsrdquo IEEETrans Wireless Commun vol 12 no 7 pp 3556ndash3565 Jul 2013

[127] A Osseiran F Boccardi V Braun K Kusume P Marsch M Mater-nia O Queseth M Schellmann H Schotten H Taoka H TullbergM A Uusitalo B Timus and M Fallgren ldquoScenarios for 5G mobileand wireless communications the vision of the METIS projectrdquo IEEECommun Mag vol 52 no 5 pp 26ndash35 May 2014

[128] A Rahmati A Sadeghi and V Shah-Mansouri ldquoPrice-based resourceallocation for full duplex self-backhauled small cell networksrdquo in ProcIEEE Int Conf Commun (ICC) 2015 pp 5709ndash5714

[129] A Rahmati V Shah-Mansouri and M Safari ldquoPrice-based resourceallocation for self-backhauled small cell networksrdquo Computer Commu-nications vol 97 pp 72ndash80 2017

[130] E Shin and G Jo ldquoUplink frame structure of short TTI systemrdquo inProc Int Conf Advanced Commun Technol (ICACT) Feb 2017 pp827ndash830

[131] Y Hwang and J Shin ldquoSlot based radio resource management forlow latency in LTE-Advanced systemrdquo in Proc Int Conf AdvancedCommun Technol (ICACT) Feb 2017 pp 244ndash246

[132] R Shreevastav and R S Carbajo ldquoDynamic RLC mode based uponlink adaptation to reduce latency and improve throughput in cellularnetworksrdquo in Proc IEEE Annual Ubiquitous Comput ElectronicsMobile Commun Conf (UEMCON) Oct 2016 pp 1ndash6

[133] H Li M Dong and K Ota ldquoControl Plane Optimization in Software-Defined Vehicular Ad Hoc Networksrdquo Proc IEEE Trans Veh Technolvol 65 no 10 pp 7895ndash7904 Oct 2016

[134] T D Assefa R Hoque E Tragos and X Dimitropoulos ldquoSDN-basedlocal mobility management with X2-interface in femtocell networksrdquoin Proc IEEE Int Workshops on Comput Aided Model and Design ofCommun Links and Netw (CAMAD) June 2017 pp 1ndash6

[135] L C Gimenez P H Michaelsen K I Pedersen T E Kolding andH C Nguyen ldquoTowards Zero Data Interruption Time with EnhancedSynchronous Handoverrdquo in Proc IEEE Veh Technol Conf June 2017pp 1ndash6

[136] J Li and J Chen ldquoPassive optical network based mobile backhaulenabling ultra-low latency for communications among base stationsrdquoIEEEOSA J Opt Commun and Netw vol 9 no 10 pp 855ndash863Oct 2017

[137] M Agiwal A Roy and N Saxena ldquoNext Generation 5G WirelessNetworks A Comprehensive Surveyrdquo IEEE Commun Surveys Tutsvol 18 no 3 pp 1617ndash1655 thirdquarter 2016

[138] Z Guizani and N Hamdi ldquoCRAN H-CRAN and F-RAN for5G systems Key capabilities and recent advancesrdquo Int J NetwManagement vol 27 no 5 pp e1973ndashna 2017 e1973 nem1973[Online] Available httpdxdoiorg101002nem1973

[139] X Hong J Wang C-X Wang and J Shi ldquoCognitive radio in 5Ga perspective on energy-spectral efficiency trade-offrdquo IEEE Communi-cations Magazine vol 52 no 7 pp 46ndash53 2014

[140] G Ding J Wang Q Wu Y-D Yao R Li H Zhang and Y Zou ldquoOnthe limits of predictability in real-world radio spectrum state dynamicsFrom entropy theory to 5G spectrum sharingrdquo IEEE Commun Magvol 53 no 7 pp 178ndash183 2015

[141] L Zhang M Xiao G Wu M Alam Y C Liang and S Li ldquoA Surveyof Advanced Techniques for Spectrum Sharing in 5G Networksrdquo IEEEWireless Commun vol 24 no 5 pp 44ndash51 Oct 2017

[142] M G S Sriyananda I Parvez I Guvene M Bennis and A I SarwatldquoMulti-armed bandit for LTE-U and WiFi coexistence in unlicensedbandsrdquo in Porc IEEE Wire Commun and Netw Conf April 2016pp 1ndash6

[143] I Parvez M Sriyananda I Guvenc M Bennis and A SarwatldquoCBRS spectrum sharing between lte-u and wifi A multiarmed banditapproachrdquo Mobile Information Systems vol 2016 2016

[144] I Parvez T Khan A I Sarwat and Z Parvez ldquoLAA LTEand WiFi based Smart Grid Metering Infrastructure in 35 GHzBandrdquo CoRR vol abs171105219 2017 [Online] Availablehttparxivorgabs171105219

[145] Y Mekonnen M Haque I Parvez A H Moghadasi and A ISarwat ldquoLTE and Wi-Fi Coexistence in Unlicensed Spectrum withApplication to Smart Grid A Reviewrdquo CoRR vol abs1708090052017 [Online] Available httparxivorgabs170809005

[146] G Ding J Wang Q Wu Y D Yao F Song and T A TsiftsisldquoCellular-base-station-assisted device-to-device communications in tvwhite spacerdquo IEEE Journal on Selected Areas in Commun vol 34no 1 pp 107ndash121 Jan 2016

[147] K Lin W Wang X Wang W Ji and J Wan ldquoQoE-driven spectrumassignment for 5G wireless networks using SDRrdquo IEEE WirelessCommun vol 22 no 6 pp 48ndash55 December 2015

[148] Y Li T Jiang M Sheng and Y Zhu ldquoQoS-Aware Admission Controland Resource Allocation in Underlay Device-to-Device Spectrum-Sharing Networksrdquo IEEE J Sel Areas Commun vol 34 no 11 pp2874ndash2886 Nov 2016

[149] N Zhang S Zhang J Zheng X Fang J W Mark and X Shen ldquoQoEDriven Decentralized Spectrum Sharing in 5G Networks PotentialGame Approachrdquo IEEE Trans Veh Technol vol 66 no 9 pp 7797ndash7808 Sep 2017

[150] Z Zhang W Zhang S Zeadally Y Wang and Y Liu ldquoCognitiveradio spectrum sensing framework based on multi-agent arc hitecturefor 5G networksrdquo IEEE Wireless Commun vol 22 no 6 pp 34ndash39Dec 2015

[151] SDN architecture overview [Online] Availablehttpswwwopennetworkingorgimagesstoriesdownloadssdn-resourcestechnical-reportsSDN-architecture-overview-10pdf[Onlineaccessed5-May-2017]

[152] E N ISG ldquoNetwork Functions Virtualization white paperrdquo in httpsportaletsiorgnfvnfv white paperpdf 2012 [Online accessed 6-July-2016] Nov 2016

[153] J Page and J M Dricot ldquoSoftware-defined networking for low-latency5G core networkrdquo in Proc Int Conf Military Commun and InformSyst (ICMCIS) May 2016 pp 1ndash7

[154] R Trivisonno R Guerzoni I Vaishnavi and D Soldani ldquoTowardszero latency Software Defined 5G Networksrdquo in Proc IEEE ConfCommun Workshop (ICCW) Jun 2015 pp 2566ndash2571

[155] D Szabo A Gulyas F H P Fitzek and D E Lucani ldquoTowardsthe Tactile Internet Decreasing Communication Latency with NetworkCoding and Software Defined Networkingrdquo in Proc Euro Conf May2015 pp 1ndash6

[156] J Heinonen T Partti M Kallio K Lappalainen H Flinck andJ Hillo ldquoDynamic tunnel switching for SDN-based cellular corenetworksrdquo in Proc ACM 4th workshop All things cellular operationsapplications and challenges 2014 pp 27ndash32

[157] J Zhang W Xie and F Yang ldquoAn architecture for 5G mobile networkbased on SDN and NFVrdquo in Proc Int Conf Wireless Mobile andMulti-Media (ICWMMN) Nov 2015 pp 87ndash92

[158] M Y Arslan K Sundaresan and S Rangarajan ldquoSoftware-definednetworking in cellular radio access networks potential and challengesrdquoIEEE Commun Mag vol 53 no 1 pp 150ndash156 January 2015

[159] X Costa-Perez A Garcia-Saavedra X Li T Deiss A de la OlivaA Di Giglio P Iovanna and A Moored ldquo5G-crosshaul An SDNNFVintegrated fronthaulbackhaul transport network architecturerdquo IEEEWireless Commun vol 24 no 1 pp 38ndash45 2017

[160] G Wang G Feng S Qin and R Wen ldquoEfficient traffic engineeringfor 5G core and backhaul networksrdquo J Commun and Netw vol 19no 1 pp 80ndash92 2017

[161] R Ford A Sridharan R Margolies R Jana and S Rangan ldquoPro-visioning low latency resilient mobile edge clouds for 5Grdquo arXivpreprint arXiv170310915 2017

[162] C C Marquezan Z Despotovic R Khalili D Perez-Caparros andA Hecker ldquoUnderstanding processing latency of SDN-based mobilitymanagement in mobile core networksrdquo in Proc IEEE Int PersonalIndoor and Mobile Radio Commun (PIMRC) 2016 pp 1ndash7

[163] M Moradi W Wu L E Li and Z M Mao ldquoSoftMoW recursiveand reconfigurable cellular WAN architecturerdquo in Proc 10th ACM IntConf emerg Netw Experiments and Technol ACM 2014 pp 377ndash390

[164] M Moradi L E Li and Z M Mao ldquoSoftMoW A dynamic andscalable software defined architecture for cellular WANsrdquo in Proc 3rdworkshop on Hot topics in software defined networking ACM 2014pp 201ndash202

[165] A S Thyagaturu Y Dashti and M Reisslein ldquoSDN-based smart gate-ways (Sm-GWs) for multi-operator small cell network managementrdquoIEEE Trans on Netw and Serv Manag vol 13 no 4 pp 740ndash7532016

[166] A Jain N Sadagopan S K Lohani and M Vutukuru ldquoA comparisonof SDN and NFV for re-designing the LTE Packet Corerdquo in Proc IEEEconf NetwFunction Virtualization and Software Defined Networks(NFV-SDN) 2016 pp 74ndash80

[167] C Liang F R Yu and X Zhang ldquoInformation-centric network func-tion virtualization over 5G mobile wireless networksrdquo IEEE Networkvol 29 no 3 pp 68ndash74 May 2015

[168] E Cau M Corici P Bellavista L Foschini G Carella A Edmondsand T M Bohnert ldquoEfficient exploitation of mobile edge computingfor virtualized 5G in EPC architecturesrdquo in Proc IEEE Mobile CloudComputing Services and Eng (MobileCloud) Mar 2016 pp 100ndash109

[169] B Martini F Paganelli P Cappanera S Turchi and P CastoldildquoLatency-aware composition of Virtual Functions in 5Grdquo in Proc IEEEConf Netw Softw (NetSoft) Apr 2015 pp 1ndash6

[170] J Heinonen P Korja T Partti H Flinck and P Poyhonen ldquoMobilitymanagement enhancements for 5G low latency servicesrdquo in Proc IEEEInt Conf on Commun Workshop (ICCW) May 2016 pp 68ndash73

[171] T Taleb ldquoToward carrier cloud Potential challenges and solutionsrdquoIEEE Wireless Commun vol 21 no 3 pp 80ndash91 2014

[172] H Hawilo A Shami M Mirahmadi and R Asal ldquoNFV state of theart challenges and implementation in next generation mobile networks(vEPC)rdquo IEEE Network vol 28 no 6 pp 18ndash26 Nov 2014

[173] H Celebi N Maxemchuk Y Li and I Guvenc ldquoEnergy reductionin small cell networks by a random onoff strategyrdquo in Proc IEEEGlobal Commun Conf (GLOBECOM) Workshop 2013 pp 176ndash181

[174] H Celebi and I Guvenc ldquoLoad Analysis and Sleep Mode Optimizationfor Energy-Efficient 5G Small Cell Networksrdquo in Proc IEEE Int ConfCommun (ICC) 2017

[175] L Pei J Huilin G Shen D Fei and P Zhiwen ldquoImpact of bs sleepingand user association scheme on delay in ultra-dense networksrdquo in 20168th Int Conf Wireless Commun Signal Process (WCSP) Oct 2016pp 1ndash6

[176] Z Alharbi A S Thyagaturu M Reisslein H ElBakoury andR Zheng ldquoPerformance Comparison of R-PHY and R-MACPHYModular Cable Access Network Architecturesrdquo IEEE Trans on Broadvol PP no 99 pp 1ndash18 2017

[177] S S Rakib ldquoVirtual converged cable access platforms for HFC cablenetworksrdquo Jan 20 2015 uS Patent 8938769

[178] A S Thyagaturu Z Alharbi and M Reisslein ldquoR-FFT FunctionSplit at IFFTFFT in Unified LTE CRAN and Cable AccessNetworkrdquo CoRR vol abs170808902 2017 [Online] Availablehttparxivorgabs170808902

[179] A S Thyagaturu A Mercian M P McGarry M Reisslein andW Kellerer ldquoSoftware Defined Optical Networks (SDONs) A Com-prehensive Surveyrdquo IEEE Commun Surv Tutor vol 18 no 4 pp2738ndash2786 Fourthquarter 2016

[180] P K Agyapong M Iwamura D Staehle W Kiess and A BenjebbourldquoDesign considerations for a 5G network architecturerdquo IEEE CommunMag vol 52 no 11 pp 65ndash75 2014

[181] A Brunstrom K-J Grinnemo J Taheri et al ldquoSDNNFV-basedMobile Packet Core Network Architectures A Surveyrdquo IEEE CommunSurveys Tuts 2017

[182] A T Z Kasgari and W Saad ldquoStochastic optimization and controlframework for 5G network slicing with effective isolationrdquo arXivpreprint arXiv180110282 2018

[183] S Wang X Zhang Y Zhang L Wang J Yang and W WangldquoA Survey on Mobile Edge Networks Convergence of ComputingCaching and Communicationsrdquo IEEE Access vol 5 pp 6757ndash67792017

[184] T Taleb M Corici C Parada A Jamakovic S Ruffino G Karagian-nis and T Magedanz ldquoEASE EPC as a service to ease mobile corenetwork deployment over cloudrdquo IEEE Network vol 29 no 2 pp78ndash88 2015

[185] A Imran A Zoha and A Abu-Dayya ldquoChallenges in 5G how toempower SON with big data for enabling 5Grdquo IEEE Network vol 28no 6 pp 27ndash33 2014

[186] D Schulz C Alexakis J Hilt M Schlosser K Habel V Jungnickeland R Freund ldquoLow Latency Mobile Backhauling using OpticalWireless Linksrdquo in Proc Symp Broadband Coverage Apr 2015 pp1ndash3

[187] J Chen and J Li ldquoEfficient mobile backhaul architecture offering ultra-short latency for handoversrdquo in Proc Int Conf Transparent Opt Netw(ICTON) Jul 2016 pp 1ndash1

[188] S Gonzalez A Oliva X Costa-Perez A Di Giglio F CavaliereT Deiszlig X Li and A Mourad ldquo5G-Crosshaul An DNNFV controland data plane architecture for the 5G integrated FronthaulBackhaulrdquoTrans Emerg Telecommun Technol vol 27 no 9 pp 1196ndash12052016

[189] M Fiorani B Skubic J Martensson L Valcarenghi P CastoldiL Wosinska and P Monti ldquoOn the design of 5G transport networksrdquoPhotonic netw commun vol 30 no 3 pp 403ndash415 2015

[190] J Zhang X Zhang and W Wang ldquoCache-Enabled Software DefinedHeterogeneous Networks for Green and Flexible 5G Networksrdquo IEEEAccess vol 4 pp 3591ndash3604 2016

[191] R J Weiler M Peter W Keusgen E Calvanese-StrinatiA De Domenico I Filippini A Capone I Siaud A-M Ulmer-MollA Maltsev et al ldquoEnabling 5G backhaul and access with millimeter-wavesrdquo in Proc IEEE Euro Conf Netw Commun (EuCNC) 2014pp 1ndash5

[192] Z Gao L Dai D Mi Z Wang M A Imran and M Z ShakirldquoMmWave massive-MIMO-based wireless backhaul for the 5G ultra-dense networkrdquo IEEE Wireless Commun vol 22 no 5 pp 13ndash212015

[193] R Taori and A Sridharan ldquoPoint-to-multipoint in-band mmwavebackhaul for 5G networksrdquo IEEE Commun Mag vol 53 no 1 pp195ndash201 2015

[194] T Levanen J Pirskanen and M Valkama ldquoRadio interface design forultra-low latency millimeter-wave communications in 5G erardquo in ProcIEEE Globecom Workshop (GC Wkshps) Dec 2014 pp 1420ndash1426

[195] J S Vardakas I T Monroy L Wosinska G Agapiou R BrenotN Pleros and C Verikoukis ldquoTowards high capacity and low latencybackhauling in 5G The 5G STEP-FWD visionrdquo in Proc 19th IntConf Transparent Opt Netw (ICTON) July 2017 pp 1ndash4

[196] K Shanmugam N Golrezaei A G Dimakis A F Molisch andG Caire ldquoFemtoCaching Wireless Content Delivery Through Dis-tributed Caching Helpersrdquo IEEE Trans Inf Theory vol 59 no 12pp 8402ndash8413 Dec 2013

[197] S Mosleh L Liu H Hou and Y Yi ldquoCoordinated Data AssignmentA Novel Scheme for Big Data over Cached Cloud-RANrdquo in ProcIEEE Global Commun Conf (GLOBECOM) Dec 2016 pp 1ndash6

[198] X Peng J C Shen J Zhang and K B Letaief ldquoBackhaul-AwareCaching Placement for Wireless Networksrdquo in Proc IEEE GlobalCommun Conf (GLOBECOM) Dec 2015 pp 1ndash6

[199] J Liu B Bai J Zhang and K B Letaief ldquoContent caching at thewireless network edge A distributed algorithm via belief propagationrdquoin Proc IEEE Int Conf Commun (ICC) May 2016 pp 1ndash6

[200] M A Maddah-Ali and U Niesen ldquoDecentralized Coded Caching At-tains Order-Optimal Memory-Rate Tradeoffrdquo IEEEACM Trans Netwvol 23 no 4 pp 1029ndash1040 Aug 2015

[201] H Hsu and K C Chen ldquoA Resource Allocation Perspective onCaching to Achieve Low Latencyrdquo IEEE Commun Lett vol 20 no 1pp 145ndash148 Jan 2016

[202] G Carofiglio L Mekinda and L Muscariello ldquoFOCAL Forwardingand Caching with Latency Awareness in Information-Centric Network-ingrdquo in Proc IEEE Global Workshop (GC Wkshps) Dec 2015 pp1ndash7

[203] Y Sun Z Chen and H Liu ldquoDelay Analysis and Optimizationin Cache-Enabled Multi-Cell Cooperative Networksrdquo in Proc IEEEGlobal Commun Conf (GLOBECOM) Dec 2016 pp 1ndash7

[204] X Zhang and Q Zhu ldquoP2P Caching Schemes for Jointly MinimizingMemory Cost and Transmission Delay over Information-Centric Net-worksrdquo in Proc IEEE Global Commun Conf (GLOBECOM) Dec2016 pp 1ndash6

[205] A Sengupta R Tandon and O Simeone ldquoCache aided wirelessnetworks Tradeoffs between storage and latencyrdquo in Proc Annu ConfInf Sci and Syst (CISS) Mar 2016 pp 320ndash325

[206] R Tandon and O Simeone ldquoCloud-aided wireless networks withedge caching Fundamental latency trade-offs in fog Radio AccessNetworksrdquo in Proc IEEE Int Symp Inf Theory (ISIT) Jul 2016pp 2029ndash2033

[207] F Xu K Liu and M Tao ldquoCooperative TxRx caching in interferencechannels A storage-latency tradeoff studyrdquo in Proc IEEE Inte Sympon Info Theo (ISIT) July 2016 pp 2034ndash2038

[208] N Naderializadeh M A Maddah-Ali and A S Avestimehr ldquoFunda-mental Limits of Cache-Aided Interference Managementrdquo IEEE TransInf Theory vol 63 no 5 pp 3092ndash3107 May 2017

[209] Y Cao F Xu K Liu and M Tao ldquoA Storage-Latency Tradeoff Studyfor Cache-Aided MIMO Interference Networksrdquo in Proc IEEE GlobalCommun Conf (GLOBECOM) Dec 2016 pp 1ndash6

[210] S M Azimi O Simeone and R Tandon ldquoFundamental Limits onLatency in Small-Cell Caching Systems An Information-TheoreticAnalysisrdquo in Proc IEEE Global Commun Conf (GLOBECOM) Dec2016 pp 1ndash6

[211] M A Maddah-Ali and U Niesen ldquoFundamental Limits ofCachingrdquo CoRR vol abs12095807 2012 [Online] Availablehttparxivorgabs12095807

[212] J Zhang and P Elia ldquoFundamental Limits of Cache-Aided WirelessBC Interplay of Coded-Caching and CSIT Feedbackrdquo IEEE TransInf Theory vol 63 no 5 pp 3142ndash3160 May 2017

[213] M Ji M F Wong A M Tulino J Llorca G Caire M Effros andM Langberg ldquoOn the fundamental limits of caching in combinationnetworksrdquo in Proc IEEE Int Workshop Signal Process Advances inWireless Commun (SPAWC) Jun 2015 pp 695ndash699

[214] M M Amiri and D Gunduz ldquoFundamental Limits of Coded CachingImproved Delivery Rate-Cache Capacity Tradeoffrdquo IEEE Trans Com-mun vol 65 no 2 pp 806ndash815 Feb 2017

[215] A Sengupta R Tandon and T C Clancy ldquoFundamental Limits ofCaching With Secure Deliveryrdquo IEEE Trans Inf Forensics Securityvol 10 no 2 pp 355ndash370 Feb 2015

[216] X Peng J C Shen J Zhang and K B Letaief ldquoJoint data assignmentand beamforming for backhaul limited caching networksrdquo in ProcIEEE Pers Indoor Mobile Radio Commun (PIMRC) Sep 2014 pp1370ndash1374

[217] E Bastug M Bennis and M Debbah ldquoCache-enabled small cellnetworks Modeling and tradeoffsrdquo in Proc Int Symp on WirelessCommun Syst (ISWCS) Aug 2014 pp 649ndash653

[218] G Carofiglio L Mekinda and L Muscariello ldquoLAC Introducinglatency-aware caching in Information-Centric Networksrdquo in ProcIEEE Conf Local Comp Netw (LCN) Oct 2015 pp 422ndash425

[219] A M Girgis O Ercetin M Nafie and T ElBatt ldquoDecentralizedCoded Caching in Wireless Networks Trade-off between Storage andLatencyrdquo arXiv preprint arXiv170106673 2017

[220] A Sengupta R Tandon and O Simeone ldquoCloud RAN and edgecaching Fundamental performance trade-offsrdquo in Proc IEEE IntWorkshop Signal Process Advances in Wireless Commun (SPAWC)Jul 2016 pp 1ndash5

[221] C Dehos J L Gonzalez A De Domenico D Ktenas and L DussoptldquoMillimeter-wave access and backhauling the solution to the exponen-tial data traffic increase in 5G mobile communications systemsrdquo IEEECommun Mag vol 52 no 9 pp 88ndash95 2014

[222] A S Thyagaturu A Mercian M P McGarry M Reisslein andW Kellerer ldquoSoftware defined optical networks SDONs A compre-hensive surveyrdquo IEEE Commun Surveys Tuts vol 18 no 4 pp 2738ndash2786

[223] M Chen Y Hao M Qiu J Song D Wu and I Humar ldquoMobility-aware caching and computation offloading in 5G ultra-dense cellularnetworksrdquo Sensors vol 16 no 7 p 974 2016

[224] S H Chae J Y Ryu T Q S Quek and W Choi ldquoCooperativeTransmission via Caching Helpersrdquo in Proc IEEE Global CommunConf (GLOBECOM) Dec 2015 pp 1ndash6

[225] X Ge B Yang J Ye G Mao C Wang and T HanldquoSpatial Spectrum and Energy Efficiency of Random CellularNetworksrdquo CoRR vol abs150105368 2015 [Online] Availablehttparxivorgabs150105368

[226] S Mukherjee and I Guvenc ldquoEffects of range expansion and interfer-ence coordination on capacity and fairness in heterogeneous networksrdquoin Proc IEEE Asilomar Conf Signals Systems and Computers (ASILO-MAR) Nov 2011 pp 1855ndash1859

[227] F Boccardi R W Heath A Lozano T L Marzetta and P PopovskildquoFive disruptive technology directions for 5Grdquo IEEE Commun Magvol 52 no 2 pp 74ndash80 Feb 2014

[228] M Gregori J Gomez-Vilardebo J Matamoros and D GunduzldquoWireless Content Caching for Small Cell and D2D Networksrdquo IEEEJ Sel Areas Commun vol 34 no 5 pp 1222ndash1234 May 2016

[229] J Goseling O Simeone and P Popovski ldquoDelivery Latency Regionsin Fog-RANs with Edge Caching and Cloud Processingrdquo arXiv preprintarXiv170106303 2017

[230] J Koh O Simeone R Tandon and J Kang ldquoCloud-aided edgecaching with wireless multicast fronthauling in fog radio access net-worksrdquo in Proc IEEE Wireless Commun and Netw Conf (WCNC)Mar 2017 pp 1ndash6

[231] A Sengupta R Tandon and O Simeone ldquoPipelined Fronthaul-EdgeContent Delivery in Fog Radio Access Networksrdquo in Proc IEEEGlobecom Workshop (GC Wkshps) Dec 2016 pp 1ndash6

[232] S M Azimi O Simeone A Sengupta and R Tandon ldquoOn-line Edge Caching in Fog-Aided Wireless Networkrdquo arXiv preprintarXiv170106188 2017

[233] C Fang F R Yu T Huang J Liu and Y Liu ldquoA survey of energy-efficient caching in information-centric networkingrdquo IEEE CommunMag vol 52 no 11 pp 122ndash129 2014

[234] M Ji G Caire and A F Molisch ldquoFundamental Limits of Cachingin Wireless D2D Networksrdquo IEEE Trans Inf Theory vol 62 no 2pp 849ndash869 Feb 2016

[235] T Wirth M Mehlhose J Pilz R Lindstedt D Wieruch B Holfeldand T Haustein ldquoAn Advanced Hardware Platform to Verify 5GWireless Communication Conceptsrdquo in Proc IEEE Veh Technol Conf(VTC Spring) May 2015 pp 1ndash5

[236] X Meng J Li D Zhou and D Yang ldquo5G technology requirementsand related test environments for evaluationrdquo China Commun vol 13no Supplement2 pp 42ndash51 2016

[237] S Parkvall J Furuskog Y Kishiyama A Harada T NakamuraP Naucler and B Halvarsson ldquo5G Wireless Access - Trial Conceptand Resultsrdquo in Proc IEEE Global Commun Conf (GLOBECOM)Dec 2015 pp 1ndash6

[238] H Cao S Gangakhedkar A R Ali M Gharba and J Eichinger ldquoA5G V2X testbed for cooperative automated drivingrdquo in Proc IEEEVeh Netw Conf (VNC) Dec 2016 pp 1ndash4

[239] X Liu H Zeng N Chand and F Effenberger ldquoEfficient MobileFronthaul via DSP-Based Channel Aggregationrdquo J Light Technolvol 34 no 6 pp 1556ndash1564 Mar 2016

[240] J Zhang W Xie F Yang and Q Bi ldquoMobile edge computing andfield trial results for 5G low latency scenariordquo China Commun vol 13no Supplement2 pp 174ndash182 2016

[241] P Kela M Costa J Salmi K Leppanen J Turkka T Hiltunen andM Hronec ldquoA novel radio frame structure for 5G dense outdoor radioaccess networksrdquo in Proc IEEE Veh Technol Conf (VTC Spring)May 2015 pp 1ndash6

[242] W Jiang G Feng and S Qin ldquoOptimal cooperative content cachingand delivery policy for heterogeneous cellular networksrdquo IEEE TransMobile Comput vol 16 no 5 pp 1382ndash1393 2017

[243] X Huang Z Zhao and H Zhang ldquoLatency Analysis of Cooper-ative Caching with Multicast for 5G Wireless Networksrdquo in ProcIEEEACM Int Conf Utility and Cloud Computing (UCC) Dec 2016pp 316ndash320

[244] L Jiang G Feng and S Qin ldquoCooperative content distribution for 5Gsystems based on distributed cloud service networkrdquo in Proc IEEE IntConf Commun Workshop (ICCW) Jun 2015 pp 1125ndash1130

[245] Y Liu C Yang Y Yao and B Xia ldquoModeling and analysis on end-to-end performance of cache-enabled networksrdquo in Proc Int ConfWireless Commun Signal Process (WCSP) Oct 2016 pp 1ndash5

[246] T S Rappaport F Gutierrez E Ben-Dor J N Murdock Y Qiao andJ I Tamir ldquoBroadband Millimeter-Wave Propagation Measurementsand Models Using Adaptive-Beam Antennas for Outdoor Urban Cel-lular Communicationsrdquo IEEE Trans Antennas Propag vol 61 no 4pp 1850ndash1859 April 2013

[247] T S Rappaport S Sun R Mayzus H Zhao Y Azar K Wang G NWong J K Schulz M Samimi and F Gutierrez ldquoMillimeter wavemobile communications for 5g cellular It will workrdquo IEEE Accessvol 1 pp 335ndash349 2013

[248] B Maaz K Khawam S Tohme S Lahoud and J Nasreddine ldquoJointUser Association Power Control and Scheduling in Multi-Cell 5GNetworksrdquo in Proc IEEE Wireless Commun Netw Conf(WCNC)March 2017 pp 1ndash6

[249] F Schaich T Wild and Y Chen ldquoWaveform Contenders for 5G -Suitability for Short Packet and Low Latency Transmissionsrdquo in ProcIEEE 79th Veh Technol Conf (VTC-Spring) May 2014 pp 1ndash5

[250] W Roh J Y Seol J Park B Lee J Lee Y Kim J Cho K Cheunand F Aryanfar ldquoMillimeter-wave beamforming as an enabling tech-nology for 5G cellular communications theoretical feasibility andprototype resultsrdquo IEEE Commun Mag vol 52 no 2 pp 106ndash113February 2014

[251] R Casellas R Munoz R Vilalta and R Martinez ldquoOrchestration ofITcloud and networks From Inter-DC interconnection to SDNNFV5G servicesrdquo in Proc Intern Conf on Optical Netw Design andModel (ONDM) May 2016 pp 1ndash6

[252] J Kakar S Gherekhloo and A Sezgin ldquoFundamental limits on latencyin transceiver cache-aided HetNetsrdquo in Proc IEEE Int Symp on InfTheory (ISIT) June 2017 pp 2955ndash2959

[253] S Zhang P He K Suto P Yang L Zhao and X ShenldquoCooperative Edge Caching in User-Centric Clustered MobileNetworksrdquo CoRR vol abs171008582 2017 [Online] Availablehttparxivorgabs171008582

[254] X Wang M Chen T Taleb A Ksentini and V C M Leung ldquoCachein the air exploiting content caching and delivery techniques for 5Gsystemsrdquo IEEE Commun Mag vol 52 no 2 pp 131ndash139 February2014

[255] C Wang Y He F R Yu Q Chen and L Tang ldquoIntegration ofNetworking Caching and Computing in Wireless Systems A SurveySome Research Issues and Challengesrdquo IEEE Commun Surveys Tutsvol PP no 99 pp 1ndash1 2017

[256] M Chen Y Hao L Hu K Huang and V K N Lau ldquoGreen andMobility-Aware Caching in 5G Networksrdquo IEEE Trans on WirelessCommun vol 16 no 12 pp 8347ndash8361 Dec 2017

I Introduction
II Low Latency Services in 5G
III Sources of Latency in a Cellular Network
IV Constraints and Approaches for Achieving Low Latency
V RAN Solutions for Low Latency
- V-A Framepacket structure
- V-B Advanced Multiple Access TechniquesWaveform
- V-C Modulation and Channel Coding
- V-D Transmitter Adaptation
- V-E Control Signaling
- V-F Symbol Detection
- V-G mmWave Communications
- V-H Location-Aware Communications for 5G Networks
- V-I QoSQoE Differentiation
- V-J CRAN and Other Aspects
- - VI Core Network Solutions for low latency
  - - VI-A Core Network Entities
    - VI-B Backhaul Solutions
    - - VI-B1 General backhaul
      - VI-B2 mmWave Backhaul
      - VII Caching Solutions for Low Latency
        
        VII-A Caching for cellular network
        
        VII-B Fundamental Latency-storage trade-off in Caching
        
        VII-C Existing Caching Solutions for 5G
        
        VIII Field Tests Trials and Experiments
        
        IX Open Issues Challenges and Future Research Directions
        
        IX-A RAN Issues
        
        IX-B Core Network Issues
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 2: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

European research projects such as METIS-II MiWaveS andmmMAGIC are working to address diverse aspects of 5Gsuch as RF block and algorithm for mmWave communicationOn the other hand projects such as ADEL FANTASTIC5G SPEED 5G and Flex5GWare address hardware andfundamental building blocks for 5G while 5G CHAMPIONEuropeanKorean research project is working to implementproof of concept (PoC) of 5G network encompassing allcutting edge radio core and satellite technologies [10] Itaims to showcase the 5G PoC with latency of 1 ms on 2018Winter Olympics in PyeongChang Korea Telecommunicationvendors such as Ericsson Huawei and Nokia Siemens areworking to bring network infrastructure and UE for 5G roll outby 2020 Besides researchers from academia are working ondifferent aspectsgoals of 5G including low latency Howeverthe real field PoC and benchmarking of performance is to bedone

In the literature surveys on 5G network including archi-tecture [3] [4] SDNNFVMEC based core network [11][12] caching [13] [14] backhaul [15] resource management[16] and data centric network [17] [18] are available Apartfrom that surveys on latency reduction approaches in Internet[19] cloud computing [20] [21] and distributed networkapplications [22] [23] are also presented however to thebest of our knowledge a comprehensive survey on latencyreduction approaches in cellular networks towards 5G is notavailable yet

In this paper we present a comprehensive survey of latencyreduction solutions particularly in the context of 5G wirelesstechnology We first present the sources and fundamentalconstraints for achieving low latency in a cellular networkWe also overview an exemplary 5G network architecturewith compliance to IMT-2020 vision Finally we providean extensive review of proposed solutions for achieving lowlatency towards 5G The goal of our study is to bring allexisting solutions on the same page along with future researchdirections We divide the existing solutions into three parts(1) RAN solutions (2) Core network solutions (3) Cachingsolutions However detailed comparison of these solutions arebeyond the scope of this work

The rest of the paper is organized as follows Section IIpresents the latency critical services in 5G The sources oflatency in a cellular network are discussed in Section III Sec-tion IV reviews the fundamental constraints and approachesfor achieving low latency Three key low latency solutions inRAN core network and caching are presented in SectionsV VI and VII respectively Section VIII presents the fieldtests trials and experiments of low latency approaches Openissues challenges and future research directions are discussedin Section IX Finally concluding remarks are provided inSection X Some of the acronyms used in this paper arepresented in Table I

II LOW LATENCY SERVICES IN 5GLatency is highly critical in some applications such as

automated industrial production controlrobotics transporta-tion health-care entertainment virtual realty education andculture In particular IoT is quickly becoming a reality which

TABLE I LIST OF ACRONYMS

AS Access StratumAR Augmented RealityBC Broadcast ChannelBLER Block Error RateCCP Communication Control PortCCSE Control Channel Sparse EncodingCSIT Channel State Information at the TransmitterDTB Delivery Time per BitD2D Device to DeviceDRB Data Radio BearereMBB Enhanced Mobile BroadbandEPC Evolved Packet CoreFFT Fast Fourier TransformFDT Fractional Delivery TimeFBMC Filter Bank Multi CarrierGFDM Generalized Frequency Division MultiplexingGP Guard PeriodGTP GPRS Tunnel ProtocolGGSN Gateway GPRS (General Packet Radio Service) Ser-

vice NodeHARQ Hybrid Automatic-Repeat-RequestICI Inter Carrier InterferenceIoT Internet of ThingsITS Intelligent Transportation SystemIFFT Inverse Fast Fourier TransformITU International Telecommunications UnionITU-R International Telecommunication Union-Radio Com-

munication SectorIMT-2020 International Mobile Telecommunication System

with a target date set for 2020ISI Inter Symbol InterferenceMEC Mobile Edge ComputingMTP Machine Type CommunicationMETIS Mobile and Wireless Communications Enablers for

Twenty-Twenty (2020) Information SocietyNGMN Next Generation Mobile NetworksMCC Mission Critical CommunicationMRC-ZF Maximum Ratio Combining Zero ForcingMAC Medium Access ControlmMTc Massive Machine Type CommunicationMME Mobility Management EntityMIMO Multiple Input Multiple OutputMDP Markov Decision ProcessNFV Network Function VirtualizationNAS Non-Access StratumNDT Normalized Delivery TimeOFDM Orthogonal Frequency Division MultiplexingOOB Out of BandOW Optical WindowOLLA Outer Loop Link AdaptationPUCCH Physical Uplink Control ChannelP2P Peer-to-PeerRAN Radio Access NetworkSC-FDMA Single Carrier Frequency Division Multiple AccessSRB Signaling Radio BearerSCMA Sparse Code Multiple AccessSGW Serving GPRS GatewayTDD Time Division DuplexRAT Radio Access TechnologyuRLLC ultra Reliable Low Latency CommunicationUFMC Universal Filtered Multi-CarrierUDN Ultra Dense NetworkVR Virtual RealityVLC Visible Light CommunicationZF Zero Forcing5G Fifth Generation Mobile Network5GETLA 5G Flexible TDD based Local Area

connects anything to any other thing anytime and anywhereSmart wearable devices (smart watches glasses bracelets

InternetCloud

UsereNB

SGSNMME

Control signaling

SDN

NFV

MEC

Fog network

Caching placement

Content delivery

Generalbackhauling

mmWavebackhauling

Symbol detection

mmWave

CRAN and others

Centralized caching

Distributed caching

Controlsignaling

Symboldetection

Cloud RAN(CRAN)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(a)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(b)

[54] Short TTI

[60] Subcarrier

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 3: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

InternetCloud

UsereNB

SGSNMME

Control signaling

SDN

NFV

MEC

Fog network

Caching placement

Content delivery

Generalbackhauling

mmWavebackhauling

Symbol detection

mmWave

CRAN and others

Centralized caching

Distributed caching

Controlsignaling

Symboldetection

Cloud RAN(CRAN)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(a)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(b)

[54] Short TTI

[60] Subcarrier

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 4: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

InternetCloud

UsereNB

SGSNMME

Control signaling

SDN

NFV

MEC

Fog network

Caching placement

Content delivery

Generalbackhauling

mmWavebackhauling

Symbol detection

mmWave

CRAN and others

Centralized caching

Distributed caching

Controlsignaling

Symboldetection

Cloud RAN(CRAN)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(a)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(b)

[54] Short TTI

[60] Subcarrier

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 5: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Control signaling

SDN

NFV

MEC

Fog network

Caching placement

Content delivery

Generalbackhauling

mmWavebackhauling

Symbol detection

mmWave

CRAN and others

Centralized caching

Distributed caching

Controlsignaling

Symboldetection

Cloud RAN(CRAN)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(a)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(b)

[54] Short TTI

[60] Subcarrier

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 6: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Control signaling

SDN

NFV

MEC

Fog network

Caching placement

Content delivery

Generalbackhauling

mmWavebackhauling

Symbol detection

mmWave

CRAN and others

Centralized caching

Distributed caching

Controlsignaling

Symboldetection

Cloud RAN(CRAN)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(a)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(b)

[54] Short TTI

[60] Subcarrier

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 7: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Controlsignaling

Symboldetection

Cloud RAN(CRAN)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(a)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(b)

[54] Short TTI

[60] Subcarrier

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 8: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Symboldetection

Cloud RAN(CRAN)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(a)

LLC LLCLLC

LCCLCC

LLCGB

LLC

GB

LLCLLC

GB

180 kHz

1 ms

Frequency

Tim

eT

ime

Frequency

(b)

[54] Short TTI

[60] Subcarrier

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 9: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

[54] Short TTI

[60] Subcarrier

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 10: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Nsubcarrier

N sub-symbols

Nsubcarrier

Nfrequencysam

ples

K time samples

M frequencysamples

OFDM

GFDM

SCFDM

N=KMTime

Frequency

Xk(N+Lminus1)times1

=

Bsumi=1

VikNtimesni

Siknitimes1

(6)

Low complexity

Low latency

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 11: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Higher Lower Lower

LDPC-PEG[71]

Polarcodes[72]

100 98 667 15

100 98 667 1107

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 12: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 13: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

F Symbol Detection

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 14: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Modulation

Precoding

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE1

Symbol

estimation

Channel

estimation

Demodulation

Received bits at

UE2

Bits for UE1

Bits for UE2

eNB Transmitter

UE receiver

Propagation

channel

Modulation

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 15: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 16: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Virtual BTSpool

Backhaul

Corenetwork

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 17: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

[52] [161] [167] [169][170] [183]

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 18: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Managem

ent and A

dm

in P

lane

DFE

VIM

VNFM

NFVO

DFE DFE

APP APP APP APP

Date

Pla

ne

Co

ntr

ol

Pla

ne

Ap

p P

lane

(a)

OSSBSS

VIM

VNFM

NFVO

EMSEMSEMS

VNF VNF VNF

Virtual

Network

Virtual

Network

Virtual

Network

(b)

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 19: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

SON

[158] [162][163]ndash[165]

[52]

[153]ndash[156]

[157] [159][160] [166]

[161]

[167] [171]

[168] [172]

[169]

[170]

[185]

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 20: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

mmWavebackhaul

for mmWave networks

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 21: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

SBS

MBS MBS

SBS

MBS

SBS SBS

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 22: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Localcaching

[227]

D2Dcaching

[223]

SBScaching [228]

MBScaching

minusγ (7)

(8)

PB =

Nsumi=1

φiPBi (9)

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 23: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

Ref Usedmetric

Description

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 24: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 25: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 26: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 27: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

C Caching Issues

X CONCLUSION

ACKNOWLEDGMENT

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 28: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

REFERENCES

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 29: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 30: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 31: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 32: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 33: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Page 34: PDF - arXiv.org e-Print archive · Arif I. Sarwat, Senior Member, IEEE, and Huaiyu Dai, Fellow, IEEE Abstract—The ﬁfth generation (5G) wireless network technol-ogy is to be standardized

I Introduction
        
        
        
        
        
        
        IX-A RAN Issues
        
        
        IX-C Caching Issues
        
        X Conclusion
        
        References

Date post:	09-Apr-2018
Category:	Documents
Upload:	buikhuong
View:	243 times
Download:	11 times