Peer-to-Peer Netw. Appl.DOI 10.1007/s12083-015-0375-6

Novel power control and collision resolution schemesfor device-to-device discovery

Jongwoo Hong1 · Seungil Park1 · Sunghyun Choi1

Received: 31 December 2014 / Accepted: 24 April 2015© Springer Science+Business Media New York 2015

Abstract Device-to-Device (D2D) communications allowdevices to communicate directly without going throughinfrastructure. It is considered a promising solution toimprove communication performance and network capacityof LTE-Advanced system. In addition, the direct prox-imity communication implies reduced delay and energyconsumption. However, from the perspective of D2D net-works, co-existence of multiple D2D User Equipments(UEs) is a challenging issue because of the proximityinterference management, especially, without a central con-troller in a distributed scheme. In this article, we proposea novel distributed power control and collision resolutionschemes, which can enhance discovery success of proxim-ity devices conforming to LTE-A networks. The simulationresults demonstrate that our proposed schemes can signif-icantly improve the performance of discovery success anddetectable coverage by mitigating D2D interference.

Keywords D2D discovery · Distributed system ·Power control · Collision resolution · LTE networks

� Sunghyun Choischoi@snu.ac.kr

Jongwoo Hongjwhong@mwnl.snu.ac.kr

Seungil Parkspark11@mwnl.snu.ac.kr

1 Department of Electrical and Computer Engineeringand INMC, Seoul National University, Seoul, Korea

1 Introduction

Recently, Device-to-Device (D2D) communication hasattracted much attention as a decent solution to cope withheavy cellular traffic caused by the proliferation of mobiledevices such as smartphones and tablet PCs along withthe increased demands for high data rate services. In thisscenario, allowing D2D User Equipments (UEs) to reusecellular resources can boost up the performance of thenetwork in terms of the system capacity. Also, reducednumber of hops and shorter communication distance viadirect communication imply reduced energy consumption.In addition, D2D communications can help offload cellu-lar traffic and avoid congestion in the cellular networks.In the past, D2D communications did not consider thehelp of external entities (e.g., cellular infrastructure), butrecently network-assistance for D2D communications hasbeen considered [1]. Figure 1 shows the concept of D2Dcommunication as an underlay to LTE networks. In D2Dnetwork, UEs can communicate directly with each otherover the D2D link.

Long Term Evolution (LTE) standard, which has beenproposed and specified by the 3rd Generation PartnershipProject (3GPP), has opened the true fourth generation (4G)cellular technology era. The main goals of LTE are tosupport reduced latency, higher user data rate, improved sys-tem capacity, and increased spectral efficiency. The abovementioned potential benefits of the D2D communicationhas led the 3GPP community to start the standardiza-tion of D2D communication in LTE-Advanced (LTE-A)networks [2, 3].

LTE-A will include D2D communication under thename, Proximity Service (ProSe) in Release 12/13. TheD2D communication in LTE networks is a promising tech-nology that is expected to fulfill future service requirements

Peer-to-Peer Netw. Appl.

Fig. 1 An example of D2D communication in LTE networks

or needs, e.g., public safety, content sharing, social network-ing, and proximity-based advertising in various use casesand scenarios [2, 4].

Even with the above mentioned potential gains, the co-existence of multiple D2D UEs reusing the same resourcewith cellular network is a challenging issue due to the dif-ficulty of interference management and resource allocation.As the demand for D2D communication increases, the con-cern of interference and resource allocation naturally grows.There is a growing interest in D2D interference in cellu-lar networks [5–7]. Meanwhile, resource allocation reusingcellular network is another growing interest [8, 9] and con-tent delivery by D2D networking is as well [10]. However,most recent studies have focused on the D2D communi-cation issues assuming that discovery procedure has beendone already in a way. Therefore, there have been only afew recent studies [11–14] investigating the D2D discoveryissue.

In this article, we study the interference problem occur-ring during D2D discovery, and propose two differentschemes to improve the performance by alleviating thisproblem. The main contributions of this article are summa-rized as follows:

– We propose a novel distributed power control schemewith an effective algorithm which aims to improve theperformance of overall D2D discovery.

– We propose a novel distributed collision resolutionscheme and protocol which can solve the D2D interfer-ence problem occurring during the discovery phase.

– We evaluate the performance of the proposed schemesand algorithms by incorporating the assumptions inrecent 3GPP ProSe technical documents.

The rest of the article is organized as follows. We firstintroduce the background of D2D discovery in Section 2,and present our system model in Section 3. In Section 4,

we present a problem formulation. Then, we describe theproposed power control scheme in Section 5, and providethe proposed contention resolution scheme in Section 6.The performance evaluation of our proposed schemes areprovided in Section 7. Finally, this article concludes inSection 8.

2 Background

2.1 Distributed D2D discovery

D2D discovery is an initial procedure in the D2D com-munication. In this phase, UEs discover the existence ofeach other before starting actual D2D communication. Sincethe discovery procedure is performed under blind circum-stance, we can assume that essential information of thereceiver (e.g., location, channel status, number of receivers)is unknown before the discovery phase.

D2D discovery is categorized into two types: centralizedand distributed. In the centralized scheme, optimal resourceand proper transmission power are allocated by evolvedNode B (eNB). However, this method requires higher sig-naling overhead and complexity for multiple transmitters tobe coordinated. For the above reasons, this study focuseson the distributed scheme which is considered a suitablestrategy under large and dense D2D networks.

2.2 Resource selection

In general, UEs may exploit both downlink and/or uplinkcellular resource for communication. In this work, weassume that D2D UE (D-UE) utilizes uplink resourcesbecause of several reasons as discussed in [15]. To guar-antee the performance of the discovery, dedicated discov-ery period is introduced periodically (e.g., every 10 sec)with reserved resources (e.g., Nf (= 44) Resource Blocks(RBs)1 and Nt (= 64) subframes in 10 MHz LTE sys-tem) [16, 17].

Therefore, a D-UE participating in discovery will selectone discovery resource unit (DRU) among the periodic dis-covery resources, where a DRU consists of two RBs. Anexample of discovery period and DRU resource is presentedin Fig. 2. The D-UE can transmit its discovery signal on itsselected DRU one time and listen to discovery signals fromother D-UEs during the rest of the discovery period. Dur-ing the discovery phase, every D-UE participates only inthe discovery process, while other types of communication(e.g., D2D or cellular) are not allowed. The UE conductinginfrastructure-based communication is denoted by CellularUE (C-UE) in this article.

1One RB consists of 12 subcarriers in the frequency domain.

Peer-to-Peer Netw. Appl.

Fig. 2 D2D discovery resources in LTE networks

D2D resource selection is also categorized into twotypes: sensing-based and random selection. In sensing-based selection, D-UEs select a DRU resource based onthe sensing results of the available discovery resources.Every D-UE assesses all DRU’s received energy level andselects the DRU which has the lowest energy level. Accord-ingly, multiple D-UEs located far away might choose thesame resource. One the other hand, according to the ran-dom selection, each D-UE randomly selects a DRU resourcefor discovery signal transmission. This work focuses on arandom selection for the following reasons.

The sensing-based selection is inefficient when the sens-ing results are outdated quickly, such as under high UEmobility scenario. Furthermore, the sensing results of D-UEs in proximity may be similar, and hence, these resultsmight lead to a collision in the resource usage. For these rea-sons, most vendors participating in 3GPP propose a simplerandom selection scheme as a distributed resource selectionmethod in recent 3GPP standardization [16, 18, 19].

2.3 Collision

Since the resource is randomly selected by each D-UE, ithas to be chosen carefully. When more than one D-UE reusethe same resource in proximity, a collision may occur dueto the simultaneous transmission. Moreover, since there isno centralized coordination or signaling during the discov-ery procedure, those involved with collision event cannot berecognized by D-UEs.

Therefore, these neighboring D-UEs can neither detecteach other nor be detected because of the mutual inter-ference. Moreover, different from cellular network, therecoexist multiple receivers (D-UEs) under D2D networktopology which are potentially exposed to suffer high inter-ference. Due to the interference, the performance of D2Ddiscovery will be severely degraded.

2.4 Motivation

Despite the advantages of the distributed scheme, thefollowing challenges might arise due to the lack of acentral controller (i.e., eNB): interference management,resource allocation, and collision resolution. In order toachieve enhanced discovery performance, mutual interfer-ence caused by collisions should be reduced or avoided.In this regard, there are two main approaches for mitigat-ing interference between D-UEs, namely, power control andcollision resolution scheme.

In general, power control is a useful solution in energysaving and interference reduction. The main focus of thepower control is to mitigate D2D interference, not to reduceenergy consumption in this article. The received inter-ference can be efficiently reduced by transmitter powercontrol. On the other hand, collision resolution schemefor avoiding interference is another approach. To avoidcollision, the resource which is being reused by multipleproximity D-UEs has to be changed. Note that the motiva-tion of this study is to mitigate interference between D-UEswith power control and collision detection scheme in a dis-tributed system to improve the overall performance of theD2D discovery as well.

3 System model

3.1 D2D system

In this section, we give a brief introduction to the sys-tem model of D2D discovery. In our study, we limit ourscope to synchronous D2D discovery, i.e., all D-UEs arein-coverage and time synchronization reference can beobtained from the eNB downlink transmission. This obvi-ously consumes much less energy and discovery time com-pared with asynchronous discovery. For the synchronousdiscovery, every D-UE can be active during predefined dis-covery time, which appear periodically, i.e., D-UE wakes upperiodically to perform discovery procedure (e.g., beacontransmission and reception) using the DRU. After finishingthe periodic discovery, D-UEs begin sleeping until the nextdiscovery period. When a D-UE has discovered a desiredtarget D-UE by receiving a beacon, it can establish a D2Dlink for direct communication.

Peer-to-Peer Netw. Appl.

In a cellular network topology, the receiver of C-UEs’transmission is always an eNB. Accordingly, if multiple C-UEs belonging to different cells reuse a common resourcein cell edge, a C-UE’s signal interferes with a neighboringeNB. On the other hand, in D2D networks, there coexistmultiple D-UEs which can be both transmitter and receiver.Under this topology, emitted signals from different trans-mitting D-UEs will arrive at proximity receiving D-UEs.Note that multiple receivers are potentially exposed to suf-fer high interference by multiple D2D links. Assuming thenumber of K D-UEs in D2D networks, the maximum num-ber of D2D links is K2 − K , which has a polynomialgrowth rate.

3.2 Criteria of discovery success

For the D2D discovery, a D-UE has to choose a DRU.Then, the D-UE can transmit and receive predefined signals,referred to as beacons. Only one beacon will be transmit-ted by each D-UE during the discovery period. By receivingbeacons periodically, a D-UE maintains a list of neighbor-ing D-UEs in order to establish direct communication linkwhen it is needed after discovery process.

The minimum unit of a beacon consist of two RBs, whichcarry 168 (with normal cyclic prefix) OFDM ResourceElements (REs) [20]. Within the beacon, each D-UE canconvey such information as its own identity, requestingservice, and interest.

Since D-UE k can be either transmitter or receiver ata given time, a D-UE is denoted by t (r) when transmit-ting (receiving), where t, r ∈ {1, ..., K}. For the successfuldecoding of a beacon, the Signal-to-Interference-plus-NoiseRatio (SINR) of the received beacon signal should be abovea certain level. The SINR of beacon b from D-UE t at D-UEr is given by

γ (b)r = Ptht,r

∑j �=t Pjhj,r + σ 2

. (1)

where Pt is the transmission power of D-UE t and ht,r

denotes the channel gain between transmitter D-UE t andreceiver D-UE r . The cumulative received interference fromthe other D-UEs is defined as

∑Jj �=t Pjhj,r and the power of

Additive White Gaussian Noise (AWGN) is denoted by σ 2.A device is assumed to be successfully discovered only

if the SINR of a received beacon signal exceeds decodingthreshold γthd . We define δb ∈ {0, 1} as a detection indicatorof beacon b ∈ {1, ..., B}. That is, δb = 1 if beacon b issuccessfully detected, and δb = 0, otherwise. Therefore,

δb ={0, γ (b)

r < γthd ,

1, γ (b)r ≥ γthd .

(2)

In order for D-UE t to be successfully discovered by D-UEr , the transmission power of D-UE t should satisfy

Pt ≥ γthd

(∑

j �=t

Pjhj,r

ht,r

+ σ 2

ht,r

)

. (3)

4 Problem formulation

The objective of conventional communication has been toachieve the maximum link capacity between transmitter andreceiver, assuming a communication link is already estab-lished. Different from the previous approach, the aim ofdiscovery may not be related to transmission rate or linkthroughput since a target link is not established. There-fore, the objective of D-UEs should be how to detect andto be detected by as many devices as possible during thediscovery period.

To deal with this problem, we propose a new metric,called discovery success ratio Sk ∈ [0, 1], which is definedas the ratio of discovered D-UEs to the total number of D-UEs participating (except itself) in discovery. The discoverysuccess ratio of D-UE k is given by

Sk ={

1

K − 1

∑K

b=1,b �=kδb

}

. (4)

The higher this ratio is, the more D-UEs have been dis-covered during the discovery period. To maximize the D2Ddiscovery success ratio, each D-UE has to select its propertransmit power considering mutual interference, and selectbeacon resource without collision. However, due to an initialstep, most essential information for discovery is unknown(e.g., distance of transmitter and receiver, channel gain,number of receivers and reusing resource). Moreover, sincemore than one proximity receiver may coexist, the opti-mal power level of transmitter cannot be determined for aspecific receiver.

Therefore, considering such constraints, the primary goalof D2D discovery is to achieve the maximum discovery suc-cess ratio during the discovery period, denoted by td . Theobjective of D2D discovery problem can be formulated asfollows:

max∞∑

td=1

K∑

k=1

Sk (5)

subject to

0 < P(b)t ≤ Pmax. (6)

Peer-to-Peer Netw. Appl.

5 Power control scheme

5.1 Power control performance

Different from cellular transmission between C-UE andeNB, D2D discovery transmission is broadcast. Due to thebroadcast topology, a transmitter D-UE will periodicallysend the beacon to every neighboring D-UEs. Under thisscenario, transmitted signal might be a desired signal forsome D-UEs, while it might be undesirable interference toother D-UEs due to the collision. By adopting power con-trol scheme, the transmitted interference among D-UEs canbe reduced.

For the purpose of the verification of power control oper-ation, we consider two simulation scenarios as shown inFigs. 3 and 4. This article conducts the following simulationas an example. We assume that there are two D-UEs whichreuse discovery resource in a cell center region. The distancebetween two D-UEs is 200 m (which are marked by reddots) and beacon detection threshold is set to 4.5 dB [21].10,000 D-UEs are placed uniformly within a cell (radius =1000 m). The path loss exponent is assumed to be 4 in thissimulation.

In Scenario 1, the two D-UEs transmit with the samemaximum power level (23 dBm). In Scenario 2, the twoD-UEs transmit the maximum and one-third of the maxi-mum (18 dBm) alternately. That is, D-UE1 transmits with23 dBm and D-UE2 transmits with 18 dBm at the firstdiscovery period, then D-UE1 transmits with 18 dBm andD-UE2 transmits with 23 dBm in the next discovery periodto alternate over time. As shown in Fig. 3, if the two D-UEstransmit with the same maximum power, the region whichis able to receive beacons (which are marked by dark bluedots) is limited due to proximal strong interference. In con-trast, as illustrated in Fig. 4, with power control, the number

−1000 −500 0 500 1000−1000

−800

−600

−400

−200

0

200

400

600

800

1000

[m]

[m]

Fig. 3 Performance of maximum power transmission

−1000 −500 0 500 1000−1000

−800

−600

−400

−200

0

200

400

600

800

1000

[m]

[m]

Fig. 4 Performance of power control transmission

of discoverable D-UEs can be greatly increased and the pos-sible range of beacon detection can be immensely extendedas well.

5.2 Proposed power control algorithm

In this section, we propose a novel power control algorithmwhich is described in Algorithm 1. Firstly, the D-UEs trans-mit and receive beacons during the D2D discovery periodwith maximum power. Then, all D-UEs measure ψ

(b)k ,

which represents the Received-Signal-Strength-Indicator(RSSI) of proximal beacons. Since the measurement ofRSSI can be simply conducted in parallel with receivingbeacons from neighboring D-UEs as part of the usual D2Ddiscovery procedure, this measurement does not consumeadditional resources or processing overhead. Then, everyD-UE figures out the sum of RSSI values with receivedmultiple beacons. In the distributed scheme, as mentionedabove, the D-UEs may not recognize whether a mutualcollision has occurred or not.

Therefore, without distinguishing the D-UEs whichinvolved in collision, we need to approach a probabilisticmethod that a relationship can be defined between the sumof high RSSI energy level and D-UE collision probability.Intuitively, we can assume that the sum of high energy levelcan be regarded as high probability of collision occurrencebecause it indicates that the number of proximal D-UEs maybe large. Specifically, when the discovery resources are pre-defined as a system configuration, the collision probabilitydepends only on the number of D-UEs participating in D2Ddiscovery procedure.

By checking E∗ which is defined as the aggregation ofreceived beacons. If E∗ of D-UE is higher than Ethd, whichis a predefined energy threshold, the D-UE will follow the

Peer-to-Peer Netw. Appl.

proposed power control scheme. For the proposed scheme,we define α as a random variable coefficient, where α ∈[0, 1]. The D-UE multiplies the maximum power level by α,when the power control works. If E∗ of D-UE is lower thanEthd, the D-UE transmits with the maximum power becauseof low density in the vicinity of D-UEs. To determine aproper Ethd, infrastructure can help in D2D networks. Notethat the eNB can also receive multiple beacons during thereserved D2D discovery period. Therefore, infrastructurecan decide variable Ethd according to the density and topol-ogy of D-UEs and it can be periodically broadcast throughthe system information.

6 Collision resolution scheme

6.1 Beacon design

For the collision resolution, collision detection and notifi-cation should be conducted in a way. Under a distributed

D2D networks, due to the absence of a central controller,such operation can be performed by neighboring D-UEs.The proposed collision detection scheme will be explainedin Section 6.2. For the collision notification, a new approachwithout any signaling or resource overhead will be required.We propose a novel beacon design which aims at notifyingthe index of colliding beacon. As mentioned in the pre-vious section, a beacon consists of two RBs, where eachbeacon carries 168 OFDM REs (12 subcarriers × 14 sym-bols, normal cyclic prefix). Therefore, within one beacon,up to 168 bits can be conveyed using QPSK modulation and1/2-rate code.

As shown in Fig. 5, one OFDM symbol (12 bits) is dedi-cated (in blue color) for the purpose of collision notification.With 12 bits, 4096 (212) beacons can be expressed, andhence, 2816 discovery resources (44 RBs × 64 subframes =2816) can be handled.

6.2 Collision resolution scheme

For collision resolution, the occurrence of a collision willbe determined by neighboring D-UEs based on the receivedsignal and interference. The signal strength of D-UE canbe measured by RSSI value of a received beacon, thus theinterference level can also be estimated by received SINRvalue of the beacon.

The main idea is that if the RSSI value of the receivedbeacon is high, whereas its SINR is low, this can be regardedas the occurrence of a collision in proximity. The procedureof the proposed approach is illustrated in Fig. 6. First of all,assuming that D-UE2 and D-UE3 reuse the same beacon inproximity, where they are not aware of the collision event.Under this scenario, D-UE1 receives high RSSI from nearbyD-UE2, while D-UE3 is generating strong interference sig-nal to neighboring D-UE1. Therefore, D-UE1 (assistingdevice) can detect a beacon collision by measuring RSSIand SINR values with thresholds −85 dBm and −1 dB,

Fig. 5 Proposed beacon design

Peer-to-Peer Netw. Appl.

Fig. 6 The collision resolution scheme

respectively. In our work, assisting D-UE informs proximityD-UEs of their collisions. To determine whether other D-UEs are located in the proximity or not, D-UE utilizes RSSI.We set RSSI threshold as −85 dBm, which correspondsto about 100 m with path-loss exponent of 4 and transmitpower of 23 dBm. In other words, assisting D-UE regardsD-UEs, which are located within 100 m from itself, as prox-imity D-UEs. In addition, we set an SINR threshold to bebelow zero (e.g., −1 dB). If SINR value is negative, aggre-gated interference plus noise is larger than desired signal,which means there might be a collision event.

Since all D-UEs transmit or receive beacons periodically,every D-UE can estimate RSSI and SINR level of beacons.Therefore, any D-UE can operate as an assisting devicewhen a collision is detected. When the collision is detected,the index of colliding beacon can be notified using the pro-posed beacon format. After the reception of the notification,the index of transmitted beacons by D-UE2 and D-UE3 willbe randomly changed in the next discovery period for thecollision resolution.

7 Performance evaluation

In this section, we present the performance evaluation ofour proposed schemes. We have implemented a system levelsimulator using MATLAB in accordance with the LTE sys-tem simulation. The main parameters are summarized inTable 1. In this work, we compare our proposed Power Con-trol (PC) and Contention Resolution (CR) schemes with abaseline scheme. In the baseline scheme, all D-UEs use thesame transmission power (23 dBm [16]) and do not reportany collision to their proximity D-UEs. The iteration of dis-covery period is set to 40 that is also represented as one ofthe simulation assumptions [16].

Since the received interference highly depends on theD-UE density or location, we assume that the locationsof D-UEs are modelled by a Poisson Point Process (PPP)

Table 1 System parameters

Parameter Values

Cell layout 7 cell-site (21 cells)

Inter Site Distance (ISD) 500 m (cell radius = 167 m)

Number of UEs 100 UEs per cell

UE dropping Uniform distribution

UE mobility 3 km/h (pedestrian)

System bandwidth 10 MHz (Nf = 44)

Number of subframes 64 ms (Nt = 64)

Carrier frequency 2 GHz

Maximum power of UE 23 dBm

Beacon threshold 4.5 dB

Beacon threshold for CR 6 dB

Path Loss exponent 4

Noise power per RB −121.44 dBm

Channel model Path loss+Shadowing+Multipath

Shadowing Log normal with 7 dB std

Modulation for beacon QPSK and 1/2 code rate

model. With such a process, the D-UEs are independentlyand uniformly distributed in a two dimensional space. Wealso assume that the D-UEs move within the deployedarea according to a pedestrian mobility model, wherethe direction of a D-UE is randomly determined between[0, 2π ].

Figure 7 shows the average discovery success ratioaccording to the number of discovery periods. The proposedschemes outperform the baseline scheme by mitigatingD2D interference. Compared with PC scheme, CR schemeachieves better performance since the collision resolutionwith maximum power transmission can be more effectivethan PC scheme. When a D-UE reduces transmission powerunder PC operation, both interference and desired signal forproximity D-UEs will be reduced. Note that the CR schemerequires higher SINR decoding threshold of beacon detec-tion. Since one symbol is allocated for colliding beaconnotification, the proposed beacon consists of fewer sym-bols. For that reason, the beacon with CR scheme has to usehigher-order modulation and code rate compared with otherschemes.

The delay of average discovery success ratio is shownin Fig. 8. We present initial 5 discovery periods for clearobservation of the performance cross-point. By controllingtransmission power and selecting beacons, the proposedschemes take some delay to satisfy the average of successratio in early stage. However, after two discovery periods inthe case of CR and four discovery periods in the case of PCare passed respectively, the proposed schemes outperformthe baseline scheme due to the power control and contentionresolution gains.

Peer-to-Peer Netw. Appl.

5 10 15 20 25 30 35 400.12

0.14

0.16

0.18

0.2

0.22

0.24

0.26

0.28

0.3

0.32

Number of discovery periods

Avg

. dis

cove

ry s

ucce

ss r

atio

Collision Resolution

Power Control

Baseline

Fig. 7 Average discovery success ratio

1 1.5 2 2.5 3 3.5 4 4.5 50.08

0.1

0.12

0.14

0.16

0.18

0.2

0.22

0.24

0.26

0.28

Number of discovery periods

Avg

. dis

cove

ry s

ucce

ss r

atio

Collision Resolution (CR)

Power Control (PC)

Baseline

Fig. 8 Delay of avg. discovery success ratio

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.450

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Discovery success ratio

CD

F

Baseline

Power Control (PC)

Collision Resolution (CR)

Fig. 9 CDF of discovery success ratio of D-UEs

240 250 260 270 280 290 3000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Avg. distance of detected D−UEs

CD

F

Baseline

Collision Resolution (CR)

Power Control (PC)

Fig. 10 CDF of avg. distance of discovered D-UEs

Figure 9 shows that a Cumulative Density Function(CDF) of discovery success ratio, and we observe that theproposed schemes outperform the baseline scheme. Themutual D2D interference will be reduced by PC scheme.In addition, the interference by collisions will be resolvedby the proposed CR scheme. These results prove that ourproposed schemes achieve to detect and be detected bylarger number of D-UEs than baseline scheme does, thusimproving overall D2D system performance. The CDF ofthe maximum distance of detected D-UEs is presented inFig. 10. We observe that the proposed schemes outper-form the baseline scheme by detecting farther D-UEs. Aspresented in Fig. 10, PC achieves higher gain than CR,regarding the average distance between the discovered D-UEs. Although the D-UEs change beacons in the CRscheme, due to the maximum power transmission, the inter-ference will affect multiple receivers under D2D networks.In contrast, the D-UEs control the transmission power,the reachable interference range will dynamically changeaccording to the power level. Therefore, when the D-UEreduces the power, the detectable region by other proximityD-UEs will be greatly extended.

8 Concluding remarks

In this article, we propose novel distributed power con-trol and contention resolution schemes for D2D discov-ery. We have investigated the performance of the pro-posed schemes compared with the baseline scheme. Thesimulation results demonstrate that the discovery suc-cess ratio is remarkably improved such that the averagedistance of detected devices is increased as well. TheCR scheme can simply overcome the D2D interferenceby collisions without a signaling or resource overhead,

Peer-to-Peer Netw. Appl.

while the PC scheme can mitigate D2D interference bytransmitter power control method. Moreover, the resultsreveal that there is a trade-off relationship between gainand delay.

We conclude that the proposed schemes can be exploitedfor various D2D discovery scenarios and applications infuture LTE-A/5G networks, while guaranteeing the differentdelay requirements.

Acknowledgments This research was supported by LG ElectronicsCo. Ltd and the Brain Korea 21 Plus Project in 2015.

References

1. Fodor G, Dahlman E, Parkvall S, Mildh G, Reider N, MiklosG, Turanyi Z (2012) Design aspects of network assisteddevice-to-device communications. IEEE Commun Mag 50(3):170–177

2. TR 3GPP22. 803 (2013) Feasibility study for Proximity Services(ProSe). ver. 12.0.0

3. TS 3GPP22. 278 (2014) Service requirements for the EvolvedPacket System (EPS). ver. 13.0.0

4. TS 3GPP23. 303 (2014) Proximity-based services (ProSe). ver.12.1.0

5. Shalmashi S, Miao G, Han Z, Slimane SB (2014) Interference con-strained device-to-device communications. In: Proceedings IEEEICC, Sydney

6. Lee N, Lin X, Andrews JG, Heath RWJr Power control forD2D underlaid cellular networks: modeling, algorithms and anal-ysis,” IEEE Selected Areas in Communications, vol. pp. [Online].Available: arXiv:1305.6161

7. Fodor G, Reider N (2011) A distributed power control scheme forcellular network assisted D2D communications. In: ProceedingsIEEE GLOBECOM, Houston

8. Phunchongharn P, Hossain E, Kim DI (2013) Resource allocationfor device-to-device communications underlaying LTE-advancednetworks. IEEE Wireless Magazine 20(4):91–100

9. Yu CH, Doppler K, Ribeiro CB, Tirkkonen O (2011) Resourcesharing optimization for device-to-device communication under-laying cellular networks. IEEE Trans Wirel Commun 10(8):2752–2763

10. Kang HJ, Kang CG (2014) Moble device-to-device (D2D) con-tent delivery networking: a design and optimization framework. JCommun Netw 16(5):568–577

11. Doppler K, Ribeiro CB, Kneckt J (2011) Advances in D2D com-munication: energy efficient service and device discovery radio.In: Proceedings of Wireless VITAE, Chennai

12. Hong J, Park S, Kim H, Choi S, Lee KB (2013) Analysis ofdevice-to-device discovery and link setup in LTE networks. In:Proceedings of IEEE PIMRC, London

13. Hong J, Park S, Choi S (2014) Neighbor device-assisted beaconcollision detection scheme for D2D discovery. In: Proceedings ofIEEE ICTC, Busan

14. Park S, Choi S (2014) Expediting D2D discovery by using tempo-rary discovery resource. In: Proceedings of IEEE GLOBECOM,Austin

15. Lin X, Andrews JG, Ghosh A, Ratasuk R (2014) An overview of3GPP device-to-device proximity services. IEEE Commun Mag52(4):40–48

16. 3GPP TR 36.843 (2014) Study on LTE device to device proximityservices radio aspects. ver. 12.0.1

17. Qualcomm (2013) 3GPP TSG RANWG1 74, Techinques for D2DDiscovery. R1-133600

18. Alcatel-Lucent Shanghai Bell (2014) 3GPP TSG RAN WG1 76,D2D discovery channel resource allocation. R1-140176

19. Samsung (2014) 3GPP TSG RAN WG1 76, resource allocationmethod for D2D discovery. R1-140393

20. 3GPP TR 36.211 (2014) Evolved universal terrestrial radio access(E-UTRA); Physical channels and modulation. ver. 12.2.0

21. Baccelli F, Khude N, Laroia R, Li J, Richardson T, ShakkottaiS, Tavildar S, Wu X (2012) On the design of device-to-device autonomous discovery. In: Proceedings of COMSNETS,Bangalore

Peer-to-Peer Netw. Appl.

Jongwoo Hong received theB.S. and M.S. degrees inElectrical Engineering fromKyunghee University andHanyang University, Koreain 2005 and 2007, respec-tively. He is currently workingtoward a Ph.D. degree in theDepartment of Electrical andComputer Engineering, SeoulNational University (SNU),Seoul, Korea. Prior to joiningSNU, he was an engineer atSamsung Electronics, Suwon,Korea from 2007 to 2010.His research interests includeLTE/LTE Advanced networks,

Device-to-Device (D2D) discovery and communication.

Seungil Park received theB.S. degree from the Depart-ment of Electronic andElectrical Engineering (EEE),POSTECH, Korea in 2011and the M.S. degree fromthe Department of ElectricalEngineering and ComputerScience, Seoul National Uni-versity (SNU), Korea in 2013,respectively. He is currentlyworking toward a Ph.D.degree in the Departmentof Electrical and Com-puter Engineering, SNU,Korea. His research interests

include wireless networks, Device-to-Device (D2D) communication,Vehicle-to-Everything (V2X) communication.

Sunghyun Choi is a profes-sor at the Department of Elec-trical and Computer Engineer-ing, Seoul National University(SNU), Korea. Before joiningSNU in 2002, he was withPhilips Research USA. Hewas also a visiting associateprofessor at Stanford Univer-sity, USA from June 2009to June 2010. He receivedhis B.S. (summa cum laude)and M.S. degrees from KoreaAdvanced Institute of Sci-ence and Technology in 1992and 1994, respectively, andreceived Ph.D. from The Uni-

versity of Michigan, Ann Arbor in 1999. He authored/coauthored over190 technical papers and book chapters in the areas of wireless/mobilenetworks and communications. He has co-authored (with B. G. Lee)a book “Broadband Wireless Access and Local Networks: MobileWiMAX and WiFi,” Artech House, 2008. He holds about 120 patents,and has tens of patents pending. He has served as a General Co-Chairof COMSWARE 2008, and a Technical Program Committee Co-Chairof ACM Multimedia 2007, IEEE WoWMoM 2007 and COMSWARE2007. He has also served on program and organization committeesof numerous leading wireless and networking conferences includingACM MobiCom, IEEE INFO-COM, IEEE SECON, and IEEE WoW-MoM. He is also currently serving as an editor of IEEE Transactionson Wireless Communications and IEEE Wireless Communications.He served as an associate editor of IEEE Transactions on MobileComputing and a guest editor of IEEE Journal on Selected Areasin Communications (JSAC) and ACM Wireless Networks (WINET).From 2000 to 2007, he was an active voting member of IEEE 802.11WLAN Working Group. He has received numerous awards includingKICS Dr. Irwin Jacobs Award (2013); the Presidential Young ScientistAward (2008); IEEK/IEEE Joint Award for Young IT Engineer (2007);Shinyang Scholarship Award (2011); the Outstanding Research Award(2008) and the Best Teaching Award (2006) from the College ofEngineering, SNU; and the Best Paper Award from IEEE WoWMoM2008. He was named IEEE fellow in 2014 for the contribution to thedevelopment of WLAN protocols.