Resource Allocation for Network-Integrated Device-to-Device Communications Using Smart Relays

Transcription

1 Resource Aocation for Network-Integrated Device-to-Device Communications Using Smart Reays Monowar Hasan and Ekram Hossain Department of Eectrica and Computer Engineering, University of Manitoba, Winnipeg, Canada arxiv:3.0843v [cs.ni] 4 Nov 203 Abstract With increasing number of autonomous heterogeneous devices in future mobie networks, an efficient resource aocation scheme is required to maximize network throughput and achieve higher spectra efficiency. In this paper, performance of network-integrated device-to-device D2D communication is investigated where D2D traffic is carried through reay nodes. An optimization probem is formuated for aocating radio resources to maximize end-to-end rate as we as conversing QoS requirements for ceuar and D2D user equipment under tota power constraint. Numerica resuts show that there is a distance threshod beyond which reay-assisted D2D communication significanty improves network performance when compared to direct communication between D2D peers. Index Terms Resource aocation, LTE-A L3 reay, D2D communication I. INTRODUCTION Device-to-device D2D communication underaying ceuar network has recenty been intensivey discussed in standardization committee and academia. Reusing the LTE-A ceuar resources, D2D communication enabes wireess peerto-peer services directy between user equipments UEs which enhances spectrum utiization and improves ceuar coverage. Possibe usage cases for D2D communication are oca voice and data services incuding content sharing i.e., exchanging photos, videos or documents through smart phones and mutipayer gaming []. In the context of D2D communication, it becomes a crucia issue to set up reiabe direct inks between the UEs whie satisfying quaity-of-service QoS of traditiona ceuar UEs CUEs and D2D UEs in the network. Besides, interference to and from CUEs and poor propagation channe may imit the advantages of D2D communication in practica scenarios. In such cases, network assisted transmission through reays coud efficienty enhance the performance of D2D communication when D2D-pairs are too far away from each other or the quaity of D2D channe is not good enough for direct communication. In this paper, we consider reay-assisted D2D communication in LTE-A ceuar networks where D2D-pairs are served by the reay node. We concentrate on the scenario in which potentia D2D UEs are ocated near to each other i.e, office bocks or university areas, concert sites etc.; however, the proximity and ink condition may not be favorabe for direct communication. Thanks to LTE-A Layer-3 L3 reay featuring with sef-backhauing configuration which makes it capabe to perform operations simiar to those of a base station i.e., Evoved Node B [enb] in an LTE-A network. We formuate the resource aocation probem with an objective to maximizing the end-to-end rate i.e., minimum achievabe rate over two hops for the UEs whie maintaining the QoS i.e., rate requirements for ceuar and D2D UEs under tota power constraint at the reay node. The resource aocation probem turns out to be a mixed-integer non-inear programming MINLP probem and to make it tractabe we reax it using the time-sharing strategy. The contribution of this paper is the anaysis of network performance under reay-assisted D2D communication. The numerica resuts show that after a distance threshod reaying D2D traffic provides significant gain in achievabe data rate. The remainder of this paper is organized as foows. A review of reated work is presented in Section II. Section III introduces LTE-A access methods and the reaying mechanisms. In Section IV, we present the system mode and formuate the resource aocation probem. The permanence evauation resuts are presented in Section V and finay we concude the paper in Section VI outining possibe future works. II. RELATED WORK AND MOTIVATIONS Resource aocation in context of D2D communication for future generation OFDMA based wireess networks is one of the active areas of research. In [2], a greedy heuristic based resource aocation scheme is devoved for both upink and downink scenarios where a D2D-pair shares same resources with traditiona user if the achieved SINR is greater than a threshod SINR. A resource aocation scheme based on a coumn generation method is proposed in [3] to maximize the spectrum utiization by finding the minimum transmission ength i.e., time sots for D2D inks whie protecting ceuar users from interference and guaranteeing QoS. A distributed suboptima joint mode seection and resource aocation scheme is proposed in [4] to reduce intrace and interce interference. In [5], authors consider reay seection and resource aocation for upink scenarios with two casses of users having different i.e., specific and fexibe rate requirements. The objective is to maximize system throughput by satisfying rate requirements for the rate-constraint users whie confining the transmit power within power-budget. In [6], performance i.e., maximum ergodic capacity and outage probabiity of cooperative reaying in reay-assisted D2D

2 2 communication is investigated considering power constraints at enb and numerica resuts show that muti-hop reaying owered the outage probabiity. However, in [2] [5], the effect of using reays in D2D communication is not studied. As a matter of fact, reaying mechanism expicity in context of D2D communication has not been considered so far in the iterature and most of the resource aocation schemes consider ony one D2D ink. Taking the advantage of L3 reays supported by the 3GPP standard, we study the network performance of network-integrated D2D communication and show that reay-assisted D2D communication provides significant performance gain for ong distance D2D inks. A brief review of radio access and reaying mechanism in the LTE-A standard is provided next. III. RADIO ACCESS AND RELAYING IN 3GPP LTE-A A. Radio Access Methods in LTE-A Networks In the LTE-A radio interface, two consecutive time sots create a subframe where each timesot spans 0.5 msec. Resources are aocated to UEs in units of resource bocks RBs over a subframe. Each RB occupies sot 0.5 msec in time domain and 80 KHz in frequency domain with subcarrier spacing of 5 KHz. The mutipe access scheme for downink i.e., enb/reay-to-ue is OFDMA whie the access scheme for upink i.e., UE-to-reay/eNB, reay-to-ue is singe carrier- FDMA SC-FDMA. In genera, SC-FDMA requires contiguous set of subcarrier aocation to UEs. Resource aocation in downink supports both bock-wise transmission ocaized aocation and transmission on non-consecutive subcarriers distributed aocation. For upink transmission, current specification supports ony ocaized resource aocation [7]. B. Reays in LTE-A Networks Reay node in LTE-A is wireessy connected to radio access network through a donor enb and serves UEs. Depending on the function, different reaying mechanisms used in LTE-A [8]. Layer L reays act as repeaters, ampifying the input signa without and decoding/re-encoding. The L reays can either use the same carrier frequency i.e., in-band reaying or an orthogona carrier frequency i.e., out-of-band reaying. The main advantages of L reays are simpicity, cost-effectiveness, and ow deay. However, with L reaying, noise and interference are aso ampified and retransmitted. Hence, the SINR of the signa may deteriorate. Layer 2 L2 reays are aso known as decode and forward DF reay which invoves decoding the source signa at the reay node. The advantage of DF reays is that noise and interference do not propagate to the destination. However, a substantia deay occurs during the reaying operation. A L2 reay does not issue any scheduing information or any contro signa i.e., HARQ and channe feedback. Hence, an L2 reay cannot generate a compete ce and from a UE s perspective, it is ony a part of donor ce. Layer 3 L3 reays with sef-backhauing configuration performs the same operation as enb except for ower transmit By the term UE, we refer to both ceuar and D2D user equipments. Fig.. L3 reay enb A singe ce with mutipe reay nodes. Ceuar UE D2D device power and smaer ce size. It contros ces and each ce has its own ce identity. The reay sha transmit its own contro signas and UE sha receive scheduing information and HARQ feedback directy from the reay node. When the ink condition between D2D peers is poor or the distance is too far for direct communication, with the support of L3 reays, scheduing and resource aocation for D2D UE can be done in reay node and D2D traffic can be transmitted through reay. We refer to this as network-integrated D2D communication which can be an aternative approach to provide better quaity of service between distant D2D-inks. In the next section, we describe the network configuration and present the formuation for resource aocation. IV. SYSTEM MODEL AND PROBLEM FORMULATION Let L = {, 2,..., L} fixed-ocation L3 reays are avaiabe in the network as shown in Fig.. The CUEs and D2Dpairs correspond to set C and D, respectivey, where the D2D-pairs are discovered during the D2D session setup. We consider ocaized resource aocation where system bandwidth is divided into N RBs denoted by N = {, 2,..., N}. We assume that the CUEs are outside the coverage region of enb and/or having bad channe condition, and therefore, CUE-eNB communications need to be supported by the reays. Besides, the direct communication between two D2D UEs coud be unfavourabe due to ong distance and/or poor ink condition, and therefore, requires the assistance of a reay node. The UEs i.e., both ceuar and D2D assisted by reay are denoted by u. The set of UEs assisted by reay is U such that U {C D}, L; U = {C D} and U =. According to our system mode, taking the advantage of an L3 reay, scheduing and resource aocation is performed in the reay node to reduce overoad at the enb. We define h n i,j the ink gain between the ink i and j over RB n. The unit power SINR for the ink between UE u U

3 3 and reay using RB n in the first hop is as foows: foows: γ n u,, = u j U j,j,j L h n u, P n u j,jh n u j, + N 0 B RB. The unit power SINR for the ink between reay and enb for CUE i.e., u {C U } in the second hop is given by γ n,u,2 = u j U j,j,j L h n,enb h n j,enb + N 0 B RB. 2 Simiary, the unit power SINR for the ink between reay and receiving D2D UE for the D2D-pair i.e., u {D U } in the second hop can be written as γ n,u,2 = u j U j,j,j L h n,u h n j,u + N 0 B RB 3 where Pi,j n is the power assigned between ink i and j over RB n, B RB is bandwidth of RB, and N 0 denotes therma noise. h n,enb is the gain between reay-enb ink; hn,u is the gain between reay and receiving D2D UE for the D2D-pair u. The achievabe data rate for the UE u in the first hop can be expressed as, ru n, = B RB og 2 +Pu n, γn u,,. Simiary, the achievabe data rate in the second hop is as foows: ru n,2 = B RB og 2 + P,u n γ,u n,2. Note that, for the CUE i.e., UEs {C U }, the SINR in the second hop is cacuated from 2; on the other hand, the SINR for the D2D UEs i.e., UEs {D U } is cacuated from 3. The end-to-end data rate on RB n for the UE u is the minimum achievabe data rate over two hops, i.e., R n u = 2 min { r n u,, r n u,2}. 4 A. Resource Bock and Power Aocation in Reay Nodes The objective of resource aocation i.e., RB and transmit power aocation is to specify for each reay, the RB and power eve assignment to the UEs which maximizes the system capacity defined as the minimum achievabe data rate over two hops. Let the maximum aowabe transmit power for UE reay is Pu max P max. The RB aocation indicator is denoted by binary decision variabe {0, }, where = if RB n is assigned to UE u and 0, otherwise. The same RBs wi N be used by reay in the second hop and R u = Ru n denotes the achievabe sum-rate over aocated RBs. Q u denotes the QoS rate requirements for a UE u. The resource aocation probem for each reay L can be formuated as N P max x n x n u,p n u,,p,u n u Ru n subject to, n N 5a N N P n u, P max u, u U 5b P,u n P max 5c Pu n,h n u ref,, Ith,, n n N 5d P,u n h n u ref,,2 Ith,2, n n N 5e R u Q u, u U 5f Pu n, 0, P,u n 0, n N, u U 5g where R n u = 2 min { BRB og 2 +P n u, γn u,,, B RB og 2 +P n,u γ n,u,2 }, SINR for the first hop, γ n u,, = u j U j,j,j L and SINR for the second hop, γ,u n,2 = u j U j,j,j L u j U j,j,j L h n u, x n u j P n u j,jh n u j, + N 0 B RB h n,enb x n u j h n j,enb + N 0 B RB, h n,u u {C U } x n u j h n j,u + N 0 B RB, u {D U }. With the constraint in 5a, each RB is assigned to ony one UE. Under the constraints in 5b and 5c, the transmit power is imited by maximum power budget. 5d and 5e constraint the amount of interference introduced to other reays and receiving D2D UEs in first and second hop, respectivey, to be ess than some threshod. Constraint 5f ensures the minimum QoS requirements for the CUE and D2D UEs. The constraint in 5g is the non-negativity condition of transmit power. Simiar to [9], we adopt the concept of reference node. For exampe, in the first hop, each UE associated with reay node chooses from among the neighbouring reays having the highest channe gain according to 6a and aocates the power eve considering the interference threshod. Simiary, in the second hop, transmit power for each reay wi be adjusted accordingy considering interference introduced to receiving D2D UEs associated with neighbouring reays according to

4 4 6b. h n u ref,,= argmax j h n u ref,,2= argmax u j h n u,j; u U, j, j L. 6a h n,u j ; j, j L, u j {D U j }. 6b From 4, the maximum rate for UE u over RB n is achieved when Pu n, γn u,, = P,u n γ,u n,2. Therefore, power aocated to reay node for UE u can be expressed as a function of power at UE as P,u n = γn u,, γ P,u n u n,2, and the rate of UE u over RB n, R n u = 2 B RB og 2 + P n u,γ n u,,. 7 The optimization probem P is a mixed-integer non-inear program MINLP which is computationay intractabe and very compex to sove. A common approach in iterature is to reax the constraint that an RB is used by ony one UE using time-sharing factor [0]. Thus 0, ] is represented as the sharing factor where each denotes the portion of time that RB n is assigned to UE u and satisfies, n. Besides, we introduce a new variabe Su n, = xn u Pu n, which denotes the actua transmit power of UE u on RB n []. Then the reaxed probem can be reformuated as foows: P2 N max,s n u, N 2 xn u B RB og 2 + Sn u, γn u,, subject to, n 8a N N γ n u,, γ n,u,2 γ n,u,2 2 xn u B RB og 2 S n u, P max u, u 8b S n u, P max 8c S n u,h n u ref,, I n th,, n 8d γ n u,, S n u,h n u ref,,2 I n th,2, n 8e + Sn u, γn u,, Q u, u 8f S n u, 0, n, u. 8g The duaity gap of any optimization probem satisfying time sharing condition is negigibe as the number of subcarrier becomes significanty arge. Since our optimization probem satisfies the time-sharing condition, the soution of the reaxed probem is asymptoticay optima [2]. The optimization probem P2 is convex; the objective function is concave, constraint 8f is convex and a the remaining constraints are affine. Therefore this probem can be soved by the interior point method [3]. To observe the nature of power aocation for a UE, we use Karush-Kuhn-Tucker KKT optimaity and define the foowing Lagrangian function: Lx, S, µ, ρ, ν, ψ, ϕ, λ = N 2 xn u B RB og 2 + Sn u, γn u,, N + µ n + N γ + ν n u,, Su n, P max + + γ,u n,2 N ψ n N ϕ n + λ u S n u,h n u ref,, Q u ρ u I n th, N γ n u,, γ n,u,2 S n u,h n u ref,,2 I n th,2 N Su n, Pu max 2 xn u B RB og 2 + Sn u, γn u,, where λ is the vector of Lagrange mutipiers associated with individua QoS requirements for ceuar and D2D UEs. Simiary, µ, ρ, ν, ψ, ϕ are the Lagrange mutipiers for constraint 8a-8e. Differentiating 9 with respect to Su n,, we obtain the foowing power aocation for UE u over RB n: where δ = P n u, = Sn u, = ρ u + γn u,, γ n,u,2 [ δ γ n u,, 2 B +λu RB n 2 9 ] + 0 ν +h n u ref,, ψn+ γn u,, γ,u n h n u,2 ref,,2 ϕn and [ε] + = maxε, 0, which is a muti-eve water fiing aocation []. B. Semi-distributed Resource Aocation Agorithm Each reay in the network independenty aocates resources to its associated UEs. Based on the mathematica formuation in the previous section, the overa resource aocation agorithm is shown in Agorithm. Agorithm Joint RB and power aocation agorithm : UEs measure interference eve from previous time sot and inform the respective reays. 2: Each reay L obtains the channe state information among a reays j; j, j L and to its schedued UEs u j U j ; j, j L. 3: For each reay and its associated UEs, obtain the reference node for the first and second hops according to 6a and 6b. 4: Sove the optimization probem P2 for each reay independenty to obtain RB and power aocation. 5: Aocate resources i.e., RB and transmit power to associated UEs for each reay and cacuate average achievabe data rate.

5 5 TABLE I SIMULATION PARAMETERS Parameter Vaues Carrier frequency 2.35 GHz System bandwidth 2.5 MHz Tota number of RBs for each reay 3 Reay ce radius 200 meter Distance between reay and enb 25 meter Minimum distance between UE and reay 0 meter Tota power avaiabe at each reay 30 dbm Tota power avaiabe at UE 23 dbm Noise power spectra density 74 dbm/hz Shadow fading standard deviation in reay-enb ink 6 db Shadow fading standard deviation in UE-reay ink 0 db Rate requirement for ceuar UEs 00 bps Rate requirement for D2D UEs 200 bps Since the L3 reays can perform the same operation as an enb, these reays can communicate using the X2 interface [4] defined in the 3GPP LTE-A standard. Therefore, in the proposed agorithm, the reays can obtain the channe state information through inter-reay message passing without increasing signaing overhead at the enb. A. Parameters V. PERFORMANCE EVALUATION The performance resuts for the resource aocation schemes obtained by a simuator written in MATLAB. In order to measure channe gain, we consider both distant-dependent path-oss and shadow fading; and the channe is assumed to be frequency-seective and experience Rayeigh fading. In particuar, we consider reaistic 3GPP propagation environment 2 presented in [5]. For exampe, UE-reay and reay-d2d ink foows the foowing path-oss equation: P L u, [db] = og + L su + 0 ogφ, where is the distance between UE and reay in kiometer; L su accounts for shadow fading and is modeed as a ognorma random variabe, and φ is an exponentiay distributed random variabe which represents Rayeigh fading. Simiary, the path-oss equation for reay-enb ink is expressed as: P L,eNB [db] = og+l sr +0 ogφ, where L sr is a og-norma random variabe accounting for shadow fading. The simuation parameters and assumptions used for obtaining the numerica resuts are isted in Tabe I. We simuate a singe three-sectored ceuar network in an rectanguar area of 700m 700m, where the enb is ocated in the centre of the ce and three reays are depoyed in the network, i.e., one reay in each sector. The CUEs and the transmitter UE of a D2D-pair are uniformy distributed within the radius of the reay ce. The other UE of the D2Dpair is distributed uniformy in the overapping area of the reay radius and a circe centred at the first D2D UE as shown in Fig. 2. The circe radius which gives the maximum distance between UEs in a D2D-pair is varied as a simuation 2 Any other propagation mode for D2D communication can be used for the proposed resource aocation method. d a Reay ce radius Fig. 2. Distribution of any D2D-pairs for two cases: a other UE of the D2D-pair is distributed anywhere on the edge of the circe with radius d ; b other UE of the D2D-pair having distance d 2 between them is uniformy distributed on the soid arc. parameter. The numerica resuts are averaged over different reaizations of simuation scenarios i.e., UE ocations and channe gains. B. Numerica Resuts In order to study network performance in presence of the L3 reay, we compare the performance of the proposed scheme with a reference scheme [2] in which an RB aocated to CUE can be shared with at most one D2D-ink. D2D UE shares the same RBs aocated to CUE by soving optimization probem and communicate directy between peers without reay ony if the QoS requirements for both CUE and D2D UE are satisfied. Achievabe data rate vs. distance between D2D-inks: In Fig. 3, we iustrate the average achievabe data rate R d 2 b u D R ach u for D2D UEs which is cacuated as R = D, where Ru ach is the achievabe data rate for UE u and denotes set cardinaity. Athough the reference scheme outperforms when the distance between D2D-ink is coser i.e., d < 60m; our proposed agorithm can greaty increase the data rate especiay when the distance increases. This is due to the fact that when the distance is higher, the performance of direct communication deteriorates due to poor propagation medium. Besides, when the D2D UEs share resources with ony one CUE, the spectrum may not utiize efficienty and decreases the achievabe rate. Consequenty, the gap between the achievabe rate of our proposed agorithm and that of the reference scheme becomes wider when the distance increases. 2 Rate gain vs. distance between D2D-inks: Fig. 4a depicts the rate gain in terms of aggregated achievabe rate for the UEs. We cacuate the gain as foows: Rate gain = R prop R ref R ref 00% where R prop and R ref is the aggregated rate for the UEs in proposed and reference scheme, respectivey. It is observed from the figure that, with the increasing distance between D2D-inks our proposed scheme provides significant gain in terms of achievabe data rate. To observe the effect of gain in different network reaization we vary the number of D2D UE in Fig. 4b. It is cear from figure that irrespective of the number of D2D UEs in the network, our proposed scheme provides considerabe rate gain for distant D2D-pairs.

6 Rate gain % oss of reaying d < 60m Maximum distance between D2D inks m a Rate Gain % Maximum distance between D2D inks m b 3 6 Number of 9 D2D inks 2 Fig. 4. Gain in aggregated achievabe data rate with varying distance for C = 5, interference threshod -70 dbm: a 3 D2D-pairs assisted by each reay i.e., D = 9; b number of D2D-pairs varies from to 4 UEs/reay i.e., D = 3, 6, 9, 2. There is a critica distance d i.e., d 60m here, beyond which reaying provides significant performance gain. Average achievabe rate bps x 05 Proposed Scheme Reference Scheme Maximum distance between D2D inks m Fig. 3. Average achievabe data rate with varying distance; number of CUE, C = 5 i.e., 5 CUEs assisted by each reay, number of D2D-pair, D = 9 i.e., 3 D2D-pair assisted by each reay and interference threshod -70 dbm. VI. CONCLUSION We have provided a mathematica formuation for resource aocation and anayzed network performance of reay-assisted D2D communication. The performance evauation resuts have shown that reay-assisted D2D communication is beneficia to provide higher rate for distant D2D-inks. Aong with the rate requirements, it can be possibe to measure additiona QoS parameters i.e., deay for observing network performance propery by other mathematica toos i.e., queuing modes. Besides when the perfect channe knowedge and the information about number of active UEs are not avaiabe, the effects of uncertainties in the system parameter need to be considered by using a robust optimization formuation. These issues wi be expored in our future works. REFERENCES x 0 5 [] L. Lei, Z. Zhong, C. Lin, and X. Shen, Operator Controed Deviceto-Device Communications in LTE-advanced Networks, IEEE Wireess Communications, vo. 9, no. 3, pp , 202. [2] M. Zuhasnine, C. Huang, and A. Srinivasan, Efficient Resource Aocation for Device-to-Device Communication Underaying LTE Network, in 200 IEEE 6th Internationa Conference on Wireess and Mobie Computing, Networking and Communications WiMob, 200, pp [3] P. Phunchongharn, E. Hossain, and D. I. Kim, Resource Aocation for Device-to-Device Communications Underaying LTE-Advanced Networks, IEEE Wireess Communications, to appear. [4] M. Beeschi, G. Fodor, and A. Abrardo, Performance Anaysis of a Distributed Resource Aocation Scheme for D2D Communications, in 20 IEEE GLOBECOM Workshops GC Wkshps, 20, pp [5] M. Aam, J. W. Mark, and X. Shen, Reay Seection and Resource Aocation for Muti-user Cooperative LTE-A Upink, in 202 IEEE Internationa Conference on Communications ICC, 202, pp [6] D. Lee, S.-I. Kim, J. Lee, and J. Heo, Performance of Mutihop Decode-and-Forward Reaying Assisted Device-to-Device Communication Underaying Ceuar Networks, in Internationa Symposium on Information Theory and its Appications ISITA, 202, pp [7] A. Ghosh, J. Zhang, J. G. Andrews, and R. Muhamed, Fundamentas of LTE, st ed. Prentice Ha Press, 200. [8] D. I. Kim, W. Choi and B.-H. Kim, Partia Information Reaying and Reaying in 3GPP LTE, in Cooperative Ceuar Wireess Networks. Cambridge University Press, 20. [9] K. Son, S. Lee, Y. Yi, and S. Chong, REFIM: A Practica Interference Management in Heterogeneous Wireess Access Networks, IEEE Journa on Seected Areas in Communications, vo. 29, no. 6, pp , 20. [0] Z. Shen, J. Andrews, and B. Evans, Adaptive Resource Aocation in Mutiuser OFDM Systems With Proportiona Rate Constraints, IEEE Transactions on Wireess Communications, vo. 4, no. 6, pp , [] M. Tao, Y.-C. Liang, and F. Zhang, Resource Aocation for Deay Differentiated Traffic in Mutiuser OFDM Systems, IEEE Transactions on Wireess Communications, vo. 7, no. 6, pp , [2] W. Yu and R. Lui, Dua Methods for Nonconvex Spectrum Optimization of Muticarrier Systems, IEEE Transactions on Communications, vo. 54, no. 7, pp , [3] S. Boyd and L. Vandenberghe, Convex Optimization. New York, NY, USA: Cambridge University Press, 2004, Chapter. [4] Acate-Lucent, The LTE Network Architecture, December 2009, White Paper. [5] Y. Yuan, LTE-A Reay Scenarios and Evauation Methodoogy, in LTE- Advanced Reay Technoogy and Standardization. Springer, 203, pp