Cooperative Full-Duplex Physical and MAC Layer Design in...

Research ArticleCooperative Full-Duplex Physical and MAC Layer Design inAsynchronous Cognitive Networks

Teddy Febrianto Jiancao Hou and Mohammad Shikh-Bahaei

Centre for Telecommunications Research Kingrsquos College London London WC2R 2LS UK

Correspondence should be addressed to Teddy Febrianto teddyfebriantokclacuk

Received 26 May 2017 Revised 29 August 2017 Accepted 17 September 2017 Published 19 November 2017

Academic Editor Gonzalo Vazquez-Vilar

Copyright copy 2017 Teddy Febrianto et alThis is an open access article distributed under the Creative CommonsAttribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

In asynchronous cognitive networks (CNs) where there is no synchronization between primary users (PUs) and secondaryusers (SUs) spectrum sensing becomes a challenging task By combining cooperative spectrum sensing and full-duplex (FD)communications in asynchronous CNs this paper demonstrates improvements in terms of the average throughput of both PUsand SUs for particular transmission schemes The average throughputs are derived for SUs and PUs under different FD schemeslevels of residual self-interference and number of cooperative SUs In particular we consider two types of FD schemes namelyFD transmit-sense-reception (FDr) and FD transmit-sense (FDs) FDr allows SUs to transmit and receive data simultaneouslywhereas in FDs the SUs continuously sense the channel during the transmission time This paper shows the respective trade-offsand obtains the optimal scheme based on cooperative FD spectrum sensing In addition SUsrsquo average throughput is analyzed underdifferent primary channel utilization and multichannel sensing schemes Finally new FD MAC protocol design is proposed andanalyzed for FD cooperative spectrum sensing We found optimum parameters for our proposed MAC protocol to achieve higheraverage throughput in certain applications

1 Introduction

Cognitive networks (CNs) which can dramatically improvespectrum efficiency using dynamic spectrum access (DSA)technology is a promising solution for the spectrum scarcityproblem [1ndash3] CNs allow the cognitive devices or secondaryusers (SUs) to use licensed or primary usersrsquo (PUs) frequencyspectrum in an opportunistic way while guaranteeing thequality of both systems

The majority of research works in this area have studiedthese problems in synchronous conditions where PUs andSUs are time slotted and synchronized However in realisticscenarios SUs have no information about the PUsrsquo signalswhich results in operations in asynchronous mode Jiang etal have studied some key issues for asynchronous CNs in [4]

Asynchronous cooperative spectrum sensing has beenproposed in [5 6] which shows improvement in the averagethroughput of SUs In-band full-duplex (FD) communication[7] is also a promising technology that can improve the per-formance of CNs FD spectrum sensing has been proposed

in [8] which allows SUs sense and transmit data at the sametime

In shared-spectrum full-duplex networking it is com-mon for the FD transceivers to operate in the transmit-sense mode that is to transmit and sense simultaneouslyand in the same frequency band [9 10] In this mode nodata is received during the data transmission period incontrast with standard noncognitive full-duplex scenariosIt has been shown that operating in transmit-sense modecan reduce the outage probability of the primary networksignificantly compared to the conventional listen-before-talkscheme (ie compared to the cognitive scenario where half-duplex transmission is performed following a short sensingperiod) [11]

An alternative approach to full-duplex networking inCNsis to combine sensing with data transmission and reception[12 13] Exploiting full-duplex communication capability ofthe transceivers data transmission takes place simultane-ously with data reception in this scheme as it does withstandard full-duplex networks However to allow spectrum

HindawiWireless Communications and Mobile ComputingVolume 2017 Article ID 8491920 14 pageshttpsdoiorg10115520178491920

2 Wireless Communications and Mobile Computing

sharing with the primary network the reception processis divided in time so that sensing and data receiving cantake place in different time slots while data transmissioncontinues over the entire time period that is using timedivision duplex (TDD) in reception

The advantage of transmit-sense mode approach is thatit allows continuous sensing which can in turn improvethe probability of detecting returning primary users Thisimprovement in the likelihood of detection is also a resultof more advanced learning algorithms that can be imple-mented in continuous sensing This is in contrast withdata transmit-sense-reception mode where sensing takesplace in short periods of time that is data reception andsensing are scheduled according to a time-division-duplexscheme Intermittent sensing in this mode does not allow theemployment of advanced and reliable learning algorithmswhich implies higher missed detection probability comparedto the transmit-sense mode On the other hand operationin transmit-sense-reception mode improves the secondaryusersrsquo throughput at the cost of deteriorating primary net-workrsquos performance

The authors in [12 13] have introduced adaptive full-duplex transmit-sense (FDs) and full-duplex transmit-receive-sense (FDr) to improve the performance of thespectrum-sharing mechanism These two works use energy-based sensing and waveform-based sensing respectivelyHowever both of these works have not considered cooper-ative spectrum sensing in order to further improve the spec-trum sensing accuracy On the other hand the effect on theprimary network can be alleviated by deploying cooperativesensing Cooperation among the secondary users in sensingthe licensed channel can improve the quality of detectingthe activity of primary users in the licensed spectrum Bycombining cooperative sensing and full-duplex communi-cation features full-duplex cooperative spectrum sensingmechanism is analyzed to improve the average throughputof SUs while guaranteeing PUsrsquo quality and throughputThissolution has been proposed in [14] only for time-slotted PUsand in [15] for cooperative acknowledgement The recentwork in [16] has analyzed SUsrsquo performance in non-time-slotted case without looking into the imposed effect on PUs

Motived by the above observation the contributions ofthis paper can be summarized as follows The throughputof both secondary and primary users is derived under FDrand FDs schemes as functions of residual self-interference(SI) and the number of cooperating users in spectrumsensing Furthermore the results are extended for differentprimary channel utilization Unlike our previous work in[17] in this paper we introduce the full-duplex cooperativemultichannel based FDr and FDs sensing schemes and alsofind the minimum number of cooperating secondary usersrequired in FDr scheme so that the achievable throughputfor the primary users is very close to that in the FDs schemeIn the literature the full-duplex cooperative multichannelscenarios have also been considered [8 18 19] Howeverthese works assume SUs perform the standard p-persistentcarrier sense multiple access (CSMA) mechanism for con-tention resolution on the selected channel Such distributedmechanism can effectively avoid the data crash and improve

TＩＨ TＩff

On state O state

Figure 1 On-off process channel model

the spectrum efficiency of the network On the other handby comparing with the cooperative multichannel allocationmechanism which will be considered in this paper thedistributed version has to be limited to a ldquosmallrdquo networkBased on the cooperative SUs design in this paper we alsopropose aMAC layer protocol design and analyze the averagethroughputs of SUs and PUs when utilizing TV white spaceas primary channels The results show that by increasingthe number of sensed channels the average throughput ofproposed FD schemesmaynot always be improved especiallyfor FDr scheme This is because the duration of sensing isconsidered as nontransmitting time which reduce the timefor data transmission

The rest of this paper is organized as follows The systemmodel and throughput analysis are presented in Section 2Spectrum sensing and self-interference effect are analyzedin Section 3 Formulation of secondary and primary averagethroughput for the cooperative full-duplex sensing scenario isin Sections 4 and 5 respectively The analysis for the averagethroughput optimization and its corresponding numericalresults are provided in Section 6 A newMACprotocol designis proposed and evaluated in Section 7 Finally conclusionsare drawn in Section 8

2 System Model

As discussed in abstract and introduction sections theasynchronous cognitive networks are defined that only theprimary and secondary networks are not synchronized butwithin the secondary network the secondary users will beassumed to be synchronized Moreover the primary usersin this case may or may not be synchronized In additionwe also assume that SUs always have packets to ldquoreceiverdquo as[9 20] and this assumption is meaningful if the number ofSUs is directly linked to the traffic demand in MAC protocoldesign

21 Primary Users One pair of PUs is considered to commu-nicate in half-duplex (HD) mode over 119882 spectrum bandswhere the bands are not overlapped and with the samespectrum utilization as 120573 PUsrsquo channel is modeled as an on-off process as [21] which is shown in Figure 1 The channel isconsidered in on state when it is occupied by PUs On theother hand off state means there is an available spectrumthat can be exploited by SUs The length of the on state (119879on)and off state (119879off ) follows the exponential distribution withaverages of 120588on and 120588off respectively In this case the primarychannel utilization 120573 can be calculated as

120573 = 120588on120588off + 120588on (1)

Wireless Communications and Mobile Computing 3

Primary user

Secondary user

Half-duplex

Full-duplexreceive-sense(FDr)

Full-duplexsense (FDs)

Tp Tp

Ts

Ts

Ts

Tt

Tt

Tt

middot middot middot


middot middot middotmiddot middot middot



Figure 2 Secondary users transmission reception and sensing schemes

22 Secondary Users We consider that there are119872 SUs thatcooperate in sensing the PUsrsquo spectrum bands Among these119872 cooperating SUs there is one coordinator node whichdecides whether the primary channel is available for SUsThe coordinator node will collect the sensed informationfrom 119872 collaborating SUs [5] and use it for ldquosoft decisionrdquo[22] If during the SUsrsquo transmission the coordinator noderealizes the PUrsquos return to the channel it would immediatelyinform the active SUs to stop opportunistic transmission inthe primary band Spectrum allocation by the coordinator isnot studied in this paper and the spectrum allocation schemewill be considered to be independent of the asynchronousspectrum sensing process

As shown in Figure 2 we consider two full-duplex sensingmechanisms for SUs to use the available primary usersrsquochannels opportunistically In addition we also include thehalf-duplex (HD) mechanisms for comparison Specificallyfor the HD communication sensing process SUsrsquo frame isdivided into two intervals that is sensing (119879119904) and trans-mitting (119879119905) with sampling rate of 120596119904 Energy-based sensingis applied for the sensing process of SUs 119879119901 = 119879119904 + 119879119905 isdefined as sensing period In the second scheme SUs use full-duplex communication to sense transmit and receive datawithin the same frame time 119879119901 This scheme is referred to asfull-duplex transmit-receive-sense (FDr) and it allows SUs toreceive data during the data transmitting process The thirdscheme is full-duplex transmit-sense (FDs) where SUs sensethe primary channel continuously during data transmission

23 AverageThroughputs We assume that each SU can senseV(le119882) channels over 119879119901 seconds where each channel issensed for 119879so seconds In this case we have119879119904 = 119881 sdot 119879so (2)In addition 119871 is defined as the average number of sensed idlechannels by119872 SUs which can be derived as [6]

119871 = 119882sumV=0

V sdot 119875V (V) (3)

where 119875V(V) is the probability that V idle channels can besensed by119872 SUs and by considering (1) we have [6]119875V (V) = (119871V) 120573119871minusV (1 minus 120573)V (4)

Therefore SUrsquos average throughput with V sensible chan-nels can be expressed as

120591(V)119904scheme = 119871 sdot 120591119904scheme119882 (5)

where 120591119904scheme is SUrsquos average throughput per channel for aspecific sensing scheme for example HD FDr or FDs Onthe other hand PUrsquos average throughput (ie 120591(V)

119901scheme) withmultiple channels will have the same behaviour as for thesingle channel case (ie 120591119901scheme) since all the channels areindependently distributed From following sections we willmathematically derive the average throughputs per channelfor both SU and PU in detail

3 Spectrum Sensing and SelfInterference Effect

The two fundamental measures to be evaluated in spectrumsensing are the detection probability (119875119889) and the false alarmprobability (119875119891) 119875119889 is the probability that SUs can detecta busy channel when PUs do use the channel 119875119891 is theprobability that SUs falsely detect a busy channel whereasthere is actually no PU activities

Residual self-interference (SI) in full-duplex communi-cation affects the detection probability and the probabilityof false alarms in sensing the activity of primary users Anenergy detection technique is widely deployed for detectingthe primary usersrsquo activity in the shared spectrum In full-duplex sensing that is simultaneous data transmission andspectrum sensing in the same frequency band the energyof the residual SI as a result of imperfect SI cancellationmay be mistaken for primary usersrsquo signal This in turn willincrease the false alarm probability and reduce the secondaryusersrsquo throughput Waveform-based sensing is an alternativesensing method that can alleviate this problem in full-duplexscenarios In this approach sensing the primary signals iscarried out by correlating the received samples with knownpattern samples

Another alternative to energy detection is cyclostationaryfeature detection which is based on the estimation of theFourier spectrum cyclic density and can detect weak signalsfrom primary users by only exploiting the cyclostationarityproperty of communication signals However this approach


is rather complex for implementation A different approachin detecting primary usersrsquo signals is based on tracking theprimary users by employing smart antennas and avoidingspatial interferencewith their signals through transmit beam-forming Cooperative sensing can improve the probabilityof detecting primary signals at the cost of higher computa-tion and networking complexities Using full-duplex radiostransmission and reception of data can be implementedsimultaneously for further increase in the secondary userthroughput In this paper cooperative energy-detection-based sensing is considered in the analysis

Hypotheses 1198670 and 1198671 correspond to the cases whereprimary channel is in the off state and on state respectivelyUnder the1198670 and1198671 conditions the SUsrsquo received signals attime instant 119899 (119903119898[119899]) are respectively given by1198670 119903119898 [119899] = 119909119898 [119899] + 119906119898 [119899] 1198671 119903119898 [119899] = 119909119898 [119899] + 119904119898 [119899] + 119906119898 [119899] (6)

where 119899 refers to the 119899th sample and subscript119898 denotes the119898th SU 119909119898 is the self-interference of 119898th SU and 119904119898 is thesignal transmitted by PU and received at the 119898th SU Thebackground noise which is assumed as circular symmetriccomplex Gaussian is denoted by 119906119898 The mean of 119906119898 is zeroand the variance is 1205902

The overall energy statistic of primary channel received atthe coordinator SU (119877) is given by [5]

119877 = 1119872120596119904119879119904 119872sum119898=1 120596119904119879119904sum119899=1 1003816100381610038161003816119903119898 [119899]10038161003816100381610038162 (7)

In addition 119875119889 and 119875119891 are given by [8]

119875119889 = Pr [119877 ge 120576th | 1198671]= 119876( 120576th1205902 minus ((119880 minus 119886) 119880) 120581SNR119904119901 minus 1radic((119880 minus 119886) 1198802) (120581SNR119904119901 + 1)2 + 1198861198802)119875119891 = Pr [119877 ge 120576th | 1198670]= 119876( 120576th1205902 minus (119889119880) 120581SNR119904119901 minus 1radic(1198891198802) (120581SNR119904119901 + 1)2 + (119880 minus 119889) 1198802)

(8)

where 120576th is the energy detection threshold119880 is the number ofprimary samples during the sensing period 119886 is the numberof samples during off state before primaryrsquos return to an active(ON) state and 119889 is the number of samples during in whichthe primary is at on state before becoming inactive SNR119909119910is the signal-to-noise ratio at receiver 119909 due to the signaltransmitted by transmitter 119910 119909 119910 isin 119901 119904 where 119901 and 119904respectively refer to a primary and secondary user 120581 (0 lt 120581 le1) represents the self-interference mitigation coefficient [823] If 120581 is high this means that self-interference is mitigatedwell On the other hand low values of 120581 imply that selfinterference at the receiver is high119876(sdot) is the complementarydistribution function of standard Gaussian which is definedby

119876 (119909) = 1radic2120587 intinfin119909 119890(minus11990522)119889119905 (9)

Considermultichannel sensing scenarios where PUsmaychange their states (ie on or off) during SUsrsquo sensing periodIn this case there are four possible cases whenwe calculate119875119889and 119875119891 in order to obtain the average throughputs

Case 1 PU is not active during the SUsrsquo sensing period Inthis case 119875119891 which is used for SUsrsquo achievable throughputcalculation can be expressed as

119875119891 = 119876((120576th1205902 minus 1)radic119872120596119904119879119904) (10)

where this equation is derived by setting 119889 = 0 and 119880 =119872120596119904119879119904Case 2 PU is always active during the SUsrsquo sensing periodIn this case 119875119889 which is used for PUrsquos achievable throughputcalculation can be expressed as

119875119889 = 119876((120576th1205902 minus 120581SNR119904119901 minus 1)radic 119872120596119904119879119904(120581SNR119904119901 + 1)2) (11)

where here we set up 119886 = 0 and 119880 = 119872120596119904119879119904Case 3 PU is firstly active until the 119889-th sample and thennot active for the rest In this case 119875119889 which is used tocalculate the PUrsquos achievable throughput is derived from (11)For calculating the SUsrsquo throughput apart from 119875119891 in (10)they still need to take the following 119875119891 into account which isgiven by

119875119891 = 119876( 120576th1205902 minus ((119889 minus lfloor119889119880rfloor) 119880) 120581SNR119904119901 minus 1radic((119889 minus lfloor119889119880rfloor) 1198802) (120581SNR119904119901 + 1)2 + (119880 minus 119889 + lfloor119889119880rfloor) 1198802) (12)

where 119880 = 119872120596119904119879119904 and lfloor119886rfloor represents the maximum integerthat is smaller than 119886 Case 4 PU is firstly not active until the 119886-th sample and

then active for the rest In this case 119875119891 which is used to


calculate the SUsrsquo average throughput is derived from (10)For calculating the PUrsquos average throughput apart from 119875119889 in (11) they also need to take the following 119875119889 into account

which is given by

119875119889 = 119876( 120576th1205902 minus ((119880 minus 119886 + lfloor119886119880rfloor) 119880) 120581SNR119904119901 minus 1radic((119880 minus 119886 + lfloor119886119880rfloor) 1198802) (120581SNR119904119901 + 1)2 + (119886 minus lfloor119886119880rfloor) 1198802) (13)

where 119880 = 1198721205961199041198791199044 Secondary Usersrsquo AverageThroughput Analysis

41 SUsrsquo Achievable Data Rate During the off state themaximum achievable data rate (1198631199040) for SUs under the effectof background noise and residual SI is

1198631199040 = log2 (1 + SNR1199041199041 + (1 minus 120581) SNR119904119904) (14)

If the coordinator falsely detects that there is no primaryactivity in the on state the achievable data rate (1198631199041) for SUsis

1198631199041 = log2 (1 + SNR1199041199041 + SNR119904119901 + (1 minus 120581) SNR119904119904) (15)

In this paper the effect of multiuser interferenceis assumed to be controlled and cancelled effectivelyusing physical layer and MAC layer techniques for exam-ple through adaptive beamforming as in [24] adaptiveratepower control and schedulingmechanisms as in [25 26]

42 SUsrsquo Average Throughput in Half-Duplex Mode In orderto compare our proposed full-duplex based sensing protocolsto the existing protocols in this subsection we first introducethe SUsrsquo average throughput in conventional half-duplexmode As illustrated in Figure 3(a) there are four differentstates that should be considered to formulate the half-duplex(HD) SUsrsquo average throughput (120591119904HD) As derived in [5]assuming asynchronicity between SUs and PUs 120591119904HD can beexpressed as

120591119904HD = 11sum119894=00

119875 [119878119894HD] 119862119894HD (16)

where 119875[119878119894HD] forall119894 are defined as probability of event 119878119894HDoccurred inHD scheme and following the assumedONOFFdistributions they can be expressed as

119875 [11987800HD] = 120588off120588off + 120588on 119890minus119879119901120588off (17)

119875 [11987801HD] = 120588off120588off + 120588on (1 minus 119890minus119879119901120588off ) (18)

119875 [11987810HD] = 120588on120588off + 120588on (1 minus 119890minus119879119901120588on) (19)

119875 [11987811HD] = 120588on120588off + 120588on 119890minus119879119901120588on (20)

In addition 119862119894HD is data rate for 119878119894HD event for HD schemewhich can be found in [5]

43 SUsrsquo Average Throughput in Full-Duplex Transmit-Sense-Reception Mode Following the similar approach as in theHD scheme under the same four different cases the averagethroughput for FDr scheme (120591119904FDr) can be obtained as

120591119904FDr = 11sum119894=00

119875 [119878119894FDr] 119862119894FDr (21)

where 119875[119878119894FDr] forall119894 are defined as probability of event 119878119894FDroccurring in FDr scheme and 119862119894FDr is the achievablethroughput for 119878119894FDr event for FDr scheme In this case asillustrated in Figure 3(b) we have

119875 [119878119894FDr] = 119875 [119878119894HD] forall119894 (22)

For an ideal single channel point-to-point communica-tion the achievable throughput in FDr is twice as high as thatinHD schemeHowever due to the SI effect (when 0 le 120581 lt 1)on11986311990401198631199041 and 119875119889 the achievable throughput will be lower


Tp

Ts Tt

On stateO state

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

Tp

Ts Tt

On stateO stateS00HD S01HD

S10HD S11HD

(a)

S00FDr S01FDr

S10FDr S11FDr

Tp

Ts Tt

On stateO stateTp

Ts

On stateO state

Tt

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

(b)

Figure 3 Four possible states in (a) HD and (b) FDr scenario

than in the perfect SI cancellation case (120581 = 1) 119862119894FDr can becalculated as

11986200FDr = 2119875119891119879119901 minus 119879119904119879119901 119863119904011986201FDr= 2119879119901 (1198631199040119875119891 minus 1198631199041119875119889) (120588off minus (119879119901 + 120588off) 119890minus119879119901120588off )1 minus 119890minus119879119901120588off+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879119904)11986210FDr

= 2119879119901 (1198631199041119875119889 minus 1198631199040119875119891) (120588on minus (119879119901 + 120588on) 119890minus119879119901120588on)1 minus 119890minus119879119901120588on+ 2119879119901 (1198751198911198631199040119879119901 minus 1198751198891198631199041119879119904)

11986211FDr = 2119875119889119879119901 minus 119879119904119879119901 1198631199041

(23)

where 119875119889 = 1 minus 119875119889 and 119875119891 = 1 minus 11987511989144 SUsrsquo Average Throughput in Full-Duplex Transmit-SenseMode In contrast with HD and FDr schemes in FDsthe transmission time during a frame is not constant SUscontinuously sense the channel and can immediately start orstop transmission based on the sensing result Hence onlytwo states are studied for average throughput calculation asshown in Figure 4 The probabilities of the events 11987800FDs and

11987811FDs are defined as the probability of being in off and onstate respectively and they can be expressed as

119875 [11987800FDs] = 120588off120588off + 120588on119875 [11987811FDs] = 120588on120588off + 120588on (24)

Average data rate during 11987800FDs and 11987811FDs can be calcu-lated by 11986200FDs = 1198751198911198631199040 (25)11986211FDs = 1198751198891198631199041 (26)

Therefore SUsrsquo average throughput (120591119904FDs) for FDsscheme is

120591119904FDs = 11sum119894=00

119875 [119878119894FDs] 119862119894FDs (27)

5 Primary Usersrsquo AverageThroughput Analysis

51 PUsrsquo Achievable Data Rate ThePUsrsquo achievable data ratewithout interference (1198631199010) from SU is

1198631199010 = log2 (1 + SNR119901119901) (28)

The PUsrsquo achievable data rate with interference (1198631199011)from SU due to miss-detection can be calculated as

1198631199011 = log2 (1 + SNR1199011199011 + SNR119901119904) (29)

52 PUsrsquo Average Throughput When Secondary Is in Half-Duplex or Full-Duplex Transmit-Sense-Reception Modes ThePUsrsquo average throughput can be calculated by modeling the


S00FDs

Ts

On state O stateO state

Ts

On state On stateO state

S11FDs



Figure 4 Two possible conditions in FDs scenario

Ts

Ts Tt

Tt

On state Off state

Primary user

Secondary userHalf-duplex

Full-duplexreceive-sense

Dp1 Dp1 Dp0Dp0

ＩＨ

Tp

Figure 5 Illustration of PUrsquos average throughput when SU uses HDor FDr schemes

on state event as a time-slotted frame of size 119879119901 as shownin Figure 5 The average number of slots (119887119879119901) is definedas ratio of the primary user on sate (120588on) to sensing period(119879119901) Furthermore the event where SUs transmit during theon state can be modeled as binomial distribution with theprobability of occurrence given by 1minus119875119889 119875[119878119895119901] is defined asprobability that 119895 frames of SUs are transmitted during 120588on ofPUwhen SUs useHDor FDr scheme which can be expressedas

119875 [119878119895119901] = (119887119879119901119895 )119875119889119895119875(119887119879119901minus119895)119889 (30)

where

119887119879119901 = [120588on119879119901 ] (31)

and [sdot] is rounded to nearest integer number operator Theaverage data rate (119862119895119901) when 119895 frames of SUs are transmittedduring 120588on of PU can be calculated as

119862119895119901 = (119895 sdot 1198631199011 sdot 120588on120588off + 120588on sdot 119879119901120588on)

Ts

On state Off state

Primary user

Secondary userFull-duplex

sense

Dp1 Dp1 Dp0Dp0

ＩＨ

Figure 6 Illustration of PUrsquos average throughput when SU uses FDsscheme

+ (1198631199010 sdot 120588on120588off + 120588on sdot 120588on minus 119895119879119901120588on )= 120588on1198631199010 minus 119895119879119901 (1198631199010 minus 1198631199011)120588off + 120588on

(32)

Finally the PUsrsquo average throughput for HD (120591119901HD) andFDr (120591119901FDr) cases can be formulated as

120591119901HD = 120591119901FDr = 119887119879119901sum119895=0

119875 [119878119895119901] 119862119895119901 (33)

53 PUrsquos Average Throughput When Secondary User Is inFull-Duplex Transmit-Sense Mode In the same fashion asin Section 52 PUsrsquo average throughput when SUs use FDsscheme can be calculated as shown in Figure 6 Instead ofdividing by 119879119901 120588on is divided by 119879119904 to estimate the numberof slots 119887119879119904 119875[119878119897119901FDs] is defined as probability that 119897 framesof SUs are transmitted during 120588on of PU when SUs use FDsscheme which can be expressed as

119875 [119878119897119901FDs] = (119887119879119904119897 ) 119875119889119897119875(119887119879119904minus119897)119889 (34)

where 119887119879119904 = [120588on119879119904 ] (35)

The average data rate (119862119897119901FDs) when 119897 frames of SU aretransmitted during 120588on of PU can be calculated as

119862119897119901FDs = (119897 sdot 1198631199011 sdot 120588on120588off + 120588on sdot 119879119904120588on)+ (1198631199010 sdot 120588on120588off + 120588on sdot 120588on minus 119897119879119904120588on )

= 120588on1198631199010 minus 119897119879119904 (1198631199010 minus 1198631199011)120588off + 120588on (36)

The PUrsquos average throughput when SUs use FDs scheme(120591119901FDs) can be calculated as

120591119901FDs = 119887119879119904sum119897=0

119875 [119878119897119901FDs] 119862119897119901FDs (37)


Table 1 Proposed MAC simulation parameters

Parameter Value Description119879119901 32ms Sensing period120581 099 Self-interference mitigation coefficient119882 11 [28] Number of primary userrsquos channels120576th1205902 103 [5] Received signal power-to-noise ratio thresholdSNR119904119904 10 dB [5] Average signal-to-noise ratio received by secondary user from secondary signalSNR119904119901 minus10 dB [5] Average signal-to-noise ratio received by secondary user from primary signalSNR119901119901 10 dB Average signal-to-noise ratio received by primary user from primary signalSNR119901119904 minus10 dB Average signal-to-noise ratio received by primary user from secondary signal120596119904 100 kHz

[5] Sampling rate120588off 640ms Average length of off state for primary user120588on 160ms Average length of on state for primary user

6 Analysis for Physical Layer Method andNumerical Results

Based on the above derived average throughputs for bothsecondary and primary users per channel the generalisedaverage throughput for multichannel case can be formulatedby inserting (14) (19) and (25) into (5) respectively forsecondary users As discussed in Section 2 for the primaryusers the average throughput for multichannel case is equalto the ones for single channel case which are given by (31)and (35) Then the optimization problems which aim tomaximize the secondary users average throughput can beexpressed as

max119898V119879119904

120591(V)119904scheme = 119871 sdot 120591119904scheme119882 st 120591(V)119901scheme ge 119877119901scheme

SNR119909119910 le SNR119909119910 forall119909 119910(38)

where 119898(le119872) is number of active secondary users 119877119901scheme

is primary userrsquos minimum rate constraint and SNR119909119910 isthe SNR upper limit for (119909 119910) link In order to obtain theoptimal solution of problem (36) we can implement thefirst-order derivation of the objective function with respectto either 119898 V or 119879119904 and then set the derived equation tozero if there is unique root Alternatively numerical resultscan be implemented to help to find the optimal solution Inthe following subsections numerical results are provided forboth primary and secondary users and the parameters usedin the numerical results are shown in Table 1 which are in linewith those in [5] for fair comparison

61 Secondary Usersrsquo Average Throughput In Figure 7 SUsrsquoaverage throughput is presented versus number of cooper-ating SUs (119872) for different schemes and 120581 values It showsthat FDr scheme achieves higher average throughput forSUs compared to FDs and HD This is due to the longertransmitting time (119879119905) in FDr compared to the other schemes

50 15 2010Number of cooperative secondary users M

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

15

2

25

3

35

4

45

5

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)

Figure 7 SUrsquos average throughput versus119872 for different schemesand different value of 120581

The SUrsquos average throughput of different schemes mono-tonically increases with the number of cooperating SUs119872As expected it shows that cooperative sensing offers betterperformance compared to the noncooperative case (119872 =1) This figure also demonstrates the effect of 120581 on the SUsrsquoaverage throughput for FDr and FDs schemes noting that inHD scheme SI is zero (120581 = 1) By decreasing 120581 the averagethroughput for both FDs and FDr deteriorates slightly

62 Primary Usersrsquo Average Throughput Figure 8 shows theaverage throughput of PU versus the number of SUs (119872) fordifferent schemes and 120581 values Although for the SUsrsquo averagethroughput FDr outperforms FDs andHD for the PUsrsquo aver-age throughput FDr gives the worse performance especially


5 10 15 200Number of cooperative secondary users M

035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

No SU activity99 guarantee

90 guarantee

Figure 8 PUrsquos average throughput versus119872 for different schemesand 120581for lower119872 Indeed it shows an average throughput trade-offbetween SUs and PUs FDs gives similar average throughputperformance as no SUs active case even the number ofcooperative SUs is equal to one This is because secondaryusers incorporate continuous sensing and cooperation inthis case does not provide much performance improvementfor primary users Furthermore it can reduce transmittingtime of the SU 119879119905 during on state As a result PUs cantransmit in the absence of SUs for most of the time duringthe on state

According to this figure the average throughput of PUincreases with 119872 especially when SUs employ FDr or HDschemes It shows that the use of cooperative sensing outper-forms noncooperative sensing from both PU and SU pointsof view The graphs also reveal that 120581 plays an importantrole in PUrsquos performance In FDr case a significant gain inPUrsquos average throughput is achieved for higher values of 120581The reason is that the higher 120581 would increase the 119875119889 whichis in turn closely related to the average throughput of thePUs When 119875119889 increases the average throughput of PUs willincrease As seen from the figure the PUsrsquo average throughputfor HD case is the same as for FDr when 120581 is equal to one63 Multi-Channel Sensing Results In Figure 9 SUsrsquo averagethroughput in multichannel sensing case is presented versusnumber of cooperating SUs (119872) for different schemes with120581 = 099 119882 = 11 119879so = 1ms and 120573 = 02 Whatis shown here is that an increasing number of channels 119881sensed by SUs will increase the average throughput This isdue to the fact that increasing 119881 will increase the sensingtime 119879119904 so that the probability that SUs detect idle channelsis increased as well In addition according to this figure FDr

0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)

2 3 4 5 6 7 8 9 101Number of cooperative secondary users M

HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1

Figure 9 SUrsquos average throughput versus119872 for different schemesand values for119881 andwith 120581 = 099119882 = 11119879so = 1ms and 120573 = 02still outperforms HD and FDs schemes for the multichannelsensing case When 119872 increases average throughput willalso increase By increasing 119872 the false alarm probabilitywill be reduced while the number of sensed idle channelswill increase This is consistent with our previous results(Figure 7) It is worthwhile to note that for multichannelsensing case the sensing time 119879119904 increases as the numberof multichannels increase In this case with a fixed 119879119901 thedetection probability 119875119889 will increase and 119887119879119904 in (33) willdecrease so that the primary usersrsquo average throughput willbe affected

7 Proposed MAC Protocol Design

71 Deployment Architecture Our MAC design is based ona limited infrastructure support architecture in cognitivevehicular networks [27] Road side units (RSU) as definedin IEEE 80211p standard are placed on the road These playthe role of coordinator nodes in the cooperative cognitiveradio network taking care of spectrum selection and accessOn the other hand vehicular nodes act as secondary users(SUs) Figure 10 illustrates the network architecture for ourproposed MAC

72 Proposed MAC Framework Our MAC framework isdeveloped based on a slotted time MAC structure illustratedin Figure 11 There is one control channel (CCC) and119882 pri-mary licensed channels within 119879119901 time duration Moreoverthe proposed MAC protocol is divided into four phases

The first phase is sensing phase (SP) Each SU whichhas packets to send senses 119881 channels from 119882 licensedchannels during this phase Energy detection technique is


Road side unit (RSU)coordinator of SUs node

VehiclesSUs node

Cooperation

Figure 10 MAC deployment topology

used to detect PUrsquos activity Furthermore SUs listen to CCCfor broadcast information

The second phase is reporting and contending phase(RCP) In this phase the SU informs the coordinator aboutthe sensing result and its intention to use licensed channelsTheCCC frame is split into119872119909minislots Each SU selects oneminislot based on the broadcast information received duringthe SP phase

The third phase is the broadcast phase (BP) After receiv-ing the sensing result information the coordinator performsspectrum decision access and SU management Spectrumdecision is performed by selecting idle channels based onoverall energy statistic calculation as in [5] Furthermore 119871identified available licensed channels are allocated among119872119888SUs If 119871 lt 119872119888 only the first of 119871 SUs can use one channelper userThe remaining SUs will be allocated in the next timeslot If 119871 gt 119872119888 then each SU may be eligible for lfloor119871119872119888rfloorchannels lfloorsdotrfloor is rounded down to nearest integer operatorIn relation to management of SUs the coordinator removesinactive SUs and adds new SUs into its list of 119872119888 SUs Inaddition broadcast is also used for acknowledging arrival ofnew SUs

Broadcast messages contain the following information

(1) Number of current SUs in the cooperative network(119872119888) this information is required for the new SU tojoin the cooperative network The new SU selects arandom minislot number from119872119888 + 1 to119872119909

(2) Available licensed channels for specific SUs trans-mission mode FDr or FDs according to Figure 11

is decided by the coordinator based on 119872119888 values119872th is defined as a threshold which allows each SUto use FDr mode based on PUrsquos performance guar-antee Coordinator selects FDs transmission modeby default However If 119872119888 gt 119872th and both thecoordinator and the SU have packets to send thenFDr can be selected

(3) Synchronization information for all SUs in the coop-erative network

The last phase is data transmission phase (DTP) If anSU uses FDr mode then data and acknowledgement canbe transmitted in both uplink and downlink directions FDsmode allows SU to send data in uplink or downlink directionand sense at the same time During the DTP phase if theSU detects primary user activity it will stop transmitting thecurrent data or acknowledgement

73 Proposed MAC Protocol Average Throughput In thissection the proposed MAC is evaluated using averagethroughput as performance metric

731 Proposed MACrsquos Average Throughput Using Full-DuplexTransmit-Sense-Reception Mode Average throughput can becalculated using the same method as in Section 43 How-ever average throughput calculation in the proposed MACrequires preparation time (119879pr) which consists of sensingtime (119879119904) reporting time (119879119903) and broadcasting time (119879119887)Figure 12 shows the frame structure of the proposed MAC

119879prFDr = 119879119904 + 119879119903 + 119879119887 = 119881 sdot 119879so +119872119909 sdot 119879ro + 119879119887 (39)

The average throughput for the proposed MAC (120591119904MFDr) canbe calculated as

120591(119881119872119909)119904MFDr = 119871 sdot sum11119894=00 119875 [119878119894FDr] 119862119894MFDr119882 (40)

where

11986200MFDr = 2119875119891119879119901 minus 119879prFDr119879119901 119863119904011986201MFDr

= 2119879119901 (1 minus 119890minus119879119901120588off )1198631199040119875119891minus 21198631199041119875119889 (120588off minus (119879119901 + 120588off) 119890minus119879119901120588off )119879119901 (1 minus 119890minus119879119901120588off )


SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp

1 middot middot middot Mx BC

Occupied

Idle

FDs

Sensing

＄ＮＮＬＨＭＧＣＭＭＣＩＨ uarr

＄ＮＮＬＨＭＧＣＭＭＣＩＨ darr

＄ＮＮＬＨＭＧＣＭＭＣＩＨ uarrdarr

Figure 11 Proposed MAC framework

+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr

= 2119879119901 1198631199041119875119889 minus 1198631199040119875119891 (120588on minus (119879119901 + 120588on) 119890minus119879119901120588on)1 minus 119890minus119879119901120588on+ 2119879119901 (1198751198911198631199040119879119901 minus 1198751198891198631199041119879prFDr)

11986211MFDr = 2119875119889119879119901 minus 119879prFDr119879119901 1198631199041(41)119862119894MFDr is the achievable throughput of the proposed

MAC for 119878119894FDr event in the FDr scheme

732 Proposed MACrsquos Average Throughput Using the Full-Duplex Transmit-Sense Mode In FDs scheme 119879pr only con-sists of reporting (119879119904) and broadcasting time (119879119887) becausesensing time (119879119904) is in parallel with transmission time (119879119905)119879prFDs = 119879119903 + 119879119887 = 119872119909 sdot 119879ro + 119879119887 (42)

Following the same method as in Section 44 the averagethroughput for the proposed MAC can be calculated as

120591(119881119872119909)119904MFDs = 119871 sdot sum11119894=00 119875 [119878119894FDs] 119862119894MFDs119882 (43)

where 119862119894MFDs is the achievable throughput of our proposedMAC for 119878119894FDs event for FDs scheme

11986200MFDs = 1198751198911198631199040 (119879119901 minus 119879prFDs)119879119901 11986211MFDs = 1198751198891198631199041 (119879119901 minus 119879prFDs)119879119901 (44)

74 Proposed MAC Protocol Numerical Results and AnalysisTables 1 and 2 summarize the parameters for evaluation of ourproposed MAC protocol In order to perform an evaluationbased on the realistic scenario (ie utilizing TV white spacechannels) the number of primary channels (119882) is establishedbased on system B TV channels inWestern Europe andmanyother countries in Africa Asia and the Pacific [28] It is


Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing

Figure 12 Proposed MAC frame structure


Parameter Value Description119879119901 32ms Sensing period119879so 1ms Sensing time of each channel119879ro 05ms Reporting time for one minislot119879119887 2ms Broadcasting time

assumed the cooperative networks are saturated when thenumber of cooperative users (119872) is equal to the maximumnumber of minislot (119872119909)

Figure 13 demonstrates average throughput of the pro-posedMAC for various number of channels sensed by SU (119881)and different number ofminislots (119872119909) It shows FDr schemehas a higher average throughput compared to FDs scheme Inboth schemes increasing 119872119909 improves average throughputuntil the optimum value of 119872119909 It deteriorates slightly afterreaching the optimum value Here 119872119909 can be linked withthe number of cooperative users which have been activatedSpecifically119872119909 increases throughput by improving the num-ber of SUs (119872) that perform cooperative spectrum sensingFurthermore cooperative spectrum sensing improves theaverage throughput At the same time minislots consume

0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

5 10 15 200Number of minislots Mx

FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1

Figure 13 Proposed MACrsquos average throughput versus 119872119909 fordifferent schemes and value for 119881


allocated time in a framework which cannot be used fordata transmission As a result increasing 119872119909 shortens thetransmission time in one frame In general when throughputgain from cooperative spectrum sensing cannot compensatefor throughput loss due to allocated minislot in a frame itreduces the average throughput

The number of sensed channels (119881) has a differenteffect for FDr and FDs schemes In FDs scheme increasingthe value of 119881 slightly improves the average throughputThis is due to the fact that increasing 119881 will increase theprobability that SUs detect idle channels Different from FDsscheme which only has throughput gain in FDr schemethere is throughput loss due to sensing time Sensing timeis considered as nontransmitting time which reduces datatransmission time As a result an increment of 119881 willdecrease slightly the average throughput When throughputgain cannot compensate for throughput loss increasing 119881deteriorates the average throughput

Based on the numerical results the optimum values for119881 and119872119909 are shown to be 3 and 11 respectively It producesthe maximum average throughput of 23436 bitssecHzwhen operating in FDr mode and 13387 bitssecHz inFDs mode In other words the stated parameter values canbe implemented for optimum proposed MAC protocol inthe cooperative cognitive network while utilizing system BTV channels It is worthwhile to note that our proposedcooperative full-duplex spectrum sensing technique needs toset up a coordinator to allocate the spectrum resource to theSUs Such scheme is quite different with the distributed usercontention based resource allocation (eg see [9]) In thiscase fair comparison between these schemes is difficult toobtain In addition like the work in [9] the author proposedthe frame fragmentation during the data transmission phasein order to protect the PUs Such design makes the datatransmission model quite different from ours On the otherhand we compare our proposed full-duplex scenarios withthe conventional half-duplex scheme in order to show theperformance improvement

8 Conclusion

Performance trade-offs for FDr FDs and HD schemes arefound by analyzing the average throughput of both PUs andSUs under multichannel spectrum sharing and consideringthe effect of residual self-interference The FDr scheme canoffer similar achievable PU throughput as in FDs by incorpo-rating a sufficient number of cooperating SUs for full-duplexcooperative sensing In addition it is shown that the resultis consistent for different primary channel utilization andparameters setup Furthermore the proposed MAC protocolbased on numerical results is designed and evaluated Theoptimum parameter sets for proposed MAC are found to beimplemented in particular cognitive networks in the future(eg the cognitive vehicular network by utilizing system BTV channels)

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

This work was partially supported by the Engineering andPhysical Science Research Council (EPSRC) through theSENSE Grant EPP0034861

References

[1] J Mitola III and G Q Maguire Jr ldquoCognitive radio makingsoftware radios more personalrdquo IEEE Personal Communica-tions vol 6 no 4 pp 13ndash18 1999

[2] S Haykin ldquoCognitive radio brain-empowered wireless com-municationsrdquo IEEE Journal on Selected Areas in Communica-tions vol 23 no 2 pp 201ndash220 2005

[3] A Shadmand K Nehra and M Shikh-Bahaei ldquoCross-layerdesign in dynamic spectrum sharing systemsrdquo EURASIP Jour-nal on Wireless Communications and Networking vol 2010article 458472 2010

[4] C Jiang N C Beaulieu L Zhang Y Ren M Peng and H-HChen ldquoCognitive radio networks with asynchronous spectrumsensing and accessrdquo IEEE Network vol 29 no 3 pp 88ndash952015

[5] C Jiang N C Beaulieu and C Jiang ldquoA novel asynchronouscooperative spectrum sensing schemerdquo in Proceedings of the2013 IEEE International Conference on Communications ICC2013 pp 2606ndash2611 Hungary June 2013

[6] C Jiang C Jiang and N C Beaulieu ldquoA contention-basedwideband DSA algorithmwith asynchronous cooperative spec-trum sensingrdquo in Proceedings of the 2013 IEEE Global Commu-nications Conference GLOBECOM 2013 pp 1069ndash1074 USADecember 2013

[7] A Sabharwal P Schniter D Guo D W Bliss S Rangarajanand R Wichman ldquoIn-band full-duplex wireless challenges andopportunitiesrdquo IEEE Journal on Selected Areas in Communica-tions vol 32 no 9 pp 1637ndash1652 2014

[8] W Cheng X Zhang and H Zhang ldquoFull-duplex spectrum-sensing and MAC-protocol for multichannel nontime-slottedcognitive radio networksrdquo IEEE Journal on Selected Areas inCommunications vol 33 no 5 pp 820ndash831 2015

[9] L T Tan and L B Le ldquoDistributed MAC Protocol Designfor Full-Duplex Cognitive Radio Networksrdquo in Proceedings ofthe GLOBECOM 2015 - 2015 IEEE Global CommunicationsConference pp 1ndash6 San Diego CA USA December 2015

[10] Y Liao T Wang L Song and Z Han ldquoListen-and-TalkProtocol Design and Analysis for Full-Duplex Cognitive RadioNetworksrdquo IEEE Transactions on Vehicular Technology vol 66no 1 pp 656ndash667 2017

[11] W Afifi and M Krunz ldquoIncorporating Self-Interference Sup-pression for Full-duplex Operation in Opportunistic SpectrumAccess Systemsrdquo IEEE Transactions on Wireless Communica-tions vol 14 no 4 pp 2180ndash2191 2015

[12] L T Tan and L B Le ldquoDesign and optimal configuration of full-duplex MAC protocol for cognitive radio networks consideringself-interferencerdquo IEEE Access vol 3 pp 2715ndash2729 2015

[13] W Afifi and M Krunz ldquoAdaptive transmission-reception-sensing strategy for cognitive radios with full-duplex capabil-itiesrdquo in Proceedings of the 2014 IEEE International Symposiumon Dynamic Spectrum Access Networks DYSPAN 2014 pp 149ndash160 USA April 2014

[14] Y Liao T Wang L Song and B Jiao ldquoCooperative spectrumsensing for full-duplex cognitive radio networksrdquo inProceedings


of the 2014 IEEE International Conference on CommunicationSystems IEEE ICCS 2014 pp 56ndash60 China November 2014

[15] V Towhidlou and M Shikh-Bahaei ldquoCooperative ARQ in fullduplex cognitive radio networksrdquo in Proceedings of the 27thIEEE Annual International Symposium on Personal Indoor andMobile Radio Communications PIMRC 2016 Spain September2016

[16] S HaW Lee J Kang and J Kang ldquoCooperative spectrum sens-ing in non-time-slotted full duplex cognitive radio networksrdquo inProceedings of the 13th IEEEAnnual Consumer Communicationsand Networking Conference CCNC 2016 pp 820ndash823 usaJanuary 2016

[17] T Febrianto and M Shikh-Bahaei ldquoOptimal full-duplex coop-erative spectrum sensing in asynchronous cognitive networksrdquoinProceedings of the 3rd IEEEAsia PacificConference onWirelessand Mobile APWiMob 2016 pp 1ndash6 Indonesia September2016

[18] L T Tan and L B Le ldquoMulti-channel MAC protocol for full-duplex cognitive radio networks with optimized access controland load balancingrdquo in Proceedings of the 2016 IEEE Interna-tional Conference on Communications ICC 2016 Malaysia May2016

[19] T-N Hoan V-V Hiep and I-S Koo ldquoMultichannel-sensingscheduling and transmission-energy optimizing in cognitiveradio networks with energy harvestingrdquo Sensors vol 16 no 4article no 461 2016

[20] G Bianchi ldquoPerformance analysis of the IEEE 80211 distributedcoordination functionrdquo IEEE Journal on Selected Areas inCommunications vol 18 no 3 pp 535ndash547 2000

[21] C Jiang Y Chen and K J R Liu ldquoA renewal-theoreticalframework for dynamic spectrum access with unknown pri-mary behaviorrdquo in Proceedings of the 2012 IEEE Global Commu-nications Conference GLOBECOM 2012 pp 1422ndash1427 USADecember 2012

[22] K J R Liu and B Wang ldquoCognitive radio networking andsecurity A Game-Theoretic viewrdquo Cognitive Radio Networkingand Security A Game-Theoretic View pp 1ndash601 2010

[23] W Cheng X Zhang and H Zhang ldquoOptimal dynamic powercontrol for full-duplex bidirectional-channel based wirelessnetworksrdquo in Proceedings of the 32nd IEEE Conference onComputer Communications IEEE INFOCOM 2013 pp 3120ndash3128 Italy April 2013

[24] J Hou N Yi and YMa ldquoJoint space-frequency user schedulingfor MIMO random beamforming with limited feedbackrdquo IEEETransactions on Communications vol 63 no 6 pp 2224ndash22362015

[25] A Shadmand and M Shikh-Bahaei ldquoMulti-user time-frequency downlink scheduling and resource allocation forLTE cellular systemsrdquo in Proceedings of the IEEE WirelessCommunications and Networking Conference 2010 WCNC2010 Australia April 2010

[26] A Kobravi and M Shikh-Bahaei ldquoCross-layer adaptive ARQand modulation tradeoffsrdquo in Proceedings of the 18th AnnualIEEE International Symposium on Personal Indoor and MobileRadio Communications PIMRCrsquo07 Greece September 2007

[27] S Basagni M Conti S Giordano and I Stojmenovic ldquoMobileAdHocNetworking Cutting Edge Directions Second EditionrdquoMobile Ad Hoc Networking Cutting Edge Directions SecondEdition 2013

[28] IEEE Standard for Information Technology-Telecommunica-tions and Information Exchange between System Wireless

Regional Area Networks (WRAN)-Specific Requirements-Part22 Cognitive Wireless RAN Medium Access Control (MAC)and Physical Layer (PHY) Specifications Policies and Proce-dures for Operation in the TV bands IEEE Standard 80222httpsstandardsieeeorgaboutget80280222html 2011

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of


International Journal of

RotatingMachinery


Hindawi Publishing Corporation httpwwwhindawicom

Journal of

Volume 201

Submit your manuscripts athttpswwwhindawicom

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics


Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks



sharing with the primary network the reception processis divided in time so that sensing and data receiving cantake place in different time slots while data transmissioncontinues over the entire time period that is using timedivision duplex (TDD) in reception

The advantage of transmit-sense mode approach is thatit allows continuous sensing which can in turn improvethe probability of detecting returning primary users Thisimprovement in the likelihood of detection is also a resultof more advanced learning algorithms that can be imple-mented in continuous sensing This is in contrast withdata transmit-sense-reception mode where sensing takesplace in short periods of time that is data reception andsensing are scheduled according to a time-division-duplexscheme Intermittent sensing in this mode does not allow theemployment of advanced and reliable learning algorithmswhich implies higher missed detection probability comparedto the transmit-sense mode On the other hand operationin transmit-sense-reception mode improves the secondaryusersrsquo throughput at the cost of deteriorating primary net-workrsquos performance

The authors in [12 13] have introduced adaptive full-duplex transmit-sense (FDs) and full-duplex transmit-receive-sense (FDr) to improve the performance of thespectrum-sharing mechanism These two works use energy-based sensing and waveform-based sensing respectivelyHowever both of these works have not considered cooper-ative spectrum sensing in order to further improve the spec-trum sensing accuracy On the other hand the effect on theprimary network can be alleviated by deploying cooperativesensing Cooperation among the secondary users in sensingthe licensed channel can improve the quality of detectingthe activity of primary users in the licensed spectrum Bycombining cooperative sensing and full-duplex communi-cation features full-duplex cooperative spectrum sensingmechanism is analyzed to improve the average throughputof SUs while guaranteeing PUsrsquo quality and throughputThissolution has been proposed in [14] only for time-slotted PUsand in [15] for cooperative acknowledgement The recentwork in [16] has analyzed SUsrsquo performance in non-time-slotted case without looking into the imposed effect on PUs

Motived by the above observation the contributions ofthis paper can be summarized as follows The throughputof both secondary and primary users is derived under FDrand FDs schemes as functions of residual self-interference(SI) and the number of cooperating users in spectrumsensing Furthermore the results are extended for differentprimary channel utilization Unlike our previous work in[17] in this paper we introduce the full-duplex cooperativemultichannel based FDr and FDs sensing schemes and alsofind the minimum number of cooperating secondary usersrequired in FDr scheme so that the achievable throughputfor the primary users is very close to that in the FDs schemeIn the literature the full-duplex cooperative multichannelscenarios have also been considered [8 18 19] Howeverthese works assume SUs perform the standard p-persistentcarrier sense multiple access (CSMA) mechanism for con-tention resolution on the selected channel Such distributedmechanism can effectively avoid the data crash and improve

TＩＨ TＩff

On state O state

Figure 1 On-off process channel model

the spectrum efficiency of the network On the other handby comparing with the cooperative multichannel allocationmechanism which will be considered in this paper thedistributed version has to be limited to a ldquosmallrdquo networkBased on the cooperative SUs design in this paper we alsopropose aMAC layer protocol design and analyze the averagethroughputs of SUs and PUs when utilizing TV white spaceas primary channels The results show that by increasingthe number of sensed channels the average throughput ofproposed FD schemesmaynot always be improved especiallyfor FDr scheme This is because the duration of sensing isconsidered as nontransmitting time which reduce the timefor data transmission

The rest of this paper is organized as follows The systemmodel and throughput analysis are presented in Section 2Spectrum sensing and self-interference effect are analyzedin Section 3 Formulation of secondary and primary averagethroughput for the cooperative full-duplex sensing scenario isin Sections 4 and 5 respectively The analysis for the averagethroughput optimization and its corresponding numericalresults are provided in Section 6 A newMACprotocol designis proposed and evaluated in Section 7 Finally conclusionsare drawn in Section 8

2 System Model

As discussed in abstract and introduction sections theasynchronous cognitive networks are defined that only theprimary and secondary networks are not synchronized butwithin the secondary network the secondary users will beassumed to be synchronized Moreover the primary usersin this case may or may not be synchronized In additionwe also assume that SUs always have packets to ldquoreceiverdquo as[9 20] and this assumption is meaningful if the number ofSUs is directly linked to the traffic demand in MAC protocoldesign

21 Primary Users One pair of PUs is considered to commu-nicate in half-duplex (HD) mode over 119882 spectrum bandswhere the bands are not overlapped and with the samespectrum utilization as 120573 PUsrsquo channel is modeled as an on-off process as [21] which is shown in Figure 1 The channel isconsidered in on state when it is occupied by PUs On theother hand off state means there is an available spectrumthat can be exploited by SUs The length of the on state (119879on)and off state (119879off ) follows the exponential distribution withaverages of 120588on and 120588off respectively In this case the primarychannel utilization 120573 can be calculated as

120573 = 120588on120588off + 120588on (1)


Primary user

Secondary user

Half-duplex



Tp Tp

Ts

Ts

Ts

Tt

Tt

Tt










119871 = 119882sumV=0

V sdot 119875V (V) (3)















119877 = 1119872120596119904119879119904 119872sum119898=1 120596119904119879119904sum119899=1 1003816100381610038161003816119903119898 [119899]10038161003816100381610038162 (7)



(8)





119875119891 = 119876((120576th1205902 minus 1)radic119872120596119904119879119904) (10)









which is given by




1198631199040 = log2 (1 + SNR1199041199041 + (1 minus 120581) SNR119904119904) (14)





120591119904HD = 11sum119894=00

119875 [119878119894HD] 119862119894HD (16)








120591119904FDr = 11sum119894=00

119875 [119878119894FDr] 119862119894FDr (21)


119875 [119878119894FDr] = 119875 [119878119894HD] forall119894 (22)



Tp

Ts Tt

On stateO state

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

Tp

Ts Tt


S10HD S11HD

(a)

S00FDr S01FDr

S10FDr S11FDr

Tp

Ts Tt

On stateO stateTp

Ts

On stateO state

Tt

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

(b)





11986211FDr = 2119875119889119879119901 minus 119879119904119879119901 1198631199041

(23)






120591119904FDs = 11sum119894=00

119875 [119878119894FDs] 119862119894FDs (27)



1198631199010 = log2 (1 + SNR119901119901) (28)


1198631199011 = log2 (1 + SNR1199011199011 + SNR119901119904) (29)



S00FDs

Ts


Ts


S11FDs




Ts

Ts Tt

Tt

On state Off state

Primary user



Dp1 Dp1 Dp0Dp0

ＩＨ

Tp



119875 [119878119895119901] = (119887119879119901119895 )119875119889119895119875(119887119879119901minus119895)119889 (30)

where

119887119879119901 = [120588on119879119901 ] (31)



Ts

On state Off state

Primary user


sense

Dp1 Dp1 Dp0Dp0

ＩＨ



(32)


120591119901HD = 120591119901FDr = 119887119879119901sum119895=0

119875 [119878119895119901] 119862119895119901 (33)


119875 [119878119897119901FDs] = (119887119879119904119897 ) 119875119889119897119875(119887119879119904minus119897)119889 (34)

where 119887119879119904 = [120588on119879119904 ] (35)





120591119901FDs = 119887119879119904sum119897=0

119875 [119878119897119901FDs] 119862119897119901FDs (37)







max119898V119879119904


SNR119909119910 le SNR119909119910 forall119909 119910(38)





HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

15

2

25

3

35

4

45

5

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)






035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095


90 guarantee



0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)


HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1








VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










Primary user

Secondary user

Half-duplex



Tp Tp

Ts

Ts

Ts

Tt

Tt

Tt










119871 = 119882sumV=0

V sdot 119875V (V) (3)















119877 = 1119872120596119904119879119904 119872sum119898=1 120596119904119879119904sum119899=1 1003816100381610038161003816119903119898 [119899]10038161003816100381610038162 (7)



(8)





119875119891 = 119876((120576th1205902 minus 1)radic119872120596119904119879119904) (10)









which is given by




1198631199040 = log2 (1 + SNR1199041199041 + (1 minus 120581) SNR119904119904) (14)





120591119904HD = 11sum119894=00

119875 [119878119894HD] 119862119894HD (16)








120591119904FDr = 11sum119894=00

119875 [119878119894FDr] 119862119894FDr (21)


119875 [119878119894FDr] = 119875 [119878119894HD] forall119894 (22)



Tp

Ts Tt

On stateO state

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

Tp

Ts Tt


S10HD S11HD

(a)

S00FDr S01FDr

S10FDr S11FDr

Tp

Ts Tt

On stateO stateTp

Ts

On stateO state

Tt

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

(b)





11986211FDr = 2119875119889119879119901 minus 119879119904119879119901 1198631199041

(23)






120591119904FDs = 11sum119894=00

119875 [119878119894FDs] 119862119894FDs (27)



1198631199010 = log2 (1 + SNR119901119901) (28)


1198631199011 = log2 (1 + SNR1199011199011 + SNR119901119904) (29)



S00FDs

Ts


Ts


S11FDs




Ts

Ts Tt

Tt

On state Off state

Primary user



Dp1 Dp1 Dp0Dp0

ＩＨ

Tp



119875 [119878119895119901] = (119887119879119901119895 )119875119889119895119875(119887119879119901minus119895)119889 (30)

where

119887119879119901 = [120588on119879119901 ] (31)



Ts

On state Off state

Primary user


sense

Dp1 Dp1 Dp0Dp0

ＩＨ



(32)


120591119901HD = 120591119901FDr = 119887119879119901sum119895=0

119875 [119878119895119901] 119862119895119901 (33)


119875 [119878119897119901FDs] = (119887119879119904119897 ) 119875119889119897119875(119887119879119904minus119897)119889 (34)

where 119887119879119904 = [120588on119879119904 ] (35)





120591119901FDs = 119887119879119904sum119897=0

119875 [119878119897119901FDs] 119862119897119901FDs (37)







max119898V119879119904


SNR119909119910 le SNR119909119910 forall119909 119910(38)





HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

15

2

25

3

35

4

45

5

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)






035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095


90 guarantee



0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)


HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1








VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation














119877 = 1119872120596119904119879119904 119872sum119898=1 120596119904119879119904sum119899=1 1003816100381610038161003816119903119898 [119899]10038161003816100381610038162 (7)



(8)





119875119891 = 119876((120576th1205902 minus 1)radic119872120596119904119879119904) (10)









which is given by




1198631199040 = log2 (1 + SNR1199041199041 + (1 minus 120581) SNR119904119904) (14)





120591119904HD = 11sum119894=00

119875 [119878119894HD] 119862119894HD (16)








120591119904FDr = 11sum119894=00

119875 [119878119894FDr] 119862119894FDr (21)


119875 [119878119894FDr] = 119875 [119878119894HD] forall119894 (22)



Tp

Ts Tt

On stateO state

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

Tp

Ts Tt


S10HD S11HD

(a)

S00FDr S01FDr

S10FDr S11FDr

Tp

Ts Tt

On stateO stateTp

Ts

On stateO state

Tt

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

(b)





11986211FDr = 2119875119889119879119901 minus 119879119904119879119901 1198631199041

(23)






120591119904FDs = 11sum119894=00

119875 [119878119894FDs] 119862119894FDs (27)



1198631199010 = log2 (1 + SNR119901119901) (28)


1198631199011 = log2 (1 + SNR1199011199011 + SNR119901119904) (29)



S00FDs

Ts


Ts


S11FDs




Ts

Ts Tt

Tt

On state Off state

Primary user



Dp1 Dp1 Dp0Dp0

ＩＨ

Tp



119875 [119878119895119901] = (119887119879119901119895 )119875119889119895119875(119887119879119901minus119895)119889 (30)

where

119887119879119901 = [120588on119879119901 ] (31)



Ts

On state Off state

Primary user


sense

Dp1 Dp1 Dp0Dp0

ＩＨ



(32)


120591119901HD = 120591119901FDr = 119887119879119901sum119895=0

119875 [119878119895119901] 119862119895119901 (33)


119875 [119878119897119901FDs] = (119887119879119904119897 ) 119875119889119897119875(119887119879119904minus119897)119889 (34)

where 119887119879119904 = [120588on119879119904 ] (35)





120591119901FDs = 119887119879119904sum119897=0

119875 [119878119897119901FDs] 119862119897119901FDs (37)







max119898V119879119904


SNR119909119910 le SNR119909119910 forall119909 119910(38)





HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

15

2

25

3

35

4

45

5

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)






035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095


90 guarantee



0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)


HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1








VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











which is given by




1198631199040 = log2 (1 + SNR1199041199041 + (1 minus 120581) SNR119904119904) (14)





120591119904HD = 11sum119894=00

119875 [119878119894HD] 119862119894HD (16)








120591119904FDr = 11sum119894=00

119875 [119878119894FDr] 119862119894FDr (21)


119875 [119878119894FDr] = 119875 [119878119894HD] forall119894 (22)



Tp

Ts Tt

On stateO state

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

Tp

Ts Tt


S10HD S11HD

(a)

S00FDr S01FDr

S10FDr S11FDr

Tp

Ts Tt

On stateO stateTp

Ts

On stateO state

Tt

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

(b)





11986211FDr = 2119875119889119879119901 minus 119879119904119879119901 1198631199041

(23)






120591119904FDs = 11sum119894=00

119875 [119878119894FDs] 119862119894FDs (27)



1198631199010 = log2 (1 + SNR119901119901) (28)


1198631199011 = log2 (1 + SNR1199011199011 + SNR119901119904) (29)



S00FDs

Ts


Ts


S11FDs




Ts

Ts Tt

Tt

On state Off state

Primary user



Dp1 Dp1 Dp0Dp0

ＩＨ

Tp



119875 [119878119895119901] = (119887119879119901119895 )119875119889119895119875(119887119879119901minus119895)119889 (30)

where

119887119879119901 = [120588on119879119901 ] (31)



Ts

On state Off state

Primary user


sense

Dp1 Dp1 Dp0Dp0

ＩＨ



(32)


120591119901HD = 120591119901FDr = 119887119879119901sum119895=0

119875 [119878119895119901] 119862119895119901 (33)


119875 [119878119897119901FDs] = (119887119879119904119897 ) 119875119889119897119875(119887119879119904minus119897)119889 (34)

where 119887119879119904 = [120588on119879119904 ] (35)





120591119901FDs = 119887119879119904sum119897=0

119875 [119878119897119901FDs] 119862119897119901FDs (37)







max119898V119879119904


SNR119909119910 le SNR119909119910 forall119909 119910(38)





HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

15

2

25

3

35

4

45

5

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)






035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095


90 guarantee



0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)


HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1








VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










Tp

Ts Tt

On stateO state

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

Tp

Ts Tt


S10HD S11HD

(a)

S00FDr S01FDr

S10FDr S11FDr

Tp

Ts Tt

On stateO stateTp

Ts

On stateO state

Tt

Tp

Ts Tt

On stateO state

Tp

Ts Tt

On state O state

(b)





11986211FDr = 2119875119889119879119901 minus 119879119904119879119901 1198631199041

(23)






120591119904FDs = 11sum119894=00

119875 [119878119894FDs] 119862119894FDs (27)



1198631199010 = log2 (1 + SNR119901119901) (28)


1198631199011 = log2 (1 + SNR1199011199011 + SNR119901119904) (29)



S00FDs

Ts


Ts


S11FDs




Ts

Ts Tt

Tt

On state Off state

Primary user



Dp1 Dp1 Dp0Dp0

ＩＨ

Tp



119875 [119878119895119901] = (119887119879119901119895 )119875119889119895119875(119887119879119901minus119895)119889 (30)

where

119887119879119901 = [120588on119879119901 ] (31)



Ts

On state Off state

Primary user


sense

Dp1 Dp1 Dp0Dp0

ＩＨ



(32)


120591119901HD = 120591119901FDr = 119887119879119901sum119895=0

119875 [119878119895119901] 119862119895119901 (33)


119875 [119878119897119901FDs] = (119887119879119904119897 ) 119875119889119897119875(119887119879119904minus119897)119889 (34)

where 119887119879119904 = [120588on119879119904 ] (35)





120591119901FDs = 119887119879119904sum119897=0

119875 [119878119897119901FDs] 119862119897119901FDs (37)







max119898V119879119904


SNR119909119910 le SNR119909119910 forall119909 119910(38)





HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

15

2

25

3

35

4

45

5

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)






035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095


90 guarantee



0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)


HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1








VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










S00FDs

Ts


Ts


S11FDs




Ts

Ts Tt

Tt

On state Off state

Primary user



Dp1 Dp1 Dp0Dp0

ＩＨ

Tp



119875 [119878119895119901] = (119887119879119901119895 )119875119889119895119875(119887119879119901minus119895)119889 (30)

where

119887119879119901 = [120588on119879119901 ] (31)



Ts

On state Off state

Primary user


sense

Dp1 Dp1 Dp0Dp0

ＩＨ



(32)


120591119901HD = 120591119901FDr = 119887119879119901sum119895=0

119875 [119878119895119901] 119862119895119901 (33)


119875 [119878119897119901FDs] = (119887119879119904119897 ) 119875119889119897119875(119887119879119904minus119897)119889 (34)

where 119887119879119904 = [120588on119879119904 ] (35)





120591119901FDs = 119887119879119904sum119897=0

119875 [119878119897119901FDs] 119862119897119901FDs (37)







max119898V119879119904


SNR119909119910 le SNR119909119910 forall119909 119910(38)





HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

15

2

25

3

35

4

45

5

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)






035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095


90 guarantee



0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)


HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1








VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation















max119898V119879119904


SNR119909119910 le SNR119909119910 forall119909 119910(38)





HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095

15

2

25

3

35

4

45

5

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)






035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095


90 guarantee



0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)


HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1








VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











035

04

045

05

055

06

065

07

Prim

ary

user

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)

HDFDs k = 1

FDs k = 099

FDs k = 095

FDr k = 1

FDr k = 099

FDr k = 095


90 guarantee



0

05

1

15

2

25

3

35

4

Seco

ndar

y us

errsquos

aver

age t

hrou

ghpu

t (bi

tsse

cH

z)


HD V = 3

HD V = 2

HD V = 1

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 3

FDr V = 2

FDr V = 1








VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











VehiclesSUs node

Cooperation
















where




SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










SP RCP BP DTP

SP RCP BP DTP

1

1



Mx

Mx

BC

BC

Data transmission

Minislot (Tms)

CCC

Ch 1

Ch W

FDr

sensing

Sensing

Tp

Tp


Occupied

Idle

FDs

Sensing





+ 2119879119901 (1198751198891198631199041119879119901 minus 1198751198911198631199040119879prFDr)11986210MFDr











Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










Ts Tr Tb

Sensing 1

1



Mx

Mx

BC

BC

V channels

FDs frame

FDr frame

Tp

Tr Tb

Tt

Tt



TＭＩ

TＪＬ

TＪＬ

Sensing






0

05

1

15

2

25

Prop

osed

MAC

rsquos av

erag

e thr

ough

put (

bits

sec

Hz)


FDs V = 5

FDs V = 4

FDs V = 3

FDs V = 2

FDs V = 1

FDr V = 5

FDr V = 4

FDr V = 3

FDr V = 2

FDr V = 1






8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation













8 Conclusion




Acknowledgments


References
































RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation


























RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation









RoboticsJournal of





Journal of



RotatingMachinery



Journal of

Volume 201


VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation









Date post:	09-Jul-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Cooperative Full-Duplex Physical and MAC Layer Design in...

Documents