+ All Categories
Home > Documents > WLANs - A Lesson 2015

WLANs - A Lesson 2015

Date post: 04-Jan-2016
Category:
Upload: omaquina
View: 33 times
Download: 1 times
Share this document with a friend
Description:
Wlan de largo alcance, nuevo protocolo IEEE 802.11ah
Popular Tags:
15
IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015 1761 Outdoor Long-Range WLANs: A Lesson for IEEE 802.11ah Stefan Aust, Member, IEEE, R. Venkatesha Prasad, Senior Member, IEEE, and Ignas G. M. M. Niemegeers Abstract—Several service applications have been reported by many who proposed the use of wireless LANs (WLANs) over a wide variety of outdoor deployments. In particular, the upcoming IEEE 802.11ah WLAN protocol will enable a longer transmission range between WLAN access points (APs) and stations (STAs) up to multiple kilometers using carrier frequencies at 900 MHz. How- ever, limitations of WLAN outdoor installations have been found of the plethora of WLAN protocols in experimental studies. This article summarizes the challenges and provides a comprehensive overview of suggested improvements. As the standardization of the IEEE 802.11ah is reaching its final stage, important protocol aspects as well as new features are to be outlined. Interference problems and issues with the WLAN configuration, the physical layer (PHY), and media access control (MAC) are of paramount importance in outdoor WLAN networks, and thus, are discussed in detail. Further, we examine the reported upper boundaries in throughput and link reliability of long-range WLANs in different environments, including sea-surfaces, unmanned aerial vehicles (UAVs), and tunnels. At the end of this study, we reflect on the major issues regarding sub-1 GHz (S1G) WLANs and propose avenues for further research. Index Terms—WLAN, IEEE 802.11ah, wireless sensor, sub-1 GHz, long-range, outdoor. I. I NTRODUCTION W IRELESS communication in the license-exempt radio- bands has become an attractive communication method because of its low cost and widespread access. It is intuitive to adapt the wireless LAN (WLAN) technology for remote sensing services as an alternative access method, e.g., to connect meter devices in outdoor utility infrastructures [1]. Other services include health care and non-intrusive remote sensing in outdoor activities. The upcoming IEEE 802.11ah WLAN protocol amendment is a potential candidate to enable outdoor long-range WLANs, which operate at the sub-1 GHz (S1G) industrial, scientific and medical (ISM) radio-band [2]. Using WLANs in outdoor locations is from vital impor- tance, but fundamental limitations of the usability of legacy WLANs have been reported in the scientific literature. First and foremost, common WLAN path loss models have been Manuscript received July 8, 2014; revised January 28, 2015; accepted April 5, 2015. Date of publication May 4, 2015; date of current version August 20, 2015. S. Aust is with NEC Communication Systems, Ltd., Kawasaki 211-8666, Japan (e-mail: [email protected]). R. V. Prasad and I. G. M. M. Niemegeers are with the Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS), Delft University of Technology, 2600 GA Delft, The Netherlands (e-mail: [email protected]; [email protected]). Digital Object Identifier 10.1109/COMST.2015.2429311 solely examined for indoor environments to enable ultra-high speed Internet access at large channel bandwidth. In contrast, outdoor path loss characterization has received less attention in WLANs [3]–[6]. The use of a WLAN in outdoor en- vironments is heavily affected by the surrounding WLANs and ambient noise; thus, interference has a major impact on the transmission performance. In this article, a special focus is made on the IEEE 802.11ah standard, which envisions a long-range WLAN network at 900 MHz serving high den- sity WLAN stations (WLAN STAs) [7]. Further, the antenna setup in outdoor WLANs is of paramount importance. Omni- directional and directional beam antennae behave differently and therefore require different physical layer (PHY) and media access control (MAC) operations. Indoor path loss models are no longer valid and PHY operations must be modified. Additionally, the required MAC enhancements must consider network densification. It is imperative that network designers require solid information about potential threats in upcom- ing IEEE 802.11ah WLAN deployments, including coverage range, throughput boundaries, and channel characteristics. Similar surveys highlighted on limited problems in legacy WLANs only, such as radio propagation in tunnels [8], through- put capacity [9], MAC design in WLAN ad-hoc networks [10], and coexistence among IEEE 802.11 and IEEE 802.15 wireless networks [11]. In contrast, this survey contributes a holistic view on the reported challenges on outdoor long-range WLANs in the scientific literature over the period 2002 to 2014. This article presents an overview of the experimental and theoretical deployment of long-range WLANs. In particular, the survey is focused on the upcoming standard 802.11ah and how the new WLAN protocol amendment must tackle some of those challenges, e.g., at different layers in all existing WLAN standards. Then, the authors explain the challenges of long- range WLAN, which is followed by long-range WLANs in exposed areas such as sea-surfaces, unmanned aerial vehicles (UAVs), and tunnels. At the end, further research topics were identified. This article is organized as follows: The IEEE 802.11ah WLAN protocol is introduced in Section II. Use case models of long-range WLANs are presented in Section III. Outdoor deployment challenges of long-range WLANs are presented in Section IV. Long-range WLANs in exposed areas are of interest in Section V; in particular, the focus is on long-range WLANs over the sea-surface and in 3D-networks. The lessons learned are discussed in Section VI, which is followed by a list of future research challenges in Section VII. Finally, concluding remarks are given in Section VIII. 1553-877X © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Transcript
Page 1: WLANs - A Lesson 2015

IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015 1761

Outdoor Long-Range WLANs: A Lessonfor IEEE 802.11ah

Stefan Aust, Member, IEEE, R. Venkatesha Prasad, Senior Member, IEEE, and Ignas G. M. M. Niemegeers

Abstract—Several service applications have been reported bymany who proposed the use of wireless LANs (WLANs) over awide variety of outdoor deployments. In particular, the upcomingIEEE 802.11ah WLAN protocol will enable a longer transmissionrange between WLAN access points (APs) and stations (STAs) upto multiple kilometers using carrier frequencies at 900 MHz. How-ever, limitations of WLAN outdoor installations have been foundof the plethora of WLAN protocols in experimental studies. Thisarticle summarizes the challenges and provides a comprehensiveoverview of suggested improvements. As the standardization ofthe IEEE 802.11ah is reaching its final stage, important protocolaspects as well as new features are to be outlined. Interferenceproblems and issues with the WLAN configuration, the physicallayer (PHY), and media access control (MAC) are of paramountimportance in outdoor WLAN networks, and thus, are discussedin detail. Further, we examine the reported upper boundaries inthroughput and link reliability of long-range WLANs in differentenvironments, including sea-surfaces, unmanned aerial vehicles(UAVs), and tunnels. At the end of this study, we reflect on themajor issues regarding sub-1 GHz (S1G) WLANs and proposeavenues for further research.

Index Terms—WLAN, IEEE 802.11ah, wireless sensor,sub-1 GHz, long-range, outdoor.

I. INTRODUCTION

W IRELESS communication in the license-exempt radio-bands has become an attractive communication method

because of its low cost and widespread access. It is intuitiveto adapt the wireless LAN (WLAN) technology for remotesensing services as an alternative access method, e.g., toconnect meter devices in outdoor utility infrastructures [1].Other services include health care and non-intrusive remotesensing in outdoor activities. The upcoming IEEE 802.11ahWLAN protocol amendment is a potential candidate to enableoutdoor long-range WLANs, which operate at the sub-1 GHz(S1G) industrial, scientific and medical (ISM) radio-band [2].Using WLANs in outdoor locations is from vital impor-tance, but fundamental limitations of the usability of legacyWLANs have been reported in the scientific literature. Firstand foremost, common WLAN path loss models have been

Manuscript received July 8, 2014; revised January 28, 2015; acceptedApril 5, 2015. Date of publication May 4, 2015; date of current versionAugust 20, 2015.

S. Aust is with NEC Communication Systems, Ltd., Kawasaki 211-8666,Japan (e-mail: [email protected]).

R. V. Prasad and I. G. M. M. Niemegeers are with the Faculty of ElectricalEngineering, Mathematics and Computer Science (EEMCS), Delft Universityof Technology, 2600 GA Delft, The Netherlands (e-mail: [email protected];[email protected]).

Digital Object Identifier 10.1109/COMST.2015.2429311

solely examined for indoor environments to enable ultra-highspeed Internet access at large channel bandwidth. In contrast,outdoor path loss characterization has received less attentionin WLANs [3]–[6]. The use of a WLAN in outdoor en-vironments is heavily affected by the surrounding WLANsand ambient noise; thus, interference has a major impact onthe transmission performance. In this article, a special focusis made on the IEEE 802.11ah standard, which envisions along-range WLAN network at 900 MHz serving high den-sity WLAN stations (WLAN STAs) [7]. Further, the antennasetup in outdoor WLANs is of paramount importance. Omni-directional and directional beam antennae behave differentlyand therefore require different physical layer (PHY) and mediaaccess control (MAC) operations. Indoor path loss modelsare no longer valid and PHY operations must be modified.Additionally, the required MAC enhancements must considernetwork densification. It is imperative that network designersrequire solid information about potential threats in upcom-ing IEEE 802.11ah WLAN deployments, including coveragerange, throughput boundaries, and channel characteristics.

Similar surveys highlighted on limited problems in legacyWLANs only, such as radio propagation in tunnels [8], through-put capacity [9], MAC design in WLAN ad-hoc networks[10], and coexistence among IEEE 802.11 and IEEE 802.15wireless networks [11]. In contrast, this survey contributes aholistic view on the reported challenges on outdoor long-rangeWLANs in the scientific literature over the period 2002 to2014. This article presents an overview of the experimental andtheoretical deployment of long-range WLANs. In particular,the survey is focused on the upcoming standard 802.11ah andhow the new WLAN protocol amendment must tackle some ofthose challenges, e.g., at different layers in all existing WLANstandards. Then, the authors explain the challenges of long-range WLAN, which is followed by long-range WLANs inexposed areas such as sea-surfaces, unmanned aerial vehicles(UAVs), and tunnels. At the end, further research topics wereidentified.

This article is organized as follows: The IEEE 802.11ahWLAN protocol is introduced in Section II. Use case modelsof long-range WLANs are presented in Section III. Outdoordeployment challenges of long-range WLANs are presented inSection IV. Long-range WLANs in exposed areas are of interestin Section V; in particular, the focus is on long-range WLANsover the sea-surface and in 3D-networks. The lessons learnedare discussed in Section VI, which is followed by a list of futureresearch challenges in Section VII. Finally, concluding remarksare given in Section VIII.

1553-877X © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

Page 2: WLANs - A Lesson 2015

1762 IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015

TABLE IIEEE 802.11ah TARGET SPECTRUM [2]

TABLE IICOMPARISON OF IEEE 802.11 AND 802.15 STANDARDS

II. THE IEEE 802.11ah WLAN PROTOCOL

The IEEE 802.11ah amendment aims to define an extendedorthogonal frequency division multiplexing (OFDM) PHY op-eration that allows for the use of the sub-1 GHz license-exemptISM-bands, which are available for radio transmission in someareas, including China, Europe, Japan, Singapore, South Korea,and the USA [12], [13]. Table I lists the IEEE 802.11ah targetfrequency spectrum and the maximum emitted radiated power(e.r.p.) and channel bandwidth (CBW) [2]. The values aredefined within the spectrum usage policies by each country.

Table II compares relevant IEEE 802.11 and IEEE 802.15standards. The table lists IEEE 802.11a, IEEE 802.11g, IEEE802.11ac, IEEE 802.11ah, and IEEE 802.15.4g as wireless per-sonal area network (WPAN). IEEE 802.15.4 does not supportmultiplexing schemes, whereas WLANs support OFDM. IEEE802.11ah provides the longest wireless access range with 1 kmradius in outdoor environments.

There have been over 10 proposed use cases for the IEEE802.11ah protocol during the standardization phase, some ofwhich pose challenges to the WLAN architecture [14]–[19]:

1) Outdoor applications over long-range communicationlinks up to 1 km;

2) Design of channel models that include both indoor andoutdoor modes;

3) The number of WLAN STAs, which is in the range of 10 sto 1000 s, which affects WLAN association, security, andthe channel access process;

4) A data rate for video streaming with 100 times largertraffic volume than metering applications;

5) The energy consumption requirement for sensors, whichis somewhat different from powered WLAN STAs;

6) Increased interference in outdoor deployments.

Fig. 1. Illustration of throughput performance versus distance, comparedfor IEEE 802.11n (Nss = 4) [4] and IEEE 802.11ah (Nss = 2) in a ruralenvironment.

A final compromise led to the agreement of three IEEE802.11ah usage models, namely, (i) sensor networks, (ii) sensorbackhaul, and (iii) WLAN coverage extension and off-loading[20]. Sensor networks represent the use of IEEE 802.11ahWLAN in areas such as metering, hospital, and environmentalmeasurements. The number of nodes varies from 10 to 100.Sensor backhaul represents the data collection and aggregationof multiple IEEE 802.11ah nodes which may vary from 10 to1000 nodes. High data rates are required for sensor backhaulconfigurations up to several Mbps. Finally, WLAN coverageextension and off-loading allow that certain amount of data traf-fic can be redirected from cellular to WLANs, e.g., in hot-spotareas. This was proposed for U.S. where up to 16 MHz channelbandwidth is available to allow high data rates. IEEE 802.11ahWLAN data rates must be optimized regarding the rate versusrange performance in a given band [17]. In particular, thisincludes transmission range ≤ 1 km and data rates ≥ 100 kpbs.The indoor coverage of sub-1 GHz sensor nodes operating at900 MHz have been analyzed in [21], indicating a 6–9 dB gainin indoor environments compared to 2.4 GHz WLANs. Thehigher power gain results in wider coverage with same tx-powerlevel. An example of the coverage range of IEEE 802.11ahcompared with IEEE 802.11n [4] is illustrated in Fig. 1. Thefigure depicts the simulated transmission performance with animplemented PHY simulator (1470 B packet size, UDP traffic,MCS 10).

A. The IEEE 802.11ah PHY

IEEE 802.11ah utilizes OFDM to achieve high data rates.The rationale to standardize OFDM in IEEE 802.11ah is tobe consistent with the IEEE 802.11ac PHY/MAC amendment,which allows a simple down-clocking of OFDM features tosmaller channel bandwidths [5]. Data are transmitted usingorthogonal sub-carriers or tones, in which data are equally dis-tributed over the sub-carriers. Each is affected by the frequencyselective behavior—due to the multipath propagation—of thewireless channel, which may degrade the received signal qual-ity. The total number of sub-carriers for 1 MHz channel band-width operation is 26, which consists of 2 pilot sub-carriers(tones) and 24 data sub-carriers per OFDM symbol. For higherchannel bandwidth the number of data and pilot tones (fixed,traveled) increases (484 tones for 16 MHz operation, with

Page 3: WLANs - A Lesson 2015

AUST et al.: OUTDOOR LONG-RANGE WLANs A LESSON FOR IEEE 802.11ah 1763

Fig. 2. Illustration of throughput performance of IEEE 802.11ah MCS rates(with short/long guard interval (GI)), 1 MHz channel bandwidth.

16 pilot and 468 data tones) [2]. The tone spacing �F is at31.25 kHz. The guard interval (GI) duration TGI is 8 μs and4 μs for short GI, respectively. The OFDM symbol durationTSYML with long GI is 40 μs, and 36 μs for short GI, respec-tively. The short training field (STF) duration TSTF and thelong training field (LTF) duration TLTF are at 160 μs whichhave been selected for robustness against multipath fading overlong distances (omni-portion of 1 MHz frame format). Theallowed error vector magnitude (EVM) is (−4) dB for BPSKand (−32) dB for 256-QAM. Minimum receiver sensitivity is(−98) dBm for BPSK and (−72) dBm for 256-QAM, respec-tively. Fig. 2 illustrates the IEEE 802.11ah PHY performancefor different modulation and coding schemes (MCS) rates at1 MHz bandwidth (number of spatial streams Nss = 1, 2). TheIEEE 802.11ah path loss models and the link budget are essen-tial to judge the performance in different outdoor environments[16]. The IEEE 802.11ah path loss PL(d) is classified into:

Macro deployment:

PL(d) = 8 + 37.6 log 10(d). (1)

Pico/Hot zone deployment:

PL(d) = 23.3 + 37.6 log 10(d), (2)

where d is the distance from the AP to the sensor node in[m], and the carrier frequency fc [Hz] is assumed to be at 900MHz. A correction factor of 21 log 10(f /900) must be added ifanother frequency is used. A short link STA-to-AP setup canbe classified as flat-fading (flat fading channel) due to the shortdelay spread and narrow bandwidth [22]. In [23], the repetitionscheme is introduced to increase the coverage for S1G WLANs.The repetition scheme is a simple method that can achieve upto 3 dB performance gain [24]. The IEEE 802.11ah repetitionscheme is introduced to provide a robust wireless link over1km with guaranteed 150 kbps data rate. Improvements of PHYtransmission efficiency have been focused on minimizing thecomplexity cyclic redundancy check (CRC) codes to reduce thenumber of transmitted parity bits [25], [26]. Other IEEE 802.11standards which consider outdoor channels are IEEE 802.11p

for wireless access in vehicular environments at 5 GHz [39]and IEEE 802.11ac on enhancements for very high throughputat 6 GHz [6].

B. The IEEE 802.11ah MAC

The IEEE 802.11ah MAC supports new PHY parameters,particularly for long-range communication. Additionally, therehave been suggestions to further increase the power efficiencyand high density access. Improved offset listen intervals forsmart grid communication [27], [28], optimized sleep modes[29], and grouping strategies for dense IEEE 802.11ah net-works [30]–[32] have been in the research focus recently.

Compared to sensor nodes, an IEEE 802.11ah standardizedsub-1 GHz WLAN will offer a higher degree of scalability(6000 nodes), improved energy efficiency, and multiple-inputmultiple-output (MIMO) OFDM features [33]. For the caseswhen the number of STAs is large, the maximum number ofSTAs that can be supported by a single WLAN AP limitsthe number of STAs, which is 2007 STAs [3]. Additionally,interference plays an important role in the sub-1 GHz ISM-band. The presence of current RFID devices, ZigBee™ andIEEE 802.15.4 devices in the 900 MHz radio-band will leadto performance limitations of wireless transmissions when thesub-1 GHz WLAN will emerge.

The IEEE 802.11ah MAC has to face these new networkchallenges. It is argued in [35] that high density WLANsof up to 6000 nodes can be covered within 1 km coverageradius. In addition, a number of 10,000 sensors/actuators canbe expected in large outdoor areas for a deployment of a typicalindustrial process automation [36]. Thus, enhanced MAC ad-dressing schemes are defined for high density deployments.Further enhancement in addressing could be done using auto-configuration [36]. In particular, the new restricted accesswindow (RAW) function utilizes a beaconing interval that isdivided into a contention-free and contention-based accessperiod. The contention-free period is further divided into pre-defined time-slots for the group-based sensor access [37], [38].The IEEE 802.11ah MAC must provide mechanisms that en-able the coexistence with other systems in the same license-exempt radio band, which include the IEEE 802.15.4 protocols[11], [17]. It is expected that wireless nodes must operatewithout battery replacement for 5 to 10 years [24]. Thus, sleepalgorithm are proposed consisting of an energy-aware sleepingalgorithm and a high priority algorithm, which is designed formachine-to-machine (M2M) networks. The algorithm favorslow energy devices and provides such devices higher priorityfor channel access [29].

III. USE CASES OF LONG-RANGE OUTDOOR WLANs

We summarize on reported WLAN usage scenarios, whichwe classified into health care and environmental sensing.

A. Health Care and Outdoor Activities

Work related to health care is reported by Abuali et al. [40],in which a long-range WLAN system is utilized. A single AP

Page 4: WLANs - A Lesson 2015

1764 IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015

for one-hop coverage of 3–4 km collects sensor data fromZigBee™ and RFID devices as part of a remote well-beingmonitoring (RWM) system. The proposed system connects upto 60 sensor nodes and 75 active RFID tags. IEEE 802.11g/n isused as the WLAN. The sending power varies from 0 to 4 W.A huge potential for the application of long-range WLANs isin the monitoring of bio-mechanics of athletes during training.The simple use—most athletes and athletic facilities already usewireless monitoring systems and WLAN access—of WLANbased wide-area sensor networks enables a huge potential mar-ket for performance assessment applications. Real-time tracingof body conditions in live-events and on-line broadcastingare envisioned applications. The non-intrusive collection ofsuch data could be beneficial in maximizing the performanceof athletes during sporting events and to prevent them fromsuffering severe injuries. The monitored parameters include thepulse rate and oxygen saturation. In addition, referee-assistedservices could benefit from an outdoor WLAN. Sivaraman et al.[41] stated that the major challenge in extracting physiologicaldata in real-time is the limited radio range in sparse and highlydynamic environments. Further, it was stated that the dynamicsof certain sports environments are unknown, such as in soccerfields, which present a large outdoor area of > 4000 m2. Suchknowledge about the environmental dynamics is essential todesign communication systems and protocols.

Wireless multi-hop communication was proposed to over-come the limited transmission rage of body-worn devices andhuman body attenuation. Interestingly, the authors in [41] se-lected a slotted media access approach to mitigate data framecollisions, in which each single body-worn device is assigned toone time-slot. A poor wireless connection was reported, basedon the measurements in the soccer field, due to the exposedphysical location of some individuals. A novel mobility modelwas developed, which includes the number of players, base-stations and wireless links. López-Matencio et al. [42] reportedon the environmental sensing application and its impact on theselection from a variety of tracks (hardness, temperature, andmoisture). Llosa et al. [43] reported the challenges of remotemonitoring of the rowing performance of athletes during train-ing cycles. Miniaturized motion sensors (accelerometers) wereused, which communicate wirelessly in real-time to enable non-invasive measurement of rigid body motions. A WLAN APcollected the monitored data of both the boat and the crewmembers.

B. Metering

The collection of meter data in smart grids is an importantapplication of long-range WLANs. Smart grid wireless systemsin the license-exempt bands are discussed in the IEEE 802 com-munity [44]. IEEE 802.11ah is a candidate to connect indoormeter devices and gateways to outdoor IEEE 802.11ah APsmounted on poles. The sensor motes were equipped with exter-nal panel antennae to mitigate the attenuation of natural obsta-cles and to increase the communication range. The control andreading of residential utility meters can be performed remotely,e.g., via a home gateway that sends the meter data to the utilitycompany. Such so-called automatic meter reading (AMR) can

be build up on a long-range and low-energy sub-1 GHz sensornetwork that operates at 900 MHz. Gas, electric and watermeters can be read remotely, thereby avoiding the monthlyreading of such meters by personnel from the utility providers[45]. Sensing can be executed as short-burst data transmissionsand covers smart metering applications, such as monitoring gas,water and energy consumption [46]. Alternatively, a WLANdevice from Atheros can be used to study some of the per-formance issues. This device is based on the IEEE 802.11gprotocol and uses proprietary channel bandwidths of 5 MHzto 20 MHz. The problem is that IEEE 802.11ah WLAN STAsmay be installed in meter boxes, i.e., behind walls and metalshields [24], [47].

C. Environmental Sensing

In [48], recent regional incidents, such as the Great EastJapan Earthquake and the resulting tsunami, are mentioned tomotivate the need for a dependable network infrastructure. Inaddition, wireless networks for smart communities are outlinedin [48], in which standardized wireless systems, including IEEE802.15.4g and IEEE 802.11ah, are considered. The communi-cation range was reported to be 150 m. Notably, the authorcalled for an energy-efficient WSN due to the unattendednode distribution in the outdoor environment [42]. Long-rangeWLANs are applicable for outdoor sensing applications of pets[49], [50]. In addition to longer coverage, some applicationsrequire energy efficient communication, which becomes animportant aspect in WLANs [51], [52].

IV. CHALLENGES IN LONG-RANGE WLANs

Next, we report on general findings in long-range WLANs,followed by PHY and MAC challenges. Reported problems oncoverage and energy usage complement our findings.

A. General Findings

Early studies on outdoor WLAN deployments by Agoayo et al.reported on findings in Roofnet [53], [54]. Their observationwas that packet losses are mainly caused by multipathinterference, in which inter-symbol interference (ISI), causedby reflection and delay—measured at > 500 ns—of the signallead to packet collisions at the receiver. In addition, they foundthat the SNR does not exhibit a significant correlation with thepacket delivery ratio.

Sayed et al. conducted wireless long-range measurementson an interesting test site, i.e., in the (flat) desert, which was160 km away from the city; thus, there was no effect of WLANinterference on the wireless link performance in their study[55]. In addition, they developed a theoretical model to validatethe findings, based on the Hata/Okumura path loss model, forcarrier frequency fc = 2.4 GHz. A 50 MB file was transferredin each measurement. They found that data files < 10 MB couldtravel up to 9 km line-of-sight (LOS). Files larger than 10 MBcould only reach up to 7 km. Interestingly, they were able todemonstrate a linear increase of the round trip time (RTT) of350 ms at 5 km distance. Although, it was not discussed in [55],

Page 5: WLANs - A Lesson 2015

AUST et al.: OUTDOOR LONG-RANGE WLANs A LESSON FOR IEEE 802.11ah 1765

their findings indicated a different initial attenuation comparedto that of the Hata/Okumura model—Sayed reported that themeasured values follow the modeled path loss. However, arough estimate in [55] proposed 110 dB for the initial measuredpath loss, compared to 115 dB from the model, which indicatesa significant difference of 5 dB at 1 km that continues overa distance of up to 7 km. A higher path loss discrepancycan be observed between 2–5 km with 10 dB. The reasonsfor this discrepancy are not clear, but we speculate that overlarge distances, the orientation of the antenna beam becomesan important factor.

Afanasyev et al. reported their findings in the freely availableoutdoor wireless Internet service “Google Wi-Fi Network”,which consists of 500 pole-top WLAN Mesh APs in MountainView, CA [56]. Whereas other reports focused on mesh proto-cols, their study focused on how users could benefit from urbanWLAN networks. Their findings include the traffic pattern ofarchetypal use cases and the different access technologies. Inparticular, they refer to the exponential increase of smart phoneusers; they found > 15.000 associated smart phones that useWLAN access within their network.

Chebrolu et al. reported that there is no direct correlationbetween the link distance and the error rate [58]. Weather con-ditions had little or no impact on the transmission performanceand were quantized to 1–2 dB changes in the link attenuation forheavy rain or fog, compared to good weather conditions (whicheven exhibited similar variations in attenuation). However, asignificant correlation of transmission failures (WLAN controlframes) and outdoor temperature was reported with a corre-lation coefficient of 0.36 in [59], suggesting that under goodweather conditions the use of WLAN increases; thus, moredata traffic occurs. The selected default packet size was 1400 Bfor each measurement with 11 Mbps tx-rate. Chebrolu et al.proposed that a simple and robust rate adaptation scheme wouldbe beneficial, based on the SNR characteristics. Surprisingly,this result is different than the findings in [53] and [60], whichsuggests that for the plethora of WLAN protocols used as long-range WLANs, different adaptation schemes are required. Notethat Chebrolu called for countermeasures against interferenceto identify the “RF pollution” inside a WLAN and amongWLAN deployments. They concluded that there should be alegal or semi-legal mechanism to detect, diagnose, and controlmutual interference [58]. Finally, they provided a list of howto avoid measurement and WLAN setup mistakes, such as thedetection of the presence of WLAN interference prior/duringthe measurements, unwanted association of terminals to thesurroundings, unknown APs, buffer overflows due to fast packetinter-arrival times, and RF leakage in the near-field.

The throughput performance decreases with the increase ofthe coverage range. In [57] a linear regression of measuredRSSI value is presented dropping from −60 dBm down to−80 dBm over a range of 300 m during UAV flyovers.The default packet size was 200 B for each measurement.Chebrolu et al. curiously concluded that their findings forIEEE 802.11b would be applicable to IEEE 802.11g and IEEE802.11a, which was disproved by Bianchi et al. in which IEEE802.11g was not found to be an appropriate WLAN protocolfor outdoor deployments [60]. The short guard interval (GI) ofIEEE 802.11g does not reflect the impulse response character-istic of outdoor multipath effects.

Trinchero et al. [61] used their modified WLAN equipment atthe European HiperLAN/2 test frequencies (5.49–5.71 GHz) tocharacterize the performance of their so-called multikilometric(MKM) point-to-point infrastructure and a 40 dBm isotropicradiated power (EIRP) limitation. They claimed that the dis-tances to be supported are in the range of 10 km to 300 km,as tested in Italian rural, urban, and mountain environments.Web traffic was used for all measurement campaigns. Thedistinct frequency, that is helpful to avoid interferences ofcommon WLAN networks is 2.4 GHz. The wireless links weretested over an 18-month period and under harsh environmentalconditions (−40 ◦C, ice falls, snow storms, and wind speedsup to 200 km/h). The problems reported are mainly basedon negative weather conditions, such as lightning (transmitterdestruction) and ice falls (damage of solar panels for the energysupply). Various types of damage due to over-voltage causedby the power regulator of the solar panels are reported. Linkoutage is reported at 0% for distances d ≤ 100 km and at 4%at d ≤ 300 km (mainly due to heavy snow storms). The authorsclaimed a high reliability of their evaluated MKM network.

Next, Paul et al. [62] reported on their observation of theWLAN link performance in open outdoor networks. The de-fault packet size was 1470 B for all reported measurementcampaigns. They achieved a maximum range of 1800 m LOSat 148 Mbps with IEEE 802.11n links in outdoor locationsfor back-haul connection among WLAN APs. In their system,polarized directional MIMO antennae (13.5 dBi) are mountedat 5 m poles, and a robust MCS with 3-streams is applied. A3 × 3 MIMO system is evaluated with 2 vertically and 1 hori-zontally polarized antenna (13.5 dBi gain, 20◦ beam-width) at5.5 GHz (ch = 149, htx = 3 m). Further, the authors suggestedincreasing the MAC driver’s default acknowledgment (ACK)timeout value from 24 μs to 38 μs for AP distances of 800 mand 1800 m, respectively. High MCS rates ≥ 18 are reportedto be unavailable for large distance link connectivity. Frameaggregation is found to be beneficial to significantly improvethe throughput performance (≥ 200%) for large distances.Channel bonding and a wider channel bandwidth (40 MHz)along with a short guard interval (SGI) further improves thetransmission performance (450%).

The findings from different antenna constellations byPaul et al. are remarkable. Different antenna (omni-directional)constellations were evaluated, including a linear array and atriangle with different antenna spacing values of 0 to 10 inches.None of the antenna patterns exhibit positive effects on thethroughput; thus, these constellations do not exhibit significantpath orthogonality for 2-stream MCS rates. To explain theMCS rate limitations, they discussed two phenomena. First,the packet error rate is increased at high MCS rates due to thestrict EVM requirements. Device manufacturers must limit thesending power Ptx for their devices at higher modulation rates.Second, the calculation of the Fresnel zone with a range of1800 m corresponds to an antenna elevation of ≥ 3000 m.Hence, the earth’s curvature becomes an obstacle of significantimportance. The measured throughput is reported at 165 Mbps(300 m), 148 Mpbs (800 m), and 40.8 Mpbs (1800 m) forchannel bandwidth 40 MHz, and 100 Mbps (300 m), 95 Mbps(800 m), and 30 Mbps (1800 m) for a channel bandwidth of

Page 6: WLANs - A Lesson 2015

1766 IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015

Fig. 3. MCS rate performance in long-range WLANs (IEEE 802.11n,20/40 MHz bandwidth, 1470 B packet size), as discussed in [62].

20 MHz, indicating a significant performance drop for largedistance ≥ 800 m. Fig. 3 summarizes the observed MCSperformance of IEEE 802.11n in outdoor environments.

B. PHY

Bianchi et al. [60] reported their findings in an IEEE802.11b/g mesh, in which surprisingly poor performance ofthe IEEE 802.11g protocol was observed. It was concludedthat - although IEEE 802.11g utilizes OFDM, which shouldprovide robust transmission performance in a rich multipathenvironment—the short physical layer convergence procedure(PLCP) preamble and the impact of the cyclic prefix or guardinterval preceding each OFDM symbol, which were designedfor indoor use case scenarios, exacerbates packet errors inoutdoor environments. A short ERP-OFDM training sequenceof 16 μs in IEEE 802.11g, compared to 144 μs in IEEE802.11b, would lead to PHY Protocol Data Unit (PPDU) errorsif the synchronization fails. They found, by classifying theframes with errors, that physical errors (PHY errors) dominateover cyclic redundancy check (CRC) and MAC errors in IEEE802.11g, confirming the hypothesis that IEEE 802.11g is notwell suited for outdoor deployment due to the short PLCPpreamble, which results in a small multipath tolerance [60]. Inaddition, Bianchi reported on the asymmetric link performance(uplink, downlink) for all measured outdoor links.

Ting et al. [64] performed an insightful study on the use ofIEEE 802.11n WLANs for multi-hop networks as the backhaullink in Malaysia’s rural areas. Their study focused on maxi-mizing the coverage and backhaul distance of IEEE 802.11nWLAN to enable various applications, e.g., to provide basictelephony access, Internet broadband connection, and commu-nity information access. They reported that co-channel interfer-ence is the main interference contributor in their rural network,which can be easily avoided by careful channel planning, e.g.,the use of orthogonal channels. The authors suggest a generallink budget LB [dB] model, given as

LB = EIRP − Rsen + Grx − FM − Lwall (3)

where EIRP is the effective isotropic radiated power in [dBm],Grx is the receiver antenna gain [dBi], FM is the fade marginin [dB], Rsen is the receiver sensitivity in [dBi], and Lwall isthe wall penetration factor [63]. Lwall is set to zero and FMto 28 dB to assure maximum link outage at 0.1% (Rayleigh

fading). Coverage performance is measured with the systemsettings of channel bandwidth of 40 MHz, transmit EIRP of30 dBm, fc = 5.8 GHz, and Grx = 15 dBi. The link layer (IPlayer) throughput efficiency is approximately 50 % (1/2) of theraw PHY data rate of IEEE 802.11n. The highest data rate(MIMO, 4-stream) is reported at 300 Mbps at a distance of200 m which gradually decreases to 25 Mbps at 1 km. Remark-ably, such distances are only achieved with the lowest MCSrate settings (most robust modulation scheme, 4-streams) [64].A large single-cell coverage area (0.43 square km) is reportedat 1 Mbps and drops gradually to 0.36 square km at 2 Mbpswith further decreases in the coverage area for increasingbandwidth. An increased number of spatial streams were foundto be beneficial to enhance the wireless coverage. The delayspread reduces with increased distance due to the dominanceof the direct link. The delay spread was measured at 500 m as3.99 μs and at 1000 m as 1.80 μs. Interestingly, two differentoptimizations are suggested: increasing the backhaul distanceis more suited for rural environments with low user density,whereas maximizing the coverage area is more suited for sub-urban or urban deployment scenarios. The WLAN power am-plifier (PA) shows a non-linear behavior for QAM modulationsat higher signal levels. Higher QAM performance requires tightlinearization of the power amplifier because any PA distortionmay lead to inter-modulation distortion (IMD) products [65],which may violate the spectral masks which was the systemdefined in [66].

C. MAC

Sheth et al. [54] reported the abysmal link performance oflong-range, point-to-point IEEE 802.11b WLANs in outdoorenvironments. The observation was that outdoor long-rangelinks exhibited an intermediate deliver probability ratio, i.e.,neither clearly bad nor good link quality was observed. In-terference was classified as external WLAN interference, non-WLAN interference, protocol induced losses, and multipathinterference. Although, in outdoor mesh WLAN deployments,the effects of multipath were identified as the major contributorto packet loss, due to the use of an omni-directional antenna, in[54], the effect of external WLAN interference was identifiedas the major contributor—including the use of direct beamantennae, which exacerbate the hidden node problem, in whicha transmitter and interferer erroneously sense the medium tobe idle, thus leading to packet collisions at the receiver. Theauthors in [54] discussed potential remedies, and concludedthat channel selection schemes would be beneficial, but therealization is too complex in large-scale WLAN deployments.They reported that, although the channel distance may be large,signal spillage from the interferer channel could affect theprimary channel, causing frame corruptions that result in CRCerrors [54]. Next, automatic rate fallback (ARF) would allowWLANs to select robust modulation schemes. However, due tothe longer airtime of the data frames, the collision probabilitywould increase, outperforming the effect of the rate adapta-tion. The conclusion was that adaptive coding schemes—herethe forward error correction (FEC) scheme—would result inhigher performance gains; in addition long burst errors are

Page 7: WLANs - A Lesson 2015

AUST et al.: OUTDOOR LONG-RANGE WLANs A LESSON FOR IEEE 802.11ah 1767

thought to be mitigated rather than short burst errors with highamplitude [54]. Note that Sheth observed optimal transmissionresults when the sender increased its sending power tx in theregime of 12–15 dB relative to the interferer.

Raman et al. reported on their design of a new MAC protocolin [67], called 2P, a bipartite media access scheme for whichthey claimed a 20-fold increase of transmission performancecompared to consumer-off-the-shelf (COTS) IEEE 802.11bcarrier sense multiple access/collision avoidance (CSMA/CA).Their focus was on the inefficiency of CSMA/CA in COTSproducts for IEEE 802.11, which provides sufficient trans-mission performance in defined indoor locations; however,by using CSMA/CA in an outdoor long-distance deployment,dramatic impairments were reported. This lack of outdoorperformance motivated the design of an alternative MAC. Theauthors in [67] were motivated by the use of WLAN as a cost-effective alternative to cellular service and sought to retain thecost advantage by redesigning the MAC in which two statesare introduced at each node: the transmission phase and thereception phase. This design requires a synchronized effort ofall nodes within the network.

Using a token introduces new problems, including the loss ofsynchrony and establishment of synchrony after the link setup.In addition, they speculated that the near-field effect cannot beavoided, even with high-precision antennae. A helpful designof the MAC could address the phenomena, which would resultin a cost-efficient solution. As the authors outlined, there is athreat of losing synchrony due to loss of the token or during theonset of the link setup phase.

In [70], the authors outlined the challenges on channel accessmechanisms in the core of the MAC design. In IEEE 802.11 thedistributed coordination function (DCF), the point coordinationfunction (PCF) and the enhanced distributed channel access(EDCA) are utilized. Further, optional access schemes, includ-ing the hybrid coordination function (HCF) and the controlledchannel access (HCCA) are defined, but are based on funda-mental principles which are the same [70]. A comprehensivesurvey on wireless coexistence between IEEE 802.11 and IEEE802.15.4 networks is given in [11], [71], [72].

D. Coverage

There is a difference in regard to wireless coverage in ruraland urban environments. The deployment of the APs of aWLAN mesh in rural areas almost always leads to a patch-work of coverage that does not provide continuous coverage.A reason for this lack of coverage is that user locations areprimarily beside streets, harbors, rivers, etc. The authors in[64] reported that a maximal coverage of IEEE 802.11n with4 spatial streams (SS) can reach up to 800 m at 50 Mbpswith the use of directional antennae of 15 dB gain, which isappropriate for the back-haul link. The use of an antenna with3 dB gain only lead to coverage of 400 m at 50 Mbps, which isappropriate for access links for mobile users. Sevani et al. [68]suggested that TDMA schemes are beneficial in long-rangeWLANs, but should not be applied in mesh-networks.

The studies in [53], [56] indicated that there is no directcorrelation between the SNR value and the link quality. This

finding is in contrast to the observations that the error ratebehaves as a function of the received signal strength in amanner that is closely described by the theory [58]. Abuali et al.[40] reported on WLAN networks which target a coveragerange of up to 3 km. A 2.4 GHz COTS IEEE 802.11g WLANrouter was connected to a 0.5–4 dB automatic gain controller(AGC) WLAN amplifier. Coverage simulations of outdoorWi-Fi hotspots with omni-directional antennae were reportedin [69], which exhibited a good agreement with experimentalresults. Table III summarizes the discussed outdoor long-rangeWLAN systems and provides direct comparisons to the resultsof the existing studies as discussed in this article. The purposeis to obtain the lessons learned by those authors and whatwas observed during their measurement campaigns. Antennaheights, distance, and prerequisites are important indicators tocompare the transmission performance of each outdoor WLAN.Additionally, observations and hypothesis are listed as stated bythe authors.

E. Energy Efficiency

Ab-Hamid et al. [74] discussed the energy efficiency ofsolar-powered long-range WLAN networks of range of 10 kmfor rural communications in which energy is scarce. In theirstudy, six nearby villages were connected to the Internet in themost remote community in Malaysia, and the system had toovercome challenges in the environment, such as mountainsand dense vegetation. Reliable power supply is important toprovide a self-sustainable WLAN with reduced requirement forthe maintenance of APs and to increase the battery lifespan.

A combination of renewable power sources that combinesboth solar and wind to power WLAN mesh networks wasreported by Bernardi et al. [75]; the dynamics in the weatherconditions revealed that this combination of renewable powersources perfectly complements each other. Energy conversionstrategies for large-scale WSNs are outlined in [76].

V. LONG-RANGE WLANs IN EXPOSED AREAS

We report on challenges, which were found in areas, in-cluding sea-surfaces, high altitude unmanned aerial vehicles(UAVs) and tunnels.

A. WLANs Over Sea-Surfaces

Considerations on the path loss over sea-surfaces have re-cently become important. Natural disasters, such as the 2011Japanese earthquake that triggered a massive Tsunami, cancause unprecedented damage to both, human life and the en-vironment. As a result, appropriate warning systems, includingthe sea-level sensing, are of great importance. In [75], theuse of over-water radio propagation is discussed. The authorsdeveloped a long-range WLAN to connect rural environmentsin coastal and remote regions, partly over (sea) water, in theScottish Highlands and Islands. They reported the challengingchannel characteristics for over-water links of: (a) multipathreflections and their dynamics due to changes of the water level(tidal patters) and (b) signal attenuation that is significantly

Page 8: WLANs - A Lesson 2015

1768 IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015

TABLE IIICOMPREHENSIVE OVERVIEW ON OUTDOOR, LONG-RANGE NETWORKS

different than the signal absorption of water, which results inunexpected measured link characteristics that are somewhatdifferent from other observed long-range WLANs in the liter-ature, e.g., from land-based rural WLAN Mesh networks [75],[97]. The WLAN distances considered include 2.5 km, 7.6 km,9.3 km, and 15.9 km, with the longest WLAN link consideredbeing 19 km point-to-point LOS over water.

In [77], it is argued that evaporation ducts and elevationducts over the sea have a significant impact on the path loss bydecreasing the path loss exponent. They concluded that the free

space loss (FSL) model is incorrect for the path loss calculationover a sea-surface.

Meng et al. discussed signal propagation over the sea-surface in [78] and for airborne altitudes (300 m–1.8 km)in [77]. Multipath statistics were analyzed for air-to-groundcommunication at 5.7 GHz (C-band) over sea locations nearSingapore (Singapore Strait) under tropical conditions. Addi-tionally, their findings indicated that the Friis model is notuseful to predict path loss, as observed from their measurementcampaign. The authors confirmed that the 3-ray multipath

Page 9: WLANs - A Lesson 2015

AUST et al.: OUTDOOR LONG-RANGE WLANs A LESSON FOR IEEE 802.11ah 1769

model fits with the observed measurement with 95% confidence(fc = 5–8 GHz). In addition, multipath fading was reportedto significantly contribute to the signal outages in the air-to-ground communication, which increases with higher attenua-tion. Spatial diversity was proposed to mitigate the multipathfading effects by applying selective combining to select theantenna with the highest signal (SNR) [78]. This multipathfading effect is alleviated at higher elevation. Similar obser-vations on multipath characteristics have been reported byBernardi et al. [75], where water levels caused significantdynamics in the signal attenuation of long-range mesh networksover the sea-surface.

B. UAV WLANs

Path loss modeling for UAVs and micro-UAVs (MUAVs), aregarnering attention because of their applicability in new remotesensing applications, surveillance and disaster missions. UAVsoperate at altitudes above ground level: hUAV ≤ 500 m [57]. Themajority of UAVs, which are used in both long-range tacticaland civil UAVs, are equipped with WLANs, making themrelevant to this study. Cheng et al. [79] reported on the wirelessperformance measurements of UAV communications to groundlocated nodes using COTS wireless devices. IEEE 802.11awireless links were evaluated based on RSSI measurementsover distances up to 300 m. The aim was to identify the bestantenna constellation among the airborne UAV and groundstations. Although, the maximum received packet ratio wasreported to be only 33%, Chen concluded that horizontal-to-horizontal configuration is the best constellation if the antennaeare elevated. Additionally, the authors recommend the use ofomni-directional antennae instead of high-gain, narrow-beamantennae for both the UAV and the ground station.

Ahmed et al. [80] reported their observations for aerialwireless sensor networks (AWSNs), which generally consistsof multiple UAVs. The study included the characterization ofcommunication links in real deployments, including ground-to-air (G-A), air-to-air (A-A), and air-to-ground (A-G). Largedistances among the WSN nodes are possible when the el-evation of the sender and receiver are at 2.8 m, reachingdistances up to 240 m. The antenna patterns of the nodes are notcompletely omni-directional in 3D. Two major contributionsare relevant for link degradation: (i) antenna configuration and(ii) multi-path fading effects due to ground reflections. The re-sults indicated a significant dependence of antenna orientationand on the RSSI performance, which was reported to vary by6 dB (0–135◦ orientation, quasi omni-directional). Horizontalmounting improved the RSSI compared to a vertical mountingof the nodes. Both RSSI and the packet reception rate (PRR)were increased when node elevation was increased. The authorsconcluded that near-ground deployments suffer from reflectionsand absorptions in aeronautical WSNs. Raising the antennaheight to 2.8 m resulted in successful PRR (80%) for distances(G-A) of up to 70 m (Ptx = 0 dBm). The maximum distancewas observed at 240 m (A-A). Additionally, the authors con-firmed that grey zones are prominent when the sender and thereceiver are placed at high altitudes, similar to the findingsas reported in [77], where significant impacts of gray zones

were observed at higher altitudes over the sea-surface. Byconsequence, protocol designer must consider remedies againstgrey zones for aerial WSNs. The best antenna configuration wasreported as horizontal transmit antenna to horizontal receiveantenna over a wide range up to 200 m. Again, worst per-formance was observed for vertical-to-vertical dipole antennaeconstellation at small antenna distance due to the null beamconstellation of each antenna beam [79]. In contrast, at largedistances, this situation is less likely to occur, thus resultingin higher UDP throughput performance compared to the useof horizontal oriented dipole antennae [57]. The observationwas that the cross-polarized antenna configuration showedimproved performance at the large distances of 300 m. Trafficoff-loading in cellular networks using a swarm of UAVs wasproposed by Rohode et al. [81]. Rohde assumed free-spacepropagation and used the Friis equation to estimate the pathloss; they determined a path loss exponent of 3.3 (Macro2UEmodel) without considering additional fading effects. A dif-ferentiation between urban and rural areas is beneficial, dueto the impact on propagation, fading and attenuation of theemitted electromagnetic radio waves. The authors argued, thatthe relevant path loss models are Okumura-Hata, COST-Hata,and COST Walfish-Ikegami (COST-WI). A problem with thepath loss models is that they are valid up to an altitude of 50 m.For higher altitudes, a generalized free space propagation model(Friis model) is applicable. At distinct distances, the LLOS

wave and the reflected ground wave Lgrw interfere and thusreduce the received signal φ at given distance d (destructiveinterference). A general estimation of the grey zone is providedin [80]. They concluded that air-to-air communication linksperform best among all types of links. The ground-to-air linkperforms better than the air-to-ground link at the minimum an-tenna heights. Increased antenna height increases the observedthroughput.

In general, cross-polarization has a negative impact on thetransmission performance, as stated in [57]. The authors con-cluded that dipole antennae exhibited the best performance(for moving UAVs) compared to narrow-beam antennae. Theauthors discussed the phenomena of a reduced path loss ex-ponent of less than 2. The authors argued that this reductionwas due to the limitation of the 802.11 equipment, whereat long distances, only packets of sufficiently high receivedsignal power are successfully received. Otherwise, packetswith weak signal strengths would not be successfully decoded,thus resulting in a skewed exponent at large distances. Highersending power is beneficial to increase the successful packetreception [57]. Similar to the observations in [80], the authorsrecommend a high elevation of the ground stations to im-prove the wireless link performance. Remarkably, the authorsreported serious problems involving interference between theUAV board-computer and the 72 MHz R/C receiver of theUAV. The authors speculated that the interference was causedby the 66 MHz bus system, which resulted in a higher noiselevel. Hardware interferences were solved by increasing thedistance between the R/C receiver and the additionally shieldedon-board computer.

In [57] different UAV path loss models were considered,including coverage evaluation of in aerial 3D-networks. The

Page 10: WLANs - A Lesson 2015

1770 IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015

TABLE IVSUMMARY OF REPORTED PROBLEMS THAT HAVE BEEN OBSERVED IN OUTDOOR LONG-RANGE WLANs

results were obtained using a captive balloon. With increasingaltitudes, the RSSI becomes higher up to 290 m, where theantenna directivity was lost. The authors reported an escalationin handovers (HOs) at altitudes of 100–150 m because morecellular base stations (different cell-IDs) are in the line of sightor detected through multi-path propagation. Higher altitudescan result in alternating HOs among the remaining BS, i.e., so-called ping ponging. The authors concluded that in the vicinityof the ground, the impact of antenna directivity and poorground coverage due to shadowing is significant. However, thiseffect diminishes at higher altitudes as the coverage becomeshomogeneous.

In [75], the use of dual polarity high performance anten-nae was proposed to reduce interference. In addition, wirelesslinks are applied with different polarizations (horizontal andvertical). It is argued that this approach can be advantageousin mitigating interference and to increase the link capacity.Table IV presents the summary of the problems found and liststhe remedies proposed that were discussed in the literature. Theaim is to provide a rule-of-thumb based on the reported find-ings in the literature. Additionally, it provides some guidelineon which metric becomes important to focus to optimize anoutdoor WLAN.

C. Wireless Propagation in Tunnels and Mines

Hrovat et al. [8] reported in their survey of radio propagationcharacteristics at 900 MHz for arched and rectangular tunnelsand in coal mines of several kilometers length. Propagation intunnels is significantly different compared to outdoor areas andis classified in three regions, namely, (i) near region with highpath loss, (ii) far region where the waveguide effect is apparent,

and (iii) extreme far region where the path loss follows thefree path loss propagation model. Note that the wave guidedeffect has been reported only for tunnels with large transversedimensions x [m]. The breakpoint location is lbp = x2/λ, whichdepends on the carrier frequency, the tunnel geometry and theantenna characteristics.

VI. LESSONS LEARNED (SO FAR. . .)

The reported performance limitations of outdoor long-rangeWLANs in the literature are classified in environmental con-ditions, PHY settings, MAC settings, and antenna settings.Additionally, wireless performance at high altitudes are ofparamount importance when UAVs are in the research focus;thus, are considered as such in the following overview:

A. Environmental Conditions

WLANs in urban environments are highly affected by twomain sources, namely, multipath and external WLAN interfer-ence. The negative effects of multipath are dominant if omni-directional antennae are used. Multipath effects lead to higherdelay spreads and attenuation (scatter); thus, leading to inter-symbol interference (ISI). If directional antennae are used, aWLAN significantly suffers from external WLAN interference,due to the hidden node effect [54]. Fewer multi-path effectsare reported; thus, a lower delay spread—within one order ofmagnitude—for long distance WLAN links when directionalantennae are used [54]. It was confirmed that antenna heightand antenna directivity are of paramount importance whenincreasing the coverage range (intuitive). Weather conditionshave little or no impact on the path loss (1–2 dB) [58]. The

Page 11: WLANs - A Lesson 2015

AUST et al.: OUTDOOR LONG-RANGE WLANs A LESSON FOR IEEE 802.11ah 1771

TABLE VSUMMARY OF COMMON AND NEWLY PROPOSED PATH LOSS MODELS

lower layers attenuate the effect of atmospheric parameters. Inaddition, error protection at upper layers minimize the influenceof meteorological effects [59]. Harsh weather conditions aremore likely to damage the energy supply (solar panels, powerregulators), e.g., due to thunderstorms or heavy snowfalls.Increasing the backhaul distance is more suited for rural en-vironments with low user density, whereas maximizing thecoverage area is more suited for sub-urban or urban deploymentscenarios [64].

B. PHY Settings

An increased link distance leads to ACK timeouts, due tothe increased propagation delay. The modification of the ACKtimeout value in the WLAN chipset (tACK = 746 μs, Atheros),results in a robust communication distance of 110 km untilthe delay exceeds the ACK timeout and the sender startsretransmitting data frames [54]. At large distances, a mismatchbetween the theoretical path loss models and the measured dataoccurs at the coverage edge, due to the imperfect receiver char-acteristics of the WLAN device at low signal strengths. TDMAschemes are beneficial in long-range networks, but not in mesh-networks, due to lack of time synchronization among multiplehops [68]. The use of a larger number of spatial streams(MIMO) is beneficial to increase the coverage [64]. 2-streamworks well for high MCS rates and long range ≤ 1800 m,whereas 3-stream only works at low MCS rate and short dis-tances ≤ 300 m [62]. The use of a higher number of spatialstreams requires lower MCS rates (most robust modulation)when a large distance is the primary concern [64].

Combining wider channel bandwidth and SGI is recom-mended (450% improvement). The frame aggregation must beenabled to observe the positive effects on wider channel band-width and SGI [62]. Design suggestions in legacy WLANs:IEEE 802.11g is recommended for radio access (high data rateat short distances), whereas IEEE 802.11a is recommendedfor backhaul link setup (more channels, less congested). Over-provisioning of the link budget by selecting higher signalstrength can help to mitigate the effect of external WLAN inter-

ference. 10–15 dB higher signal strength can offer a significantincrease of traffic performance.

The path loss characteristic is highly affected by the sur-roundings; thus, it requires a careful examination of the WLANoutdoor location. Table V summarizes the discussed path lossmodels.

C. MAC Settings

Long-distance WLAN communication significantly suffersfrom CSMA/CA limitations. The increased propagation delayin addition to bidirectional traffic can lead to an increasednumber of packet collisions. High MCS rates (≥ 18, Nss ≥ 2)fail to establish link connectivity over a large coverage [62]. Atlong distances, two effects limit the MCS rate, namely, EVMrestrictions and the Earth’s curvature [62]. Further, in caseswhere ARF changes to a lower MCS rate, the airtime of the dataframes increases, thus leading to higher collision probability indense WLANs [54]. As a result, ARF exacerbates the numberof packet collisions. The decision feedback equalization (DFE)demodulation process optimizes the receiver performance un-der any given channel condition to provide the best possibleSNR for any given environment. A Reed-Solomon (RS) errorcorrection is suggested to mitigate errors, which is capable ofcorrecting up to 2 byte-errors per frame. Packet aggregation isbeneficial over long-range wireless links (200% improvement)[62]. The use of block ACKs can significantly improve theperformance in long-range WLANs, due to reduced overheadand ACK collisions.

D. Antenna Settings

When using directional antennae, researchers suggest usingpolarized antennae to further increase the diversity gain, e.g.,in MIMO systems [82]. This approach has been found to bebeneficial when path loss orthogonally cannot be achieved overlong distances, even in a LOS situation. Linear array antennaeonly exhibit path orthogonality for very short distances [62],[82]. Antenna configuration and the antenna null beam are

Page 12: WLANs - A Lesson 2015

1772 IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015

affecting the transmission. A horizontal antenna configurationis recommended for narrow range, a vertical antenna configura-tion for wide range, respectively.

E. High Altitude

Multipath and ground reflections become dominant withincreased altitude; thus, grey zones increase. This is particularlytrue over a sea-surface. Additionally, high altitudes cause alter-nating handovers (ping ponging) due to the LOS of multipleAPs or BSs. Further, near-ground antenna mounting was foundto be detrimental, whereas high elevation antenna mountingpositively contributes on the coverage range. At high altitudes,the impact of obstacles on the path loss is negligible. Finallyit is noted that using WLANs in scenarios, which were notconsidered during the PHY/MAC standardization process, willresult in unexpected performance loss, e.g., IEEE 802.11bexhibits limitations in point-to-multipoint scenarios [54].

VII. FUTURE RESEARCH CHALLENGES

To optimize the system performance of long-range WLANsfurther research is encouraged, and should be focused on theimprovements of the antenna metric, wireless link metric, andnetwork metric which have a significant impact on the wirelesstransmission characteristic.

A. Antenna Metric

The outdoor use of multiple antennae, either as phased arrayor as MIMO spatial streaming, requires further investigation.There are reported merits and demerits to the use of multipleantenna arrays in long-range outdoor deployments. Althoughlarge antenna heights have been applied to some outdoorWLANs, no guideline is given for the appropriate height to use.Further research is required regarding the transmission charac-teristics when obstacles or persons are crossing the Fresnel-zone. In [66], it has been reported that channel data rates of320 kbps in 25 kHz channels are possible, providing coveragediameters in excess of 60 kilometers with antennae of moderateheight. Thus, the practical upper boundary of 1 MHz must beat 12.8 Mbps. Future measurement studies may show if suchperformance boundaries are realistic, in particular for IEEE802.11ah WLANs.

B. Link Metric

At altitudes of ≥ 50 m common path loss models havebeen found useless. In particular, the Friis model is not rec-ommended. The cause of this effect is reduced shadowing andthe impact of antenna directivity at higher altitude. Instead, newpath loss modes are encouraged, which can be based on 2-ray-ground or 3-ray-ground models. Next, long-range WLANswould be an appropriate choice, to link offshore objects. Re-search is required on the impact of the transmission character-istics over water surface, e.g., when WLANs are deployed astsunami warning systems. Further, the effect of beamformingwhen using consecutive links leads to an increase of link

instability, thereby increasing the received noise variance; thus,studies on beamforming in long-range ad-hoc networks arerequired [83]. Noise floor measurements have been reported,but different results were found. In some measurements, alinearly increasing noise floor was correlated with an increaseof packet errors. In large outdoor deployments, wide-band noisewas not observed [54].

C. Network Metric

There has been little or no research on IEEE 802.15.4 andIEEE 802.11ah network coexistence at 900 MHz. Initial studiessuggest that there is a threat of asynchronous channel access inwhich non-harmonized carrier sense access (CCA) thresholdsresult in high number of packet collisions. Further, there arevarious sources of non-WLAN interference present, includingmicrowave ovens, baby phones, and cordless phones, whichdo not follow the IEEE 802.11 channel access protocol, thusleading to significant packet collisions. Additionally, massiveaccess is considered as a wireless media access scenario indense WLANs [84], [85]. Massive Access is discussed as ausage model in IEEE 802.11ax [86], IEEE 802.16p and LTE-Aand it may appear in M2M and Internet-of-Things (IoT) as anew access scheme [22], [87]. Massive access schemes havebeen proposed in 3GPP networks [90]–[92], but not thoroughlyevaluated. New codebook selection strategies and sectorizationof the coverage area can further mitigate coexistence problems[88], [89]. Finally, the advice is to conduct field tests to iden-tify limitations in transmission performance, e.g., by using aWLAN prototype [93]–[96].

VIII. CONCLUSION

This survey provided a comprehensive summary of the chal-lenges and implications of long-range WLAN deploymentswhich were reported in the scientific literature over the period2002 to 2014. Potential use cases of outdoor WLANs weresuggested, including health care, outdoor activities, metering,and environmental sensing. However, outdoor WLANs haveshown significant performance limitations. Unexpected com-munication problems in outdoor environments as well as invalidpath loss models were found in exposed communication areas.This is particularly true at high altitudes over sea-surfaces.Path loss characteristics were discussed and changes of pathloss coefficients were motivated to model an accurate pathloss performance. So-called grey zones are the cause of signaloutages at high altitudes that otherwise could not be explained.As such, UAVs and 3D-networks were included in this surveyto identify new communication challenges.

Additionally, measured long-distance communication exhib-ited the problem of inaccuracy due to significant hardwarelimitations of the WLAN modules. Two-ray-ground path lossmodels were found to be accurate, whereas the Friis-model isonly applicable for short distances of multiple wavelengths.Multi-stream MIMO is beneficial to extend the coverage, butrobust MCS rates are required; thus, a throughput limitationis the result. A 2-stream configuration is able to connectwireless terminals over a very large distance up to 1800 m,

Page 13: WLANs - A Lesson 2015

AUST et al.: OUTDOOR LONG-RANGE WLANs A LESSON FOR IEEE 802.11ah 1773

whereas the 3-stream configuration only provides short dis-tances. Whereas IEEE 802.11b at 2.4 GHz was reported tobe useful for outdoor deployments, IEEE 802.11g was foundto be useless due to the significant short PHY preambles.In particular, the IEEE 802.11ah protocol is of interest be-cause this emerging sub-1 GHz WLAN protocol amendmentis targeting long-range outdoor deployments at 900 MHz as aworldwide standard. The IEEE 802.11ah PHY is well-designedfor outdoor communication, including robust MCS rates andlong training preambles. However, due to the lack of certi-fied IEEE 802.11ah WLAN modules, little or no informa-tion is reported of outdoor performance of this new WLANstandard.

Our findings are helpful for network designers of sub-1 GHzlong-range outdoor WLANs in constrained areas, includingtunnels, high altitudes, and sea-surfaces. A gap analysis inthis article identified required actions for future research ofIEEE 802.11ah outdoor deployments. Sub-1 GHz WLANs willbe deployed in the near future in various regions, includingChina, Europe, Japan, Singapore, South Korea, and USA. SuchWLANs will exploit a larger coverage area and high wirelesspenetration of obstacles. In anticipation of this deployment,this paper presented the advances of outdoor WLANs andconsidered the challenges that must be addressed so that sub-1 GHz WLANs can be further optimized, making the study oflong-range WLANs a demanding task.

REFERENCES

[1] W. Y. Chen, Home Networking Basis—Transmission Environments andWired/Wireless Protocols. Upper Saddle River, FL, USA: Prentice-Hall,2004.

[2] Draft Standard for Information Technology—Telecommunications andInformation Exchange Between Systems—Local and Metropolitan AreaNetworks—Specific Requirements—Part 11: Wireless LAN Medium Ac-cess Control (MAC) and Physical Layer (PHY) Specifications—Am. 6:S1G License Exempt Operation, IEEE P802.11ah /D5.0, Jun. 2014.

[3] IEEE Standard for Information Technology—Telecommunications andInformation Exchange Between Systems Local and Metropolitan AreaNetworks, Specific Requirements, Part 11: WLAN Medium Access Con-trol (MAC) and Physical Layer (PHY) Specifications, IEEE 802.11,Mar. 2012.

[4] IEEE Standard for Information Technology—Telecommunications andInformation Exchange Between Systems—Local and Metropolitan AreaNetworks, Specific Requirements, Part 11: WLAN Medium Access Control(MAC) and Physical Layer (PHY) Specifications—Am. 5: Enhancementsfor Higher Throughput, IEEE 802.11n, Oct. 2009.

[5] S. H. Aust, “Advanced wireless local area networks in the unlicensed sub-1 GHz ISM-bands,” Ph.D. dissertation, Embedded Softw. Dept., EEMCS,TU Delft, Delft, The Netherlands, Oct. 2014.

[6] Draft Standard for Information Technology—Telecommunications andInformation Exchange Between Systems—Local and Metropolitan AreaNetworks, Specific Requirements, Part 11: WLAN Medium Access Con-trol (MAC) and Physical Layer (PHY) Specifications—Amendment 4:Enhancements for Very High Throughput for Operation in Bands Below6 GHz, IEEE P802.11ac/D5.0, Jan. 2013.

[7] IEEE P802.11—Task Group AH, IEEE 802.11ah, last access: 2014/11/20.[Online]. Available: http://www.ieee802.org/11/Reports/tgah_update.htm

[8] A. Hrovat, G. Kandus, and T. Javornik, “A survey of radio propagationmodeling for tunnels,” IEEE Commun. Surveys Tuts., vol. 16, no. 2,pp. 658–669, 2nd Quart. 2014.

[9] N. Lu and X. Shen, “Scaling laws for throughput capacity and delay inwireless networks—A survey,” IEEE Commun. Surveys Tuts., vol. 16,no. 2, pp. 642–657, 2nd Quart. 2014.

[10] M. Natkaniec, K. Kosek-Szott, S. Szott, and G. Bianchi, “A survey ofmedium access mechanisms for providing QoS in ad-hoc networks,” IEEECommun. Surveys Tuts., vol. 15, no. 2, pp. 592–620, 2nd Quart. 2013.

[11] D. Yang, Y. Xu, and M. Gidlund “Wireless coexistence between IEEE802.11 and.15.4-based networks: A survey,” Int. J. Distrib. Sensor Netw.,vol. 2011, Apr. 2011, Art. ID 912152.

[12] M. Park, “IEEE 802.11ah Specification framework document,” version14, DCN IEEE 802.11-13/0014r0, May 2013.

[13] 950 MHz-Band Telemeter, Telecontrol and Data Transmission RadioEquipment for Specified Low Power Radio Station, ARIB STD-T108ver. 1.0, Association of Radio Industries and Business, Feb. 14, 2012.

[14] S. Aust, “Additional use cases for sub-1 GHz license-exempt frequencybands,” DCN IEEE802.11-10/1458r0, Dec. 2010.

[15] D. Halasz, “Sub-1 GHz license-exempt use cases,” DCN IEEE 802.11-10/1044r0, Sep. 2010.

[16] R. Porat, “Link budget,” DCN IEEE 802.11-11/0552r2, Apr. 19, 2011.[17] D. Halasz, “Sub-1 GHz license-exempt PAR and 5C,” IEEE P802.11

Wireless LANs, DCN IEEE 802.11-10/0001r13, Jul. 2010.[18] C.-C. Wang et al., “Usage cases for 802.11ah,” DCN IEEE 802.11-

11/0341r0, Mar. 14, 2011.[19] D. Halasz, “Categories of TGah use cases and straw polls,” DCN IEEE

802.11-11/0301r0, Mar. 8, 2011.[20] R. de Vegt, “Potential compromise for 802.11ah use case document,”

DCN IEEE 802.11-11/0457r0, Mar. 2011.[21] S. Aust, R. V. Prasad, and I. G. M. M. Niemegeers, “Performance evalu-

ation of sub-1 GHz wireless sensor networks for the smart grid,” in Proc.37th IEEE LCN, Clearwater, FL, USA, Oct. 22–25, 2012, pp. 292–295.

[22] C.-Y. Ho and C.-Y. Huang, “Energy-saving massive access control andresource allocation schemes for M2M communications in OFDMA cellu-lar networks,” IEEE Wireless Commun. Lett., vol. 1, no. 3, pp. 209–212,Jun. 2012.

[23] S. Aust, R. V. Prasad, and I. G. M. M. Niemegeers, “IEEE 802.11ah:Advantages in standards and further challenges for sub-1 GHz Wi-Fi,” in Proc. IEEE ICC, Ottawa, ON, Canada, Jun. 10–15, 2012,pp. 6885–6889.

[24] A. Hazmi, J. Rinne, and M. Valkama, “Feasibility study of IEEE802.11ah radio technology for IoT and M2M use cases,” in Proc.2nd IEEE GLOBECOM Workshop, Anaheim, CA, USA, Dec. 2012,pp. 1687–1692.

[25] L. C. Choo and Z. Lei, “CRC codes for short control frames in IEEE802.11ah,” in Proc. 80th IEEE VTC Fall, Sep. 2014, pp. 1–5.

[26] M. Ghosh and F. LaSita, “Puncturing of CRC codes for IEEE 802.11ah,”in Proc. 78th IEEE VTC Fall, Sep. 2013, pp. 1–5.

[27] R. P. Liu, G. J. Sutton, and I. B. Collings, “Power save with offset listeninterval for IEEE 802.11ah smart grid communications,” in Proc. IEEEICC, Jun. 2013, pp. 4488–4492.

[28] R. P. Liu, G. J. Sutton, and I. B. Collings, “WLAN power save with offsetlisten interval for machine-to-machine communications,” IEEE Trans.Wireless Commun., vol. 13, no. 5, pp. 2552–2562, May 2014.

[29] H.-H. Lin, H.-Y. Wei, and R. Vannithamby, “DeepSleep: IEEE 802.11 en-hancement for energy-harvesting machine-to-machine communications,”in Proc. IEEE GLOBECOM, Dec. 2012, pp. 5231–5236.

[30] J. O. Seo, C. Nam, S. G. Yoon, and S. Bahk, “Group-based contentionin IEEE 802.11ah networks,” in Proc. Int. Conf. ICTC, Oct. 2014,pp. 709–710.

[31] Y. Yang and S. Roy, “Grouping-based MAC protocols for EV charg-ing data transmission in smart metering network,” IEEE J. Sel. AreasCommun., vol. 32, no. 7, pp. 1328–1343, Jul. 2014.

[32] L. Zheng, L. Cai, J. Pan, and M. Ni, “Performance analysis of groupingstrategy for dense IEEE 802.11 networks,” in Proc. IEEE GLOBECOM,Dec. 2013, pp. 219–224.

[33] “About that MiMOMax data rate!—Explaining the MiMOMax datarate calculations,” MIMOMAX White Paper, Jun. 2013, last access:2014/11/20. [Online]. Available: www.mimomax.com

[34] S. Aust and T. Ito, “Sub-1 GHz wireless LAN deployment scenarios anddesign implications in rural areas,” in Proc. IEEE GLOBECOM, Houston,TX, USA, Dec. 2011, pp. 1045–1049.

[35] M. Iwaoka, “IEEE 802.11ah coverage requirement of industrial processautomation use case,” DCN IEEE 802.11-11/0547, Apr. 2011.

[36] N. I. C. Wangi, R. V. Prasad, M. Jacobson, and I. G. M. M. Niemegeers,“Address auto-configuration in wireless ad hoc networks: Protocolsand techniques,” IEEE Wireless Commun., vol. 15, no. 1, pp. 70–80,Feb. 2008.

[37] O. Raeesi, J. Pirskanen, A. Hazmi, T. Levanen, and M. Valkama, “Per-formance evaluation of IEEE 802.11ah and its restricted access windowmechanism,” in Proc. IEEE ICC Workshops, Jun. 2014, pp. 460–466.

[38] O. Raeesi, J. Pirskanen, A. Hazmi, J. Talvitie, and M. Valkama, “Perfor-mance enhancement and evaluation of IEEE 802.11ah multi-access pointnetwork using restricted access window mechanism,” in Proc. IEEE Int.Conf. DCOSS, May 2014, pp. 287–293.

Page 14: WLANs - A Lesson 2015

1774 IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 17, NO. 3, THIRD QUARTER 2015

[39] IEEE Standard for Information Technology—Telecommunications andInformation Exchange Between Systems—Local and Metropolitan AreaNetworks—Specific Requirements, Part 11: Wireless LAN Medium Ac-cess Control (MAC) and Physical Layer (PHY) Specifications, Amend-ment 6: Wireless Access in Vehicular Environments, IEEE Std. 802.11p,Jul. 2010.

[40] N. Abuali, “Power management in solar-powered long range Wi-Fi test-bed,” in Proc. IEEE ICC Workshop, Jun. 2012, pp. 5983–5987.

[41] V. Sivaraman, S. Grover, A. Kurusingal, A. Dhamdhere, and A. Burdett,“Experimental study of mobility in the soccer field with applicationto real-time athlete monitoring,” in Proc. 6th IEEE Int. Conf. WiMobComput., Netw. Commun., Oct. 2010, pp. 337–345.

[42] P. López-Matencio, J. V. Alonso, F. J. González-Castaño, J. L. Sieiro, andJ. J. Alcaraz, “Ambient intelligence assistant for running sports based onk-NN classifiers,” in Proc. 3rd Conf. HSI, Oct. 2010, pp. 605–611.

[43] J. Llosa et al., “REMOTE, a wireless sensor network based system tomonitor rowing performance,” Sensors, vol. 9, no. 9, pp. 7069–7082,Apr. 2010. [Online]. Available: www.mdpi.com/journal/sensors

[44] IEEE, “IEEE 802.24 Smart Grid TAG,” last access: 2014/11/20]. [Online].Available: www.ieee802.org

[45] S. Farahani, ZigBee Wireless Networks and Transceivers. Amsterdam,The Netherlands: Elsevier, 2008.

[46] Priority Action Plan 2, Guidelines for Assessing Wireless Standardsfor Smart Grid Applications, ver. 1.0, NIST, Gaithersburg, MD, USA,Dec. 31, 2010.

[47] S. Aust and T. Ito, “Sub-1 GHz wireless LAN propagation path lossmodels for urban smart grid applications,” in Proc. ICNC, Maui, HI, USA,Jan. 30–Feb. 2, 2012, pp. 116–120.

[48] K. Mase, “Information and communication technology and electricvehicles—Paving the way towards a smart community,” IEICE Trans.Commun., vol. E95-B, no. 6, pp. 1902–1910, Jun. 2012.

[49] NTT DoCoMo, “Petfit: Smart life with your dog,” last access: 2014/11/20. [Online]. Available: https://www.mydocomo.com/onlineshop/products/

[50] NEC, “M2M serivce: Petfit,” last access: 2014/11/20. [Online]. Available:http://jpn.nec.com/press/201402/20140213_01.html

[51] J. Polastre, J. Hill, and D. Culler, “Versatile low power media access forwireless sensor networks,” in Proc. 2nd Int. Conf. Embedded Netw. SensorSyst., Oct. 2010, pp. 3–5.

[52] T. Tanaka, S. Aust, and T. Ito, “Enhanced energy consumption reductionschemes in radio-on-demand WLANs,” in Proc. 24th Annu. IEEE Int.Symp. PIMRC, London, U.K., Sep. 2013, pp. 1–6.

[53] D. Aguayo, J. Bicket, S. Biswas, G. Judd, and R. Morris, “Link-levelmeasurements form an 802.11b mesh network,” in Proc. SIGCOMMConf. Appl., Technol., Architect., Protocols Comput. Commun., Nov. 2004,pp. 121–132.

[54] A. Sheth, S. Nedevshi, R. Patra, and L. Subramanian, “Packet loss char-acterization in Wi-Fi-based long distance networks,” in Proc. 26th IEEEINFOCOM, Mar. 2007, pp. 312–320.

[55] H. El-Sayed, S. Zeadally, and M. Boulmalf, “Experimental evaluationand characterization of long-distance 802.11g links,” in Proc. 7th ICN,Mar. 2008, pp. 511–516.

[56] M. Afanasyev, T. Chen, G. M. Voelker, and A. C. Snoeren, “Usagepatterns in an urban Wi-Fi network,” IEEE/ACM Trans. Netw., vol. 18,no. 5, pp. 1359–1372, Oct. 2010.

[57] N. Goddemeier, K. Daniel, and C. Wietfeld, “Coverage evaluation of wire-less networks for unmanned aerial systems,” in Proc. IEEE GLOBECOMWorkshop, Dec. 2010, pp. 1760–1765.

[58] K. Chebrolu, B. Raman, and S. Sen, “Long-distance 802.11b links: Per-formance measurements and experience,” in Proc. 12th Annu. Int. Conf.MobiCom Netw., Mar. 2006, pp. 74–85.

[59] D. Bri, M. Fernández-Diego, M. Garcia, F. Ramos, and J. Lloret, “Howthe weather impacts on the performance of an outdoor WLAN,” IEEECommun. Lett., vol. 16, no. 8, pp. 1184–1187, Aug. 2012.

[60] G. Bianchi, F. Formisano, and D. Giustiniano, “802.11b/g link level mea-surements for an outdoor wireless campus network,” in Proc. Int. Symp.WoWMoM Netw., Sep. 2006, pp. 1–6.

[61] D. Trinchero, R. Stefanelli, and A. Galardini, “Reliability and scala-bility analysis of low cost long distance IP-based wireless networks,”in Proc. ITU-T Kalaidoscope—Innov. Digit. Inclusions, Oct. 2009,pp. 1–6.

[62] U. Paul, R. Crepaldi, J. Lee, S. J. Lee, and R. Etkin, “Characterizing Wi-Fi link performance in open outdoor networks,” in Proc. 8th Annu. IEEECommun. Soc. Conf. Sensor, Mesh Ad Hoc Commun. Netw., Oct. 2011,pp. 251–259.

[63] J. S. Seybold, Introduction to RF Propagation. Hoboken, NJ, USA:Wiley, 2005.

[64] A. Ting, D. Chieng, and K. H. Kwong, “Capacity and coverage analysisof rural multi-radio multi-hop network deployment using IEEE802.11nradios,” in Proc. 10th IEEE MICC, Oct. 2011, pp. 77–82.

[65] M. O’Droma, Y. Lei, E. Bertran, and P. Gilabert, “Analysis of inter-modulation products and nonlinear distortion in RF OFDM transmittersystems,” in Proc. IEEE 69th VTC Spring, Apr. 2009, pp. 1–5.

[66] Buyers Guide, “UHF data pipeline—New technology promises data ratesof 320 kilobits per second (kbps) in 25-kilohertz channels in the 400 MHzband,” Radio Resource International, Centennial, CO, USA, May 2009,pp. 12–16.

[67] B. Raman and K. Chebrolu, “Design and evaluation of a new MACprotocol for long-distance 802.11 mesh networks,” in Proc. 11th Annu.Int. Conf. MobiCom Netw., Mar. 2005, pp. 156–169.

[68] V. Sevani, B. Raman, and P. Joshi, “Implementation-based evaluation ofa full-fledged multihop TDMA-MAC for WiFi mesh networks,” IEEETrans. Mobile Comput., vol. 13, no. 2, pp. 392–406, Feb. 2014.

[69] C. K. Chio, S. W. Ting, K. W. Tam, and T. K. Sarkar, “Coverage analysisof outdoor public Wi-Fi hotspot using full-wave electromagnetic simula-tion,” in Proc. IEEE APSURSI, Jul. 2013, pp. 1538–1539.

[70] Y. Zhou, H. Wang, S. Zheng, and Z. Z. Lei, “Advances in IEEE 802.11ahstandardization for machine-type communications in sub-1 GHz WLAN,”in Proc. 2nd IEEE ICC Workshop, Jun. 2013, pp. 1289–1293.

[71] C. S. Metcalf, T. Camp, M. Colagrosso, and O. Chase, “TestbedProfiler: Avalidation tool for wireless sensor network testbed deployment,” in Proc.35th IEEE LCN, Oct. 2010, pp. 188–191.

[72] J.-P. Sheu, C.-J. Chang, C.-Y. Sun, and W.-K. Hu, “WSNTB: A testbedfor heterogeneous wireless sensor networks,” in Proc. 1st IEEE Int. Conf.Ubi-Media Comput., Aug. 2008, pp. 338–343.

[73] K. Hanada, K. Yamamoto, M. Morikura, K. Ishihara, and R. Kudo,“Game-theoretic analysis of multi-bandwidth channel selection by co-ordinated APs in WLANs,” IEICE Trans. Commun., vol. E96-B, no. 6,pp. 1277–1287, Jun. 2013.

[74] K. Ab-Hamid, C. E. Tan, and S. P. Lau, “Self-sustainable energy effi-cient long range WiFi network for rural communities,” in Proc. IEEEGLOBECOM Workshop, Dec. 2011, pp. 1050–1055.

[75] G. Bernardi, P. Buneman, and M. K. Marina, “Tegola tiered mesh networktestbed in rural scotland,” in Proc. MobiCom, Oct. 2008, pp. 1–6.

[76] T. V. Prabhakar, “An empirical approach towards zero energy networks(zen),” Ph.D. dissertation, Embedded Softw. Dept., EEMCS, TU Delft,Delft, The Netherlands, 2012.

[77] Y. S. Meng and Y. H. Lee, “Measurements and characterizations of air-to-ground channel over sea-surface at C-band with low airborne altitudes,”IEEE Trans. Veh. Technol., vol. 60, no. 4, pp. 1943–1948, May 2011.

[78] Y. S. Meng and Y. H. Lee, “Multipath characterization and fade mitigationof air-to-ground propagation channel over tropical sea-surface at C band,”in Proc. IEEE Int. Symp. APSURSI, Nov. 2010, pp. 1–4.

[79] C.-M. Cheng, P.-H. Hsiao, H. T. Kung, and D. Vlah, “Performancemeasurement of 802.11a wireless links from UAV to ground nodeswith various antenna orientations,” in Proc. 15th ICCCN, Oct. 2006,pp. 303–308.

[80] N. Ahmed, S. S. Kanhere, and S. Jha, “Link characterization for aerialwireless sensor networks,” in Proc. IEEE GLOBECOM, Dec. 2011,pp. 1274–1279.

[81] S. Rohde and C. Wietfeld, “Interference aware positioning of aerial relaysfor cell overload and outage compensation,” in Proc. IEEE VTC Fall,Apr. 2012, pp. 1–5.

[82] D. Gesbert, H. Bolcskei, D. Gore, and A. Paulraj, “Outdoor MIMOwireless channels: Models and performance prediction,” IEEE Trans.Commun., vol. 50, no. 12, pp. 1926–1934, Dec. 2002.

[83] E. A. Gharavol, Y.-C. Liang, and K. Mouthaan, “Robust linear beamform-ing for MIMO relay with imperfect channel state information,” in Proc.21st IEEE Int. Symp. Pers. Indoor Mobile Radio Commun., Apr. 2010,pp. 510–514.

[84] S. Aust and T. Ito, “Sub 1 GHz wireless LAN propagation path lossmodels for urban smart grid applications,” in Proc. ICNC Workshop,Maui, HI, USA, Jan. 2012, pp. 1–8.

[85] H. Ma and S. Roy, “Contention window and transmission opportunityadaptation for dense IEEE 802.11 WLAN based on loss differentiation,”in Proc. IEEE ICC, Beijing, China, May 2008, pp. 2556–2560.

[86] “Status of IEEE 802.11 HEW Study Group—High Efficiency WLAN(HEW),” IEEE P802.11—High efficiency WLAN study group—Meetingupdate, last access: 2014/11/20. [Online]. Available: http://www.ieee802.org/11/Reports/hew_update.htm

[87] S. Aust, R. V. Prasad, and I. G. M. M. Niemegeers, “A framework formassive access and radio resource management in urban WLANs,” inProc. 38th IEEE LCN, Sydney, Australia, Oct. 2013, pp. 1–6.

[88] S. Aust, R. V. Prasad, and I. G. M. M. Niemegeers, “Codebook selectionstrategies in long-range sub-1 GHz WLANs,” Procedia Comput. Sci.,vol. 32, pp. 133–140, 2014.

Page 15: WLANs - A Lesson 2015

AUST et al.: OUTDOOR LONG-RANGE WLANs A LESSON FOR IEEE 802.11ah 1775

[89] S. Aust, R. V. Prasad, and I. G. M. M. Niemegeers, “Sector-based RTC/CTS access scheme for high density WLAN sensor net-works,” in Proc. 39th IEEE LCN, Edmonton, AB, Canada, Sep. 2014,pp. 697–701.

[90] Service Requirements for Machine-Type Communications, 3GPP TS22.368 V10.1.0, Jul. 2010.

[91] System Improvement for Machine-Type Communications, 3GPP TR23.888 V0.5.1, Jul. 2010.

[92] Machine-to-Machine (M2M) Communication Study Report, IEEE80216p-10 005, May 2010.

[93] S. Aust, R. V. Prasad, and I. G. M. M. Niemegeers, “Performance studyof MIMO-OFDM platform in narrow-band sub-1 GHz wireless LANs,”in Proc. WiOpt/WinMee, 2013, pp. 1–5.

[94] S. Aust, R. V. Prasad, and I. G. M. M. Niemegeers, “Analysis of the per-formance boundaries of sub-1 GHz WLANs in the 920 MHz ISM-band,”in Proc. 10th Int. Symp. Wireless Commun. Syst., Ilmenau, Germany,Aug. 2013, pp. 1–6.

[95] S. Aust, R. V. Prasad, and I. G. M. M. Niemegeers, “Advances in wirelessM2M and IoT: Rapid SDR-prototyping of IEEE 802.11ah,” in Proc. 39thIEEE LCN, Edmonton, AB, Canada, Sep. 8–11, 2014, pp. 1–3.

[96] H. Murata et al., “Software radio-based distributed multi-user MIMOtestbed: Towards green wireless communications,” IEICE Trans.Fundam., vol. E96-A, no. 1, pp. 247–254, Jan. 2013.

[97] S. Uludag, T. Imboden, and K. Akkaya, “A taxonomy and evaluationfor developing 802.11-based wireless mesh network testbeds,” Int. J.Commun. Syst., vol. 25, no. 8, pp. 963–990, Aug. 2011.

Stefan Aust (M’01) received the Dipl.-Ing. degree inelectronics and information technology engineeringfrom the University of Bremen, Germany, in 2001.He received the Ph.D. degree from the Faculty ofElectrical Engineering, Mathematics and ComputerScience, at the Delft University of Technology, TheNetherlands, in 2014. His doctoral research focusedon the physical layer and medium access control inlong-range 920 MHz WLANs. His research interestsinclude information dissemination in smart sensornetworks, high efficiency wireless WLANs, and au-

tomotive software platforms. From 2001 to 2005, he was a Research Associateat the ComNets Group, University of Bremen, working on the European Unionfunded projects xMotion and NOMAD. He has worked in the German BMBFfunded project IPonAir together with the Siemens AG, Munich and Berlin.From 2005 to 2008, he was with ATR International in Kyoto, Japan, as groupleader in the Japanese Government funded project Cognitive Radio. Since2008, he has been with NEC Communication Systems, Ltd., Kawasaki, Japan.In 2012 he was promoted to the R&D Manager position. He has been anorganizing committee member of IEEE LCN and goSMART. Dr. Aust is activein standardization, including IEEE DySPAN-SC, IEEE 802.11, and .15. He is amember of the IEICE.

R. Venkatesha Prasad (SM’12) received the Ph.D.degree from Indian Institute Science (IISc), Banga-lore, India, in 2004, during which he designed a scal-able VoIP conferencing platform. Part of this thesislead to a startup venture, Esqube CommunicationSolutions, where he worked on a part-time basis asa Senior Design Consultant, from 2006 to 2009, andlead a team of up to 10 engineers, developing manyreal-time applications including bridging anonymousVoIP calls called Click-to-Talk for portals. Whileat Esqube, eight patent applications and three PCT

applications were filed along with his colleagues. In 2005, he joined TUDelftas a PostDoc to work on the EU FP7 Magnet Project and the Dutch project PNP-2008 on personal networks (PNs). His work involved evolving PN networkarchitecture and foreign communication with TUDelft team and resulted in anECMA report. He also has worked on cognitive radio networks (CRNs) and60GHz networks for future homes. He is contributing to IEEE standards onCRNs. Currently, he is involved in Internet of Things (IoT), cyber physicalsystems (CPS), and energy harvesting networks. He is working on EU fundediCore project on IoTs. At TUDelft, he has been supervising Ph.D. and M.Sc.students and his work here has resulted in over 150 publications. He wasCo-Founder and a main organizer of the International Workshop E2Nets inconjunction with IEEE ICC 2010–15. He is General Co-Chair, IEEE ICC 2015Workshop on Next Generation Green ICT. He was Track Chair of CognitiveRadio and Spectrum Sensing of IEEE VTC2013-Spring in Dresden, and is aTutorial Chair for IEEE Globecom-2015. He was TPC chair of IEEE Sympo-sium on Communications and Vehicular Technology in the Benelux (SCVT)2014. He is a contributor to the academic community by leading many IEEEactivities, such as memberships of standards boards and technical committees,as well as being a reviewer and organizer of conferences and workshops. Heis also a member of TCCN, AHSNTC, TCGCC, and TCCC. He is a SeniorMember of ACM.

Ignas G. M. M. Niemegeers received the degree inelectrical engineering from the University of Ghent,Belgium, in 1970, the M.Sc.E. degree in computerengineering in 1972, and the Ph.D. degree fromPurdue University, West Lafayette, IN, USA, in1978. From 1978 to 1981, he was a designer ofpacket switching networks at Bell Telephone Man-ufacturing, Antwerp, Belgium. From 1981 to 2002,he was a Professor in the Computer Science andthe Electrical Engineering Departments, Universityof Twente, Enschede, The Netherlands. From 1995

to 2001, he was the Scientific Director of the Centre for Telematics and Infor-mation Technology at the University of Twente. Since May 2002, he has beenthe Chair of the Wireless and Mobile Communications Group, Delft Universityof Technology, where he heads the Telecommunications Department. He is amember of the Expert Advisory Group of the European technology platformeMobility and IFIP TC-6 on networking. He has (co)authored nearly 300scientific publications and coauthored a book on personal networks.


Recommended