Coordinated Dynamic Spectrum Management ofLTE-U and Wi-Fi Networks

Shweta Sagari∗, Samuel Baysting∗, Dola Saha†, Ivan Seskar∗, Wade Trappe∗, Dipankar Raychaudhuri∗∗WINLAB, Rutgers University, {shsagari, sbaysting, seskar, trappe, ray}@winlab.rutgers.edu

†NEC Labs America, dola@nec-labs.com

Abstract—This paper investigates the co-existence of Wi-Fiand LTE in emerging unlicensed frequency bands which areintended to accommodate multiple radio access technologies. Wi-Fi and LTE are the two most prominent access technologiesbeing deployed today, motivating further study of the inter-systeminterference arising in such shared spectrum scenarios as wellas possible techniques for enabling improved co-existence. Ananalytical model for evaluating the baseline performance of co-existing Wi-Fi and LTE is developed and used to obtain baselineperformance measures. The results show that both Wi-Fi andLTE networks cause significant interference to each other andthat the degradation is dependent on a number of factors suchas power levels and physical topology. The model-based resultsare partially validated via experimental evaluations using USRPbased SDR platforms on the ORBIT testbed. Further, inter-network coordination with logically centralized radio resourcemanagement across Wi-Fi and LTE systems is proposed as apossible solution for improved co-existence. Numerical resultsare presented showing significant gains in both Wi-Fi andLTE performance with the proposed inter-network coordinationapproach.

Keywords—Wi-Fi, LTE-U, dynamic spectrum management,inter-network coordination, optimization

I. INTRODUCTION

Exponential growth in mobile data usage is driven by thefact that Internet applications of all kinds are rapidly migratingfrom wired PCs to mobile smartphones, tablets, mobile APsand other portable devices [1]. Industry is already gearing upfor the 1000x increase in data capacity, which has given rise tothe concept of the 5th Generation (5G) mobile network. The5G vision, though, is not limited to matching the increase inmobile data demand, but it also includes an improved overallservice-oriented user experience with immersive applications,such as high definition video streaming, real-time interactivegames, applications in wearable mobile devices, ubiquitoushealth care, mobile cloud, etc. [2]–[4]. For such applications,the system needs to provide improved Quality of Experience(QoE), which can be factored in different ways: better cell/edgecoverage (availability of service), lower latency (round triptime), lower power consumption (longer battery life), reliableservices, cost-effective network, and support for mobility.

To meet such a high Quality-of-Service and system ca-pacity demand, there have been three main solutions pro-posed [5]: a) addition of more radio spectrum for mobileservices (increase in MHz), b) deployment of small cells(increase in bits/Hz/km2), and c) efficient spectrum utilization(increase in bits/second /Hz/km2). Several spectrum bands, asshown in figure 1, have been opened up for mobile and fixed

55 - 698MHz 2.4 -2.5GHz

5.15 - 5.835GHz

3.55 - 3.7GHz

57 - 64GHz

TV White Space 2.4GHz ISM 3.5GHz Shared band

5GHz UNII/ISM

60GHz mmWave Band

Fig. 1. Proposed spectrum bands for deployment of LTE/Wi-Fi small cells.

wireless broadband services. These include 2.4 and 5 GHzunlicensed bands for the proposed unlicensed LTE operationas a secondary LTE carrier [6]. These bands are currentlyutilized by unlicensed technologies such as Wi-Fi/Bluetooth.The 3.5 GHz band, which is currently utilized for militaryand satellite operations has also been proposed for small cell(Wi-Fi/LTE based) services [7]. Another possibility is the 60GHz band (millimeter wave technology), which is well suitedfor short-distance communications including Gbps Wi-Fi, 5Gcellular and peer-to-peer communications [8]. In addition,opportunistic spectrum access is also possible in TV whitespaces for small cell/backhaul operations [9].

These emerging unlicensed band scenarios will lead toco-channel deployment of multiple radio access technologies(RATs) by multiple operators. These different RATs, designedfor specific purposes at different frequencies, now must coexistin the same frequency, time and space. This causes increasedinterference to each other and degradation of the overallsystem performance is eminent due to the lack of inter-RATcompatibility. Figure 2 shows two such scenarios, where thetwo networks interfere with each other. When Wi-Fi AccessPoint is within the transmission zone of LTE, it senses themedium and postpones transmission due to detection of LTEHome eNodeB’s (HeNB) transmission power in the spectrumband as shown in figure 2(a). Consequently, the Wi-Fi linkfrom AP to Client suffers in presence of LTE transmission.The main reason for this disproportionate share of the mediumis due to the fact that LTE does not sense other transmissionsbefore transmitting. On the other hand, Wi-Fi is designed tocoexist with other networks as it senses the channel beforeany transmission. However, if LTE works as supplementaldownlink only mode, UEs do not transmit at all. So, a Wi-Fi AP, which cannot sense LTE HeNB’s transmission, willtransmit and cause interference at the nearby UEs, as shownin figure 2(b). This problem also exists in multiple Wi-Fi linkswith some overlap in collision domain, but the network canrecover packets quickly as a) packets are transmitted for a very

HeNB

UE1 UE2

AP

Client

(a) Interference caused by LTE.

HeNB

UE1 UE2AP

Client

(b) Interference caused by Wi-Fi.

Fig. 2. Scenarios showing challenges of coexistence of LTE and Wi-Fi inthe same unlicensed spectrum.

short duration in Wi-Fi, compared to longer frames in LTE andb) all the nodes perform carrier sensing before transmission.Therefore, to fully utilize the benefits of new spectrum bandsand deployments of HetNets, efficient spectrum utilizationneeds to be provided by the dynamic spectrum coordinationframework and the supporting network architecture.

It is reasonable to forecast that Wi-Fi and LTE will beamong the dominant technologies used for radio access pur-poses over the next few years. Thus, this paper focuses on thecoordinated coexistence between these two technologies. LTEis designed to operate solely in a spectrum, which when oper-ating in unlicensed spectrum, is termed LTE-U. It is suggestedin 3GPP, that LTE-U will be used as a supplemental downlink,whereas the uplink will use licensed spectrum. This makes thedeployment even more challenging as the UE’s do not transmitin unlicensed spectrum yet experience interference from Wi-Fi transmissions. To alleviate these problems, we extend theinterference characterization of co-channel deployment of Wi-Fi and LTE using simplistic but accurate analytical model [10].Then, we validate this model through experimental analysis ofco-channel deployment in the 2.4 GHz band, using the ORBITtestbed and LTE on USRP platforms available at WINLAB.

To support the co-existence of a multi-RAT network, wepropose a dynamic spectrum coordination framework, whichis enabled by a Software Defined Network (SDN) architecture.SDN is technology-agnostic, can accommodate different radiostandards and does not require change to existing standards orprotocols. In contrast to existing technology-centric solutions,this is a desirable feature, especially in the rapid develop-ment of upcoming technologies and spectrum bands [11]–[13].Furthermore, the proposed framework takes advantage of theubiquitous Internet connectivity available at wireless devicesand provides the pseudo-global network with the ability toconsider policy requirements in conjunction with improvedvisibility of each of the technologies, spectrum bands, clientsand/or operators. Thus, it offers significant benefits for spec-trum allocation over radio based control channels [14] orcentralized spectrum servers [15].

SDN enabled inter-network cooperation can be achieved byoptimizing several spectrum usage parameters such as powercontrol, channel selection, rate allocation, and duty cycle, etc.In this paper, we focus on power control at both LTE andWi-Fi, which maximizes aggregate throughput at all clientsacross both Wi-Fi and LTE networks along with considerationof throughput requirement at each client [16], [17]. We also

propose to apply validated interference characterization ofWi-Fi-LTE coexistence in the optimization framework, whichcaptures the specific requirements of each of the technologies.In general, we adopt the geometric programming frameworkdeveloped in [18] for the LTE-only network and enhance it toaccommodate Wi-Fi network.

The major contributions of this work are as follows:

• We introduce an analytical model to characterize theinterference between Wi-Fi and LTE networks, whenthey coexist and share the medium in time, frequencyand space. We have also validated the model by per-forming experimental analysis using USRP based LTEnodes and commercial off-the-shelf (COTS) IEEE802.11g devices in the ORBIT testbed.

• We propose a coordination framework to facilitatedynamic spectrum management among multi-operatorand multi-technology networks over a large geograph-ical area.

• We propose a logically centralized optimizationframework that involves dynamic coordination be-tween Wi-Fi and LTE networks by exploiting powercontrol and time division channel access diversity.

• We evaluate the proposed optimization frameworkfor improved coexistence between Wi-Fi and LTEnetworks.

The rest of the paper is organized as follows. In §II,we discuss previous work on this topic and distinguish ourwork from existing literature. In §III, we propose an analyticalmodel to characterize the interference between Wi-Fi andLTE networks followed by partial experimental validation ofthe model. In §IV, we propose an SDN-based inter-networkcoordination architecture, which can be used for transferringcontrol messages between the different entities in the network.We use two approaches - power control and channel accesstime sharing methods to jointly optimize the spectrum sharingamong Wi-Fi and LTE networks, which is described in §VI,followed by their evaluation in §VII. We conclude in §VIII.

II. BACKGROUND ON WI-FI/LTE CO-EXISTENCE

Coordination between multi-RAT networks with LTE andWi-Fi is challenging due to the difference in the medium accesscontrol (MAC) layer of the two technologies.

Wi-Fi is based on the distributed coordination function(DCF) where each transmitter senses the channel energy fortransmission opportunities and collision avoidance. In partic-ular, clear channel assessment (CCA) in Wi-Fi involves twofunctions to detect any on-going transmissions [19], [20] -

1) Carrier sense: Defines the ability of the Wi-Fi node todetect and decode other nodes’ preambles, which mostlikely announces an incoming transmission. In such cases,Wi-Fi nodes are said to be in the CSMA range of eachother other. For the basic DCF with no RTS/CTS, theWi-Fi throughput can be accurately characterized usingthe Markov chain analysis given in Bianchi’s model [21],assuming a saturated traffic condition (at least 1 packetis waiting to be sent) at each node. Wi-Fi channel rates

used in the [21] can be modeled as a function of Signal-to-Interference-plus-Noise ratio. Our throughput analysisgiven in the section III is based on Bianchi’s model.

2) Energy detection: Defines the ability of Wi-Fi to detectnon-Wi-Fi (in this case, LTE) energy in the operatingchannel and back off the data transmission. If the in-band signal energy crosses a certain threshold, the channelis detected as busy (no Wi-Fi transmission) until thechannel energy is below the threshold. Thus, this func-tion becomes the key parameter for characterizing Wi-Fithroughput in the co-channel deployment with LTE.

LTE has both frequency division (FDD) and time division(TDD) multiplexing modes to operate. But to operate inunlicensed spectrum, supplemental downlink and TDD accessis preferred. In either of the operations, data packets are sched-uled in the successive time frames. LTE is based on orthogonalfrequency-division multiple access (OFDMA), where a subsetof subcarriers can be assigned to multiple users for a certainsymbol time. This offers LTE an additional diversity in the timeand frequency domain that Wi-Fi lacks, since Wi-Fi assignsfixed bandwidth to a single user at any time. Further, due tocurrent exclusive band operation, LTE does not employ anysharing features in the channel access mechanisms. Thus, thecoexistence performance of both Wi-Fi and LTE is largelyunpredictable and may lead to unfair spectrum sharing or thestarvation of one of the technologies [22], [23].

In the literature, spectrum management in shared fre-quency bands have been discussed for multi-RAT heteroge-neous networks primarily focusing on IEEE 802.11/16 net-works [12]–[14]. For instance, Cognitive WiMAX achievescooperative resource management in hierarchical network withpower/frequency assignment optimization, Listen-before-Talk(LBT), etc. along with guaranteed QoE [24], [25]. These prin-ciples need to be extended to Wi-Fi and LTE coexistence andmodified specific to their protocols. Wi-Fi/LTE coexistence hasbeen studied in the context of TV white space [26], in-devicecoexistence [27], and LTE unlicensed (LTE-U) [28]–[30].Studies [29]–[31] propose CSMA/sensing based modificationsin LTE such as LBT, RTS/CTS protocol, and slotted channelaccess. Other solustions such as blank LTE subframes/LTEmuting (feature in LTE Release 10/11) [26], [32], carriersensing adaptive transmission (CSAT) [29], interference awarepower control in LTE [33] require LTE to transfer its resourcesto Wi-Fi. These schemes give Wi-Fi transmission opportunitiesbut also lead to performance tradeoffs for LTE. Further, timedomain solutions often require time synchronization betweenWi-Fi and LTE and increase channel signaling. Frequency andLTE bandwidth diversities are explored in studies [29] and[34], respectively.

In this paper, we propose Wi-Fi/LTE coordination algo-rithms based on optimization in power and frequency domain,which does not require modifications in existing MAC layerof Wi-Fi/LTE. Our time division channel access (TDCA) al-gorithm resembles CSAT, but TDCA is a centralized approachwith a joint consideration of Wi-Fi/LTE QoE requirementsfor fairness. Furthermore, no/limited details of LTE-U co-existence mechanisms (adaptive duty cycle/switch-OFF) andWi-Fi/LTE interference model are available in public domain.Also previous studies have yet to consider dense Wi-Fi andLTE deployment scenarios in detail. Notably, in the literature,

there are no previous studies experimentally evaluating thecoexistence performance of Wi-Fi and LTE. Thus, this paperfocuses on these aspects to provide a complete evaluation.

III. INTERFERENCE CHARACTERIZATION

A. Interference Characterization Model

We propose an analytical model to characterize the inter-ference between Wi-Fi and LTE, while considering the Wi-Fi sensing mechanism (clear channel assessment (CCA)) andscheduled and persistent packet transmission at LTE. To illus-trate, we focus on a co-channel deployment involving a singleWi-Fi and a single LTE cell, which involves disseminatingthe interaction of both technologies in detail and establish abuilding block to study a complex co-channel deployment ofmultiple Wi-Fis/LTEs.

In a downlink deployment scenario, a single client anda full buffer (saturated traffic condition) is assumed at eachAP under no MIMO. Transmit powers are denoted as Pi, i ∈{w, l} where w and l are indices to denote Wi-Fi and LTElinks, respectively. We note that the maximum transmissionpower of an LTE small cell is comparable to that of the Wi-Fi, and thus is consistent with regulations of unlicensed bands.

The power received from a transmitter j at a receiver iis given by PjGij where Gij ≥ 0 represents a channel gainwhich is inversely proportional to dγij where dij is the distancebetween i and j and γ is the path loss exponent. Gij mayalso include antenna gain, cable loss and wall loss. Signal-to-Interference-plus-Noise (SINR) of the link i is given as

Si =PiGii

PjGij +Ni, i, j ∈ {w, l}, i 6= j (1)

where Ni is noise power for receiver i. For the case of a singleWi-Fi and LTE, if i represents the Wi-Fi link, then j is theLTE link, and vice versa.

Throughput, Ri, i ∈ {w, l}, is represented as a function ofSi as

Ri = αiB log2(1 + βiSi), i ∈ {w, l}, (2)

where B is a channel bandwidth; βi is a factor associated withthe modulation scheme. For LTE, αl is a bandwidth efficiencydue to factors adjacent channel leakage ratio and practicalfilter, cyclic prefix, pilot assisted channel estimation, signalingoverhead, etc. For Wi-Fi, αw is the bandwidth efficiency ofCSMA/CA, which comes from the Markov chain analysis ofCSMA/CA [21] with

ηE =TEE[S]

, ηS =TSE[S]

, ηC =TCE[S]

, (3)

where E[S] is the expected time per Wi-Fi packet transmission;TE , TS , TC are the average times per E[S] that the channel isempty due to random backoff, or busy due to the successfultransmission or packet collision (for multiple Wi-Fis in theCSMA range), respectively. αw is mainly associated with ηS .

{αi, βi} is approximated based on throughput models givenin [10] so that for LTE, Rl matches with throughput achievedunder variable channel quality index (CQI), and for Wi-Fi, Rwmatches throughput according to Biachi’s CSMA/CA model.

1Throughput the paper, LTE home-eNB (HeNB) is also referred as accesspoint (AP) for the purpose of convenience

1) Characterization of Wi-Fi Throughput: Assuming λc isa threshold of CCA energy detection mechanism, if channelenergy at the Wi-Fi node is higher than λc, Wi-Fi would holdback the data transmission, otherwise it transmits at a datarate based on the SINR of the link. Wi-Fi throughput with andwithout LTE is given as

Model 1: Wi-Fi Throughput Characterization

Data: Pw: Wi-Fi Tx power; Gw: channel gainof Wi-Fi link; Pl: LTE Tx power; Gwl:channel gain(LTE AP, Wi-Fi UE); N0:noise power; Ec: channel energy at theWi-Fi (LTE interference + N0).

Parameter: λc: Wi-Fi CCA thresholdOutput : Rw: Wi-Fi throughputif No LTE then

Rw = αwB log2

(1 + βw

PwGwN0

).

else When LTE is presentif Ec > λc then

No Wi-Fi transmission with Rw = 0else

Rw = αwB log2

(1 + βw

PwGwPlGwl +N0

).

endend

2) Characterization of LTE Throughput: Due toCSMA/CA, Wi-Fi is active for an average ηS fraction oftime (Eq. (3)). Assuming that LTE can instantaneously updateits transmission rate based on the Wi-Fi interference, itsthroughput can be modeled as follows-

Model 2: LTE Throughput Characterization

Data: Pl: LTE Tx power; Gl: channel gain ofLTE link; Pw: Wi-Fi Tx power; Glw:channel gain(Wi-Fi AP,LTE UE); N0:noise power; Ec: channel energy at Wi-Fi(LTE interference + N0);

Parameter: λc: Wi-Fi CCA thresholdOutput : Rl: LTE throughputif No Wi-Fi then

RlnoW = αlB log2

(1 + βl

PlGlN0

).

else When Wi-Fi is presentif Ec > λc then

No Wi-Fi transmission/interference

Rl = RlnoW .

else

Rl = αlB log2

(1 + βl

PlGlPlGlw +N0

).

Using (3) and ηC = 0 (a single Wi-Fi)

Rl = ηERlnoW + ηSRl

endend

Wi-Fi AP Associated

Wi-Fi UEInterfering AP

UE-AP dist

(fixed) = 0.25 m

Inter-AP dist (variable)

range = [1, 20] m

Fig. 3. Experimental scenario to evaluate the throughput performance ofWi-Fi w1 in the presence of interference (LTE/other Wi-Fi/white noise) whenboth w1 and interference operated on the same channel in 2.4 GHz

0 5 10 15 20

0

5

10

15

20

25

Distance[m]

Thr

ough

put[M

bps]

Exp ErrorbarExperimental ThroughputAnalytical Throughput

Fig. 4. Comparative results analytical model and experiments to show theeffect of LTE on the throughput of Wi-Fi 802.11g when distance between LTEeNB and Wi-Fi link is varied.

B. Experimental Validation

In this section, we experimentally validate proposed in-terference characterization models using experiments involv-ing the ORBIT testbed and USRP radio platforms availableat WINLAB [35], [36]. An 802.11g Wi-Fi link is set upusing Atheros AR928X wireless network adapters [37] andan AP implementation with hostapd [38]. For LTE, we useOpenAirInterface, an open-source software implementation,which is fully compliant with 3GPP LTE standard (release8.6) and set in transmission mode 1 (SISO) [39]. Currently,OpenAirInterface is in the development mode for USRP basedplatforms with limited working LTE operation parameters. Dueto limitations in the available setup, we perform experimentsin the 2.4 GHz spectrum. We note that, though channel fadingcharacteristics differ in other spectrum bands, Wi-Fi/LTE coex-istence throughput behavior remains same with appropriatelyscaled distance.

In our experiment, depicted as the scenario shown infigure 3, we study the effect of interference on the Wi-Fi linkw1. For link w1, the distance between the AP and client isfixed at 0.25 m (very close so that the maximum throughputis guaranteed when no interference is present. Experimentally,we observe maximum throughput as 22.2 Mbps). The distancebetween the interfering AP and Wi-Fi AP is varied in the rangeof 1 to 20 m. The throughput of w1 is evaluated under twosources of interference - LTE and Wi-Fi, when both w1 and theinterference AP is operated on the same channel in the 2.4 GHzspectrum band. These experiments are carried in the 20 m-by-20 m ORBIT room in WINLAB, which has an indoor Line-

0 5 10 15 200

5

10

15

20

25

Distance[m]

Thr

ough

put[M

bps]

Wi−FiLTE 5MHzLTE 10MHz

No interference WiFi Throughput

Fig. 5. Comparative results analytical model and experiments to show theeffect of LTE on the throughput of Wi-Fi 802.11g when distance between LTEHeNB (AP) and Wi-Fi link is varied.

TABLE I. NETWORK PARAMETERS OF WI-FI/LTE DEPLOYMENT

Parameter Value Parameter ValueScenario Downlink Tx power 20 dBmSpectrum band 2.4 GHz Channel bandwidth 20 MHzTraffic model Full buffer via saturated UDP flowsAP antenna height 10 m User antenna height 1 mPath loss model 36.7log10(d[m]) + 22.7 + 26log10(frq [GHz])Noise Floor -101 dBm, (-174 dBm thermal noise/Hz)Channel No shadow/Rayleigh fadingWi-Fi 802.11n: SISOLTE FDD, Tx mode-1 (SISO)

of-Sight (LoS) environment. For each source of interference,Wi-Fi throughput is averaged over 15 sets of experiments withvariable source locations and trajectories between interferenceAP and w1.

In the first experiment, we perform a comparison studyto evaluate the effect of LTE interference on w1, observedby experiments and computed by interference characterizationmodel. In this case, LTE signal is lightly loaded on 5 MHz ofbandwidth mainly consist of control signals. Thus, the impactof such LTE signal over the Wi-Fi band is equivalent to thelow power LTE transmission. Thus, we incorporate these LTEparameters in our analytical model. As shown in figure 4, weobserve that both experimental and analytical values matchthe trend very closely, though with some discrepancies. Thesediscrepancies are mainly due to the fixed indoor experiment en-vironment and lack of a large number of experimental data sets.Additionally, we note that even with the LTE control signal(without any scheduled LTE data transmission), performanceof Wi-Fi gets impacted drastically.

In the next set of experiments, we study the throughputof a single Wi-Fi link in the presence of different sources ofinterference - (1) Wi-Fi, (2) LTE operating at 5 MHz, and (3)LTE operating at 10 MHz, evaluating each case individually.For this part, full-band occupied LTE is considered withthe maximum power transmission of 100 mW. As shown infigure 5, when the Wi-Fi link operates in the presence of otherWi-Fi links, they share channel according to the CSMA/CAprotocol and throughput is reduced approximately by half.In the both the cases of LTE operating at 5 and 10 MHz,due to lack of coordination, Wi-Fi throughput gets impactedby maximum upto 90% compared to no interference Wi-Fithroughput and 20 − 80% compared to Wi-Fi thorughput in

(0,0)

UEiInterfering APj Interfering APj Associated APi

+x-axis

dA

-|dI| +|dI|

-x-axis

Fig. 6. Experimental scenario to evaluate the throughput performance ofWi-Fi w1 in the presence of interference (LTE/other Wi-Fi/white noise) whenboth w1 and interference operated on the same channel in 2.4 GHz

the presence of other Wi-Fi link. These results indicate asignificant effect of inter-network interference on throughput inthe baseline case without any coordination between networks.

C. Motivational Example

We extend our interference model to complex scenarios in-volving co-channel deployment of a single link Wi-Fi and LTEfor the detailed performance evaluation. As shown in figure 6,UEi, associated APi and interfering APj , i, j ∈ {w, l}, i 6= j,are deployed in a horizontal alignment. The distance, dA,between UEi and APi is varied between 0 and 100 m. Ateach value of dA, the distance between UEi and APj is variedin the range of −100 to 100 m. Assuming UEi is located atthe origin (0, 0), if APj is located on the negative X-axis thenthe distance is denoted as −dI , otherwise as +dI , where dI isan Euclidean norm ‖UEi,APj‖. In the shared band operationof Wi-Fi and LTE, due to the CCA sensing mechanism at theWi-Fi node, the distance between Wi-Fi and LTE APs (underno shadow fading effect in this study) decides the transmissionor shutting off of Wi-Fi. Thus, the above distance convention isadopted to embed the effect of distance between APi and APj .Simulation parameters for this set of simulations are given inTable I.

Figure 7 shows the Wi-Fi performance in the presenceof LTE interference. As shown in figure 7(a), the Wi-Fithroughput is drastically deteriorated in the co-channel LTEoperation, leading to zero throughput for 80% of the casesand an average 91% of throughput degradation compared tostandalone operation of Wi-Fi. Such degradation is explainedby figure 7(b). Region CCA-busy shows the shutting off of theWi-Fi AP due to the CCA mechanism, where high energy issensed in the Wi-Fi band. This region corresponds to caseswhen Wi-Fi and LTE APs are within ∼ 20m of each other.In the low SINR region, the Wi-Fi link does not satisfythe minimum SINR requirement for data transmission, thusthe Wi-Fi throughput is zero. High SINR depicts the datatransmission region that satisfies SINR and CCA requirementsand throughput is varied based on variable data rate/SINR.

On the other hand, figure 8 depicts the LTE throughput inthe presence of Wi-Fi interference. LTE throughput is observedto be zero in the low SINR regions, which is 45% of the overallarea and the average throughput degradation is 65% comparedto the standalone LTE operation. Under identical networkparameters, overall performance degradation for LTE is muchlower compared to that of Wi-Fi in the previous example. The

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

0

10

20

30

40

50

60

(a) A heat map of Wi-Fi throughput (Mbps)

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

High SINR Low SINR CCA-Busy

(b) High SINR: non-zero throughput, Low SINR:SINR < minimum SINR requirement, CCA-busy:

shutting off of Wi-Fi (channel sensed as busy)

Fig. 7. Wi-Fi performance as a function of distance(Wi-Fi AP, associatedWi-Fi UE) dA and distance(Interfering LTE AP, Wi-Fi UE) dI

reasoning for such a behavior discrepancy is explained withrespect to figure 8(b) and the Wi-Fi CCA mechanism. In theCCA-busy region, Wi-Fi operation is shut off and LTE operatesas if no Wi-Fi is present. In both LTE and the previous Wi-Fi examples, low SINR represents the hidden node problemwhere two APs do not detect each other’s presence and datatransmission at an UE suffers greatly.

IV. SYSTEM ARCHITECTURE

In this section, we describe an architecture for coordinatingbetween multiple heterogeneous networks to improve spectrumutilization and facilitate co-existence [11]. Figure 9 shows theproposed system, which is built on the principles of a SoftwareDefined Networking (SDN) architecture to support logically-centralized dynamic spectrum management involving multipleautonomous networks. The basic design goal of this architec-ture is to support the seamless communication and informa-tion dissemination required for coordination of heterogeneousnetworks. The system consists of two-tiered controllers: theGlobal Controller (GC) and Regional Controllers (RC), whichare mainly responsible for the control plane of the architecture.The GC, owned by any neutral/authorized organization, is themain decision making entity, which acquires and processes

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

0

10

20

30

40

50

60

(a) A heat map of LTE throughput (Mbps)

dA[m]

20 40 60 80 100d

I[m]

-100

-50

0

50

100

High SINR Low SINR CCA-Busy

(b) High SINR: non-zero throughput, Low SINR:SINR < minimum SINR requirement, CCA-busy:

shutting off of Wi-Fi (channel sensed as busy)

Fig. 8. LTE performance as a function of distance(LTE AP, associated LTEUE) dA and distance(Interfering Wi-Fi AP, LTE UE) dI

network state information and controls the flow of informationbetween RCs and databases based on authentication and otherregulatory policies. Decisions at the GC are based on differentnetwork modules, such as radio coverage maps, coordinationalgorithms, policy and network evaluation matrices. The RCsare limited to network management of specific geographicregions and the GC ensures that RCs have acquired localvisibility needed for radio resource allocation at wirelessdevices. A Local Agent (LA) is a local controller, co-locatedwith an access point or base-station. It receives frequentspectrum usage updates from wireless clients, such as devicelocation, frequency band, duty cycle, power level, and datarate. The signaling between RC and LAs are event-driven,which occurs in scenarios like the non-fulfillment of quality-of-service (QoS) requirements at wireless devices, request-for-update from an RC and radio access parameter updates from anRC. The key feature of this architecture is that the frequencyof signaling between the different network entities is less inhigher tiers compared to lower tiers. RCs only control theregional messages and only wide-area network level signallingprotocols are handled at the higher level, GC. Furthermore, thisarchitecture allows adaptive coordination algorithms based onthe geographic area and change in wireless device density and

Tier 1 – Regional Controller (RC)

Mobility

Mgmt.

Wireless

access control

QoS

controlRRM

Inter-Network

Coordination

Tier 2- Global Controller (GC)

BS/AP Control Network

Data Plane

Internet

LA

LALA

Tier 1 – Regional Controller (RC)

Data Apps Control AppsData Apps Control Apps

Mobility

Management

Radio

Coverage

Map

Policy

Negotiation

Network

Utility

Evaluation

Tx Power

ControlCh Selection

Load

Balancing Coordination

Algorithm

Repository

Local Agent (LA)

Datapath Per device parameters-

Unique ID (Type)

Tx Power, Frequency, BW,

Duty cycle

Location: (x,y) coordinates

Radio Map

Control/

Event Handling

Local Performance

Evaluation (QoS,

Client throughput)

Radio

Interference

Model

Fig. 9. SDN based achitecture for inter-network cooperation on radio resource management

traffic patterns. We use this architecture to exchange controlmessages required for the optimization model, as described in§VI.

V. SYSTEM MODEL

As seen in the previous section, when two (or more) APsof different Wi-Fi and LTE networks are deployed in thesame spectrum band, APs can cause severe interference toone another. In order to alleviate inter-network interference,we propose joint coordination based on (1) power, and (2)time division channel access optimization. We assume thatboth LTE and Wi-Fi share a single spectrum channel andoperate on the same amount of bandwidth. We also note thatclients associated to one AP cannot join other Wi-Fi or LTEAPs. This is a typical scenario when multiple autonomousoperators deploy APs in the shared band. With the help ofthe proposed SDN architecture, power level and time divisionchannel access parameters are forwarded to each networkbased on the throughput requirement at each UE. To the best ofour knowledge, such an optimization framework has not yetreceived much attention for the coordination between Wi-Fiand LTE networks.

We consider a system with N Wi-Fi and M LTE networks.W and L denote the sets of Wi-Fi and LTE links, respec-tively. We maintain all assumptions, definitions and notationsas described in Section III-A. For notational simplicity, weredefine Ri = αiB log2(1 + βiSi), i ∈ {W,L} as Ri =αi log2(1 + βiSi), where constant parameter B is absorbedwith αi. Additional notation are summarized in Table II.

In order to account for the co-channel deployment ofmultiple Wi-Fi networks, we assume that time is sharedequally when multiple Wi-Fi APs are within CSMA rangedue to the Wi-Fi MAC layer. We denote the set of Wi-FiAPs within the CSMA range of APi, i ∈ {W} as Ma

i andthose outside of carrier sense but within interference range asM bi . When APi shares the channel with |Ma

i | other APs, itsshare of the channel access time get reduced to approximately1/(1 + |Ma

i |). Furthermore, M bi signifies a set of potential

TABLE II. DEFINITION OF NOTATIONS

Notation Definitionw, l indices for Wi-Fi and LTE network, respectivelyW the set of Wi-Fi linksL the set of LTE linksPi Transmission power of i-th AP, where i ∈ {W,L}Gij Channel gain between nodes i and jRi Throughput at i-th link, where i ∈ {W,L}Si SINR at i-th link, where i ∈ {W,L}B Channel BandwidthN0 Noise levelαi, βi Efficiency parameters of system i ∈ {W,L}Ma

i Set of Wi-Fi APs in the CSMA range of AP i ∈ {W}Mb

i Set of Wi-Fi APs in the interference range of AP i ∈ {W}ζ Hidden node interference parameterη Fraction of channel access time for network i, i ∈ {w, l} when

j, j ∈ {w, l}, j 6= i, access channel for 1− η fraction of timeλc threshold of Wi-Fi CCA energy detection mechanism

hidden nodes for APi,∀i. To capture the effect of hidden nodeinterference from APs in the interference range, parameter ζ isintroduced which lowers the channel access time and thus, thethroughput. Average reduction in channel access time at APiis 1/(1 + ζ|M b

i |) where ζ falls in the range [0.2, 0.6] [40].Therefore, the effective Wi-Fi throughput can be written as

Ri = aibiαw log2(1 + βwSi), i ∈ W,

with ai =1

1 + |Mai |

and bi =1

1 + ζ|M bi |.

(4)

SINR of Wi-Fi link, i, i ∈ W , in the presence of LTE and noLTE is described as

Si =

PiGiiN0

, if no LTE;

PiGii∑j∈L PjGij +N0

, if LTE,(5)

where the term∑j∈L PjGij is the interference from all LTE

networks at a Wi-Fi link i.

The throughput definition of the LTE link i, i ∈ L remainsthe same as in Section III-A:

Ri = αl log2(1 + βlSi), i ∈ L.

The SINR of the LTE link, i,∀i, in the presence of Wi-Fi andno Wi-Fi is described as

Si =

PiGii∑

j∈L,j 6=i PjGij +N0, if no Wi-Fi;

PiGii∑j∈L,j 6=i PjGij +

∑k∈W akPkGik +N0

, if Wi-Fi,

(6)where terms

∑j∈L,j 6=i PjGij and

∑k∈W akPkGik signifies

the interference contribution from other LTE links and Wi-Filinks, (assuming all links in W are active). For the k-th Wi-Filink, ∀k, the interference is reduced by a factor ak to capturethe fact that the k-th Wi-Fi is active approximately for onlyak fraction of time due to the CSMA/CA protocol at Wi-Fi.

For a given model, inter-network coordination is employedto assure a minimum throughput requirement, thus the guaran-teed availability of the requested service at each UE. For thispurpose, we have implemented our optimization in two stagesas described in following subsections.

VI. COORDINATION VIA JOINT OPTIMIZATION

A. Joint Power Control Optimization

Here, the objective is to optimize the set of transmissionpower Pi, i ∈ {W,L} at Wi-Fi and LTE APs, which maxi-mizes the aggregated Wi-Fi+LTE throughput. Conventionally,LTE supports the power control in the cellular network. Bydefault, commercially available Wi-Fi APs/routers are set tomaximum level [41]. But adaptive power selection capabilityis incorporated in available 802.11a/g/n Wi-Fi drivers, eventhough it is not invoked very often. Under the SDN architec-ture, transmission power level can be made programmable tocontrol the influence of interference from any AP at neighbor-ing radio devices based on the spectrum parameters [42].

For the maximization of aggregated throughput, we pro-pose a geometric programming (GP) based power control [18].For the problem formulation, throughput, given by Eq. 2, canapproximated as

Ri = αi log2(βiSi), i ∈ {W,L}. (7)

This equation is valid when βiSi is much higher than 1. In ourcase, this approximation is reasonable considering minimumSINR requirements for data transmission at both Wi-Fi andLTE. The aggregate throughput of the WiFi+LTE network is

R =∑i∈W

aibiαw log2(βwSi) +∑j∈L

αl log2(βlSj)

= log2

(∏i∈W

(βwSi)aibiαw

)∏j∈L

(βlSi)αl

. (8)

In the coordinated framework, it is assumed that WiFiparameters ai and bi are updated periodically. Thus, these areconsidered as constant parameters in the formulation. Also,αi, βi, i ∈ {w, l} are constant in the network. Therefore, aggre-gate throughput maximization is equivalent to maximization ofa product of SINR at both WiFi and LTE links. Power control

optimization formulation is given by:

maximize

(∏i∈W

(βwSi)aibiαw

)∏j∈L

(βlSi)αl

subject to Ri ≥ Ri,min, i ∈ W,

Ri ≥ Ri,min, i ∈ L,∑k∈Mb

i

PkGik +∑j∈L

PjGij +N0 < λc, i ∈W,

0 < Pi ≤ Pmax, i ∈ W,

0 < Pi ≤ Pmax, i ∈ L.(9)

Here, the first and second constraints are equivalent to Si ≥Si,min,∀i which ensures that SINR at each link achieves aminimum SINR requirement, thus leading to non-zero through-put at the UE. The third constraint assures that channel energyat a WiFi (LTE interference + interference from WiFis in theinterference zone + noise power) is below the clear channelassessment threshold λc, thus WiFi is not shut off. The fourthand fifth constraints follow the transmission power limits ateach link. Unlike past power control optimization formulationsfor cellular networks, WiFi-LTE coexistence requires to meetthe SINR requirement at a WiFi UE and, additionally, CCAthreshold at a WiFi AP.

For multiple Wi-Fi and LTE links, to ensure the feasibilityof the problem where all constrains are not satisfied, notablyfor WiFi links, we relax the minimum data requirement con-straint for LTE links. In our case, we reduce the minimum datarequirement to zero. This is equivalent to shutting off certainLTE links which cause undue interference to neighboring WiFidevices.

B. Joint Time Division Channel Access Optimization

The relaxation of minimum throughput constraint in thejoint power control optimization leads to throughput depri-vation at some LTE links. Thus, joint power control is notsufficient when system demands to have non-zero throughputat each UE. In such cases, we propose a time divisionchannel access optimization framework where network of eachRAT take turns to access the channel. Assuming networki, i ∈ {w, l} access the channel for η, eta ∈ [0, 1], fraction oftime, network j, j ∈ {w, l}, j 6= i, holds back the transmissionand thus no interference occurs at i from j. For remaining 1−ηfraction of time, j access the channel without any interferencefrom i. This proposed approach can be seen as a subset ofpower assignment problem, where power levels at APs ofnetwork i, i ∈ {w, l}, is set to zero in their respective timeslots. The implementation of the protocol is out of scope ofthis paper.

In this approach, our objective is to optimize η, thetime division of channel access, such that it maximizes theminimum throughput across both WiFi and LTE networks. Wepropose the optimization in two steps -

1) Power control optimization across network of same RAT:Based on the GP-formulation, the transmission power of theAPs across the same network i, i ∈ {w, l}, is optimized for

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

0

10

20

30

40

50

60

(a) A heat map of Wi-Fi throughput when jointpower Coordination (Mbps)

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

Infeasible Feasible

(b) Feasibility region of joint powerCoordination

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

0

10

20

30

40

50

60

(c) A heat map of Wi-Fi throughput when timedivision channel access coordination (Mbps)

Fig. 10. Wi-Fi performance under joint Wi-Fi and LTE coordination (dA: dist(Wi-Fi AP, associated UE), dI : dist(Interfering LTE AP, Wi-Fi UE))

Wi-Fi and LTE, respectively, as

maximize∑i∈W

Ri

subject to Ri ≥ Ri,min, i ∈ W0 ≤ Pi ≤ Pmax, i ∈ W,∑k∈Mb

i

PkGik +N0 < λc, i ∈ W.

(10)

andmaximize

∑i∈L

Ri

subject to Ri ≥ Ri,min, i ∈ L0 ≤ Pi ≤ Pmax, i ∈ L.

(11)

Here, the objective function is equivalent to maximizing theproduct of SINRs at the networks i, i ∈ {w, l}. The first andsecond constraints ensure that we meet the minimum SINRand transmission power limits requirements at all links of i.In this formulation, SINR at WiFi and LTE respectively givenas

Si =PiGiiN0

, i ∈ W,

Si =PiGii∑

j∈L,j 6=i PjGij +N0, i ∈ L.

which are first cases in equations (5) and (6), respectively.

2) Joint time division channel access optimization: Thisis the joint optimization across both WiFi and LTE networkswhich is formulated using max-min fairness optimization asgiven below

maximize min (ηRi∈W , (1− η)Rj∈L)subject to 0 ≤ η ≤ 1.

(12)

Here, throughput values at all WiFi and LTE nodes areconsidered as a constant, which is the output of the previousstep. Time division channel access parameter η is optimizedso that it maximizes the minimum throughput across all UEs.

VII. EVALUATION OF JOINT COORDINATION

A. Single Link Co-channel Deployment

We begin with the motivational example of co-channeldeployment of one Wi-Fi and one LTE links, as described in§ III-C. Figure 10 shows the heatmap of improved throughputof Wi-Fi link, when joint Wi-Fi and LTE coordination isprovided in comparison with the throughput with no coor-dination as shown in figure 7 . Similarly, figure 11 showsthe heatmap of improved throughput of LTE link, when jointcoordination is provided in comparison with the throughputwith no coordination, as shown in figure 8.

For both the figures 10 and 11, in their respective scenarios,though joint power control improves the overall throughputfor most of topological scenarios (see Figure (a) of 10 and11), it is not an adequate solution for topological combinationmarked by infeasible region as given in figure (b) of 10 and11. The infeasible region signifies the failure to attain the CCAthreshold at Wi-Fi AP and link SINR requirement when theUE and interfering AP are very close to each other. When weapply time division channel access optimization for a givenscenario, we do not observe any infeasible region, in factoptimization achieves almost equal and fair throughput at bothWi-Fi and LTE link, as shown in figure (c) of 10 and 11. On thedownside, this optimization does not consider cases when Wi-Fi and LTE links can operate simultaneously without causingsevere interference to each other. In such cases, throughput atboth Wi-Fi and LTE get degraded.

Figure 12 summarizes the performance of Wi-Fi and LTElinks in terms of 10 percentile and mean throughput. We notethat the 10 percentile throughput of both Wi-Fi and LTE isincreased to 15 − 20 Mbps for time division coordinationcompared to ∼ zero throughput for no and power coordination.We observe 200% and 350% Wi-Fi mean throughput gainsdue to power and time division channel access, respectively,compared to no coordination. For LTE, throughput gainsfor both of these coordination is ∼ 25 − 30%. It appearsthat time division channel access coordination does not offerany additional advantage to LTE in comparison with powercoordination. But it brings the throughput fairness between

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

0

10

20

30

40

50

60

(a) A heat map of LTE throughput when jointpower Coordination (Mbps)

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

Infeasible Feasible

(b) Feasibility region of joint powerCoordination

dA[m]

20 40 60 80 100

dI[m

]

-100

-50

0

50

100

0

10

20

30

40

50

60

(c) A heat map of LTE throughput when timedivision channel access coordination (Mbps)

Fig. 11. LTE performance under joint Wi-Fi and LTE coordination (dA: dist(LTE AP, associated UE), dI : dist(Interfering Wi-Fi AP, LTE UE))

WiFi LTE0

5

10

15

20

25

Thro

ughp

ut [M

bps]

WiFi LTE0

5

10

15

20

25

30

Thro

ughp

ut [M

bps]

Mean throughput

No Interference Pwr Control TimeDivCh Access

10 percentile throughput

Fig. 12. 10 percentile and mean LTE throughput for a single link Wi-Fi andLTE co-channel deployment

Wi-Fi and LTE which is required for the co-existence in theshared band.

B. Multiple Links Co-channel Deployment

Multiple overlapping Wi-Fi and LTE links are randomlydeployed in 200-by-200 sq. meter area which depicts thetypical deployment in residential or urban hotspot. The numberof APs of each Wi-Fi and LTE networks are varied between2 to 10 where number of Wi-Fi and LTE links are assumedto be equal. For the simplicity purpose, we assume that onlysingle client is connected at each AP and their associationis predefined. The given formulation can be extended formultiple client scenarios. In the simulations, the carrier senseand interference range for Wi-Fi devices are set to 150 metersand 210 meters, respectively. The hidden node interferenceparameter is set to 0.25.

Figure 13(a) and 13(a) show the percentile and meanthroughput values of Wi-Fi and LTE links, respectively, forwhen number of links for each Wi-Fi and LTE networks is setat N = {2, 5, 10}. The throughput performance is averagedover 10 different deployment topologies of Wi-Fi and LTE

links. From figure 13(a), it is clear that 10 percentile Wi-FiUEs get throughput starved due to LTE interference with nocoordination. This is consistent with results from single linksimulations. With coordination, both joint power control andtime division channel access, we achieve a large improvementin the 10 percentile throughput. Joint power control improvesmean Wi-Fi throughput by 15-20% for all N . On the otherhand, time division channel access achieves throughput gain(40-60%) only at higher values of N = {5, 10}.

Throughput performance of LTE, on the other hand, getdeteriorates in the presence of coordination compared to whenno coordination is provided. This comes from the fact that,in case of no coordination, LTE causes undue impact atWi-Fi which makes them to hold off data transmission andLTE experiences no Wi-Fi interference. The joint coordinationbetween Wi-Fi and LTE networks brings the notion of fairnessin the system and allocates spectrum resources to sufferedWi-Fi links. In the joint power control optimization, thoughcertain LTE links (maximum 1 link for N = 10) have to bedropped from network with zero throughput, the overall meanthroughput is greater than 150 to 400% than Wi-Fi throughput.

We observe that for small numbers of Wi-Fi links, jointtime division channel access degrades the performance of bothWi-Fi and LTE. But as number of links grows, coordinatedoptimization results in allocation of orthogonal resources (e.g.separate channels) gives greater benefit than full sharing ofthe same spectrum space, as is the case for power controloptimization.

VIII. CONCLUSION

This paper investigates inter-system interference in sharedspectrum scenarios with both Wi-Fi and LTE operating inthe same band. An analytical model has been developed forevaluation of the performance and the model has been partiallyverified with experimental data. The results show that sig-nificant performance degradation results from uncoordinatedoperation of Wi-Fi and LTE in the same band. To address thisproblem, we further presented an architecture for coordinationbetween heterogeneous networks, with a specific focus on

2 5 100

2

4

6

8

10T

hrou

ghpu

t [M

bps]

10 percentile throughput

2 5 100

5

10

15

20

25

Thr

ough

put [

Mbp

s]

Mean throughput

No Interference Pwr Control TimeDivCh Access

(a) 10 percentile and mean Wi-Fi throughput for N = {2, 5, 10}

2 5 100

5

10

15

20

Thro

ughput [M

bps]

10 percentile throughput

2 5 100

10

20

30

40

50

Thro

ughput [M

bps]

Mean throughput

No Interference Pwr Control TimeDivCh Access

(b) 10 percentile and mean LTE throughput for N = {2, 5, 10}

Fig. 13. Multi-link throughput performance under power control and timedevision channel access optimization. N = no. of LTE links = no. of Wi-Filinks.

LTE-U and Wi-Fi, to cooperate and coexist in the same area.This framework is used to exchange information between thetwo networks for a logically centralized optimization approachthat improves the aggregate throughput of the network. Ourresults show that, with joint power control and time divisionmultiplexing, the aggregate throughput of each of the networksbecomes comparable, thus realizing fair access to the spectrum.In future work, we plan to extend our analytical model andoptimization framework to study realistic user applications forwhich full buffer traffic conditions can not be assumed. Wefurther plan to extend the optimization framework to exploitthe frequency diversity for joint coordination of Wi-Fi andLTE.

Acknowledgment: Research is supported by NSF EARSprogram- grant CNS-1247764.

REFERENCES

[1] “Cisco Visual Networking Index: global mobile data traffic forecastupdate, 2013-2018,” Cisco White Paper, Feb 2014.

[2] Samsung, “Vision and key features for 5th generation (5G) cellular,”2014, http://tinyurl.com/lo5gg53.

[3] Shahram Giri, “Exploring 5G: Performance targets, technologies andtimelines,” 2014, http://tinyurl.com/mznelg8.

[4] June Kamran Etemad, LTE World Summit, “Improving cell capacityof 5G systems through opportunistic use of unlicensed and sharedspectrum,” 2014, http://tinyurl.com/k4bwjw4.

[5] Agilent Technologies Moray Rumney, “Taking 5G from vision toreality,” 2014, The 6th Future of Wireless International Conference,http://tinyurl.com/o8hug8j.

[6] RP-140060, “Summary of a workshop on LTE in Unlicensed Spectrum,”3GPP TSG-RAN Meeting 63, 2014.

[7] Federal Communications Commission, “Enabling innovative small celluse in 3.5 GHz band NPRM and order,” FCC 12-148, 2012.

[8] Intel Corporation Ali Sadri, “mmWave technology evolution fromWiGig to 5G small cells,” 2013, http://tinyurl.com/lejlgfg.

[9] S.J. Shellhammer, A.K. Sadek, and Wenyi Zhang, “Technical challengesfor cognitive radio in the TV white space spectrum,” in InformationTheory and Applications Workshop, 2009, Feb 2009, pp. 323–333.

[10] S. Sagari, I. Seskar, and D. Raychaudhuri, “Modeling the coexistence ofLTE and WiFi heterogeneous networks in dense deployment scenarios,”accepted at IEEE ICC workshop 2015 on LTE in Unlicensed Bands:Potentials and Challenges.

[11] D. Raychaudhuri and A. Baid, “NASCOR: Network assisted spectrumcoordination service for coexistence between heterogeneous radio sys-tems,” IEICE Trans. Commun., vol. 97, no. 2, pp. 251–260, 2014.

[12] G. Nychis, C. Tsourakakis, S. Seshan, and P. Steenkiste, “Centralized,measurement-based, spectrum management for environments with het-erogeneous wireless networks,” in Dynamic Spectrum Access Networks(DYSPAN), 2014 IEEE International Symposium on, April 2014, pp.303–314.

[13] A. Baid, S. Mathur, I. Seskar, S. Paul, A. Das, and D. Raychaudhuri,“Spectrum MRI: Towards diagnosis of multi-radio interference inthe unlicensed band,” in Wireless Communications and NetworkingConference (WCNC), 2011 IEEE, March 2011, pp. 534–539.

[14] Xiangpeng Jing and Dipankar Raychaudhuri, “Spectrum co-existenceof IEEE 802.11b and 802.16a networks using reactive and proactiveetiquette policies,” Mobile Networks and Applications, vol. 11, no. 4,pp. 539–554, 2006.

[15] O. Ileri, D. Samardzija, and N.B. Mandayam, “Demand responsivepricing and competitive spectrum allocation via a spectrum server,” inNew Frontiers in Dynamic Spectrum Access Networks, 2005. DySPAN2005. 2005 First IEEE International Symposium on, Nov 2005, pp.194–202.

[16] A. Baid and D. Raychaudhuri, “Understanding channel selectiondynamics in dense Wi-Fi networks,” Communications Magazine, IEEE,vol. 53, no. 1, pp. 110–117, January 2015.

[17] S. S. Sagari, “Coexistence of LTE and WiFi heterogeneous networksvia inter network coordination,” in Proceedings of the 2014 Workshopon PhD Forum, New York, NY, USA, 2014, PhD forum ’14, pp. 1–2,ACM.

[18] Mung Chiang, Chee Wei Tan, D.P. Palomar, D. O’Neill, and D. Julian,“Power control by geometric programming,” Wireless Communications,IEEE Transactions on, vol. 6, no. 7, pp. 2640–2651, July 2007.

[19] IEEE, “Part 11: Wireless LAN medium access control (MAC) andphysical layer (PHY) specifications,” 2012.

[20] Kyle Jamieson, Bret Hull, Allen Miu, and Hari Balakrishnan, “Under-standing the real-world performance of carrier sense,” in Proceedingsof the 2005 ACM SIGCOMM Workshop on Experimental Approachesto Wireless Network Design and Analysis, New York, NY, USA, 2005,E-WIND ’05, pp. 52–57, ACM.

[21] G. Bianchi, “Performance analysis of the IEEE 802.11 distributed coor-dination function,” IEEE Journal on Selected Areas in Communications,vol. 18, no. 3, pp. 535 –547, 2000.

[22] Chaves F. S., Cavalcante A. M., Almeida E. P. L., Abinader F. M. Jr.,Vieira R. D., Choudhury S., and Doppler K., “LTE/Wi-Fi coexistence:Challenges and mechanisms,” 2013, XXXI SIMPOSIO BRASILEIRODE TELECOMUNICACOES - SBrT2013.

[23] F.M. Abinader, E.P.L. Almeida, F.S. Chaves, A.M. Cavalcante, R.D.Vieira, R.C.D. Paiva, A.M. Sobrinho, S. Choudhury, E. Tuomaala,K. Doppler, and V.A. Sousa, “Enabling the coexistence of LTE andWi-Fi in unlicensed bands,” Communications Magazine, IEEE, vol. 52,no. 11, pp. 54–61, Nov 2014.

[24] Jin Jin and Baochun Li, “Cooperative resource management in cognitiveWiMAX with femto cells,” in INFOCOM, 2010 Proceedings IEEE,March 2010, pp. 1–9.

[25] A.E. Leu, B.L. Mark, and M.A. McHenry, “A framework for cognitiveWiMAX with frequency agility,” Proceedings of the IEEE, vol. 97, no.4, pp. 755–773, April 2009.

[26] E. Almeida, A.M. Cavalcante, R.C.D. Paiva, F.S. Chaves, F.M. Abi-nader, R.D. Vieira, S. Choudhury, E. Tuomaala, and K. Doppler,“Enabling LTE/WiFi coexistence by LTE blank subframe allocation,” in2013 IEEE International Conference on Communications (ICC), June2013, pp. 5083–5088.

[27] S.K. Baghel, M.A. Ingale, and G. Goyal, “Coexistence possibilities ofLTE with ISM technologies and GNSS,” in Communications (NCC),2011 National Conference on, Jan 2011, pp. 1–5.

[28] A.M. Cavalcante, E. Almeida, R.D. Vieira, F. Chaves, R.C.D. Paiva,F. Abinader, S. Choudhury, E. Tuomaala, and K. Doppler, “Performanceevaluation of LTE and Wi-Fi coexistence in unlicensed bands,” in 2013IEEE 77th Vehicular Technology Conference (VTC Spring), June 2013,pp. 1–6.

[29] Inc Qualcomm Technologies, “LTE in unlicensed spectrum: Harmo-nious coexistence with Wi-Fi,” 2014, White paper.

[30] R. Ratasuk, M.A. Uusitalo, N. Mangalvedhe, A. Sorri, S. Iraji, C. Wijt-ing, and A. Ghosh, “License-exempt LTE deployment in heterogeneousnetwork,” in Wireless Communication Systems (ISWCS), 2012 Interna-tional Symposium on, Aug 2012, pp. 246–250.

[31] Feilu Liu, E. Bala, E. Erkip, and Rui Yang, “A framework for femtocellsto access both licensed and unlicensed bands,” in Modeling andOptimization in Mobile, Ad Hoc and Wireless Networks (WiOpt), 2011International Symposium on, May 2011, pp. 407–411.

[32] T. Nihtila, V. Tykhomyrov, O. Alanen, M.A. Uusitalo, A. Sorri,M. Moisio, S. Iraji, R. Ratasuk, and N. Mangalvedhe, “System perfor-mance of LTE and IEEE 802.11 coexisting on a shared frequency band,”

in Wireless Communications and Networking Conference (WCNC),2013 IEEE, April 2013, pp. 1038–1043.

[33] F.S. Chaves, E.P.L. Almeida, R.D. Vieira, A.M. Cavalcante, F.M.Abinader, S. Choudhury, and K. Doppler, “LTE UL power controlfor the improvement of LTE/Wi-Fi coexistence,” in 2013 IEEE 78thVehicular Technology Conference (VTC Fall), Sept 2013, pp. 1–6.

[34] R.C.D. Paiva, P. Papadimitriou, and S. Choudhury, “A physical layerframework for interference analysis of LTE and Wi-Fi operating in thesame band,” in 2013 Asilomar Conference on Signals, Systems andComputers, Nov 2013, pp. 1204–1209.

[35] D. Raychaudhuri, I. Seskar, M. Ott, S. Ganu, K. Ramachandran,H. Kremo, R. Siracusa, H. Liu, and M. Singh, “Overview of the ORBITradio grid testbed for evaluation of next-generation wireless networkprotocols,” in Wireless Communications and Networking Conference,2005 IEEE, March 2005, vol. 3, pp. 1664–1669 Vol. 3.

[36] Ettus Research, “USRP B200 and B210,” http://tinyurl.com/k7w6zh2.[37] Linux Wireless ath9k, ,” https://wireless.wiki.kernel.org/en/users/drivers/ath9k.[38] “hostapd: IEEE 802.11 AP, IEEE 802.1X/WPA/WPA2/EAP/RADIUS

authenticator,” http://w1.fi/hostapd/.[39] Navid Nikaein, Mahesh K. Marina, Saravana Manickam, Alex Dawson,

Raymond Knopp, and Christian Bonnet, “OpenAirInterface: A flexibleplatform for 5G research,” SIGCOMM Comput. Commun. Rev., vol. 44,no. 5, pp. 33–38, Oct. 2014.

[40] A. Baid, M. Schapira, I. Seskar, J. Rexford, and D. Raychaudhuri,“Network cooperation for client-AP association optimization,” in IEEEWiOpt, 2012, pp. 431 –436.

[41] Linux Wireless, ,” https://wireless.wiki.kernel.org/en/users/documentation/iw.[42] Aditya Gudipati, Daniel Perry, Li Erran Li, and Sachin Katti, “Soft-

RAN: Software defined radio access network,” in Proceedings of theSecond ACM SIGCOMM Workshop on Hot Topics in Software Defined

Networking, New York, NY, USA, 2013, HotSDN ’13, pp. 25–30, ACM.