BULGARIAN ACAEMY OF SCIENCES CYBERNETICS AN INFORMATION TECHNOLOGIES Volume 15, No 6 Special Issue o Logistics, Iformatics ad Service Sciece Sofia 015 Prit ISSN: 1311-970; Olie ISSN: 1314-4081 OI: 10.1515/cait-015-0070 Joit Resource Allocatio Scheme for evice-to-evice Commuicatio uder a Cellular Networ Chewei Feg 1, Pigbo Che, Zewag Zhag 1 1 School of Opto-Electroic ad Commuicatio Egieerig, Xiame Uiversity of Techology, Xiame, 36104 Chia School of Iformatio Sciece ad Egieerig, Xiame Uiversity, Xiame, 361005 Chia Emails: cwfeg@xmut.edu.c mejie005@163.com zwzhag@xmut.edu.c Abstract: evice-to-evice () commuicatio is a ew techology that allows the mobile termials to directly commuicate with each other by sharig the resources of the cellular etwor uder the cotrol of a cellular system. commuicatio was itroduced ito the cellular etwor i order to improve the efficiecy of resource utilizatio ad icrease the system throughput. However, there exist iterfereces betwee lis ad cellular lis due to the resource reuse. I order to efficietly limit the cases of iterferece, a joit resource allocatio scheme for commuicatio uder a cellular etwor is proposed. The scheme, based o the user s Quality of Service (QoS), defies a ew utility fuctio to balace the icreasig system throughput ad the iterferece i the cellular users commuicatio. The optimal trasmittig power is obtaied through optimizig the utility fuctio by usig the covex optimizatio theory. The simulatio shows that the scheme ca fulfill the user s QoS requiremets efficietly, improve the system throughput ad the reliability of the resources as well. Keywords:, cellular etwor, resource allocatio, power cotrol, covex optimizatio. 1. Itroductio The wireless commuicatio etwor has bee developed for higher data rate, greater resource utilizatio ad higher etwor capacity [1]. However, due to the limited spectrum resources available i the wireless etwor, the problem how to achieve this goal has become the focus of the research i idustry. I order to mae use of the badwidth resource efficietly ad improve the etwor throughput, 91 Uautheticated owload ate 7/19/18 5:40 PM
more ad more researchers pay attetio to commuicatio techology. commuicatio ca co-exist with the existig cellular wireless commuicatio systems ad few chages are eeded i the system structure. commuicatio techology maes users use the cellular resources uder the cotrol of the cellular etwor, which ca ehace the spectrum efficiecy of the cellular commuicatio system, offload the data of the cellular etwor, icrease the data rate ad reduce the eergy cosumptio of the mobile termials. There are two mai types of resource allocatios for commuicatio i the cellular etwor: orthogoal resource allocatio ad sharig resource allocatio. The iterferece caused by the orthogoal resource allocatio has lower resource utilizatio; o the cotrary, the iterferece caused by the sharig resource allocatio has higher resource utilizatio. ue to the shortage of the spectrum resource, the sharig resource allocatio is curretly domiatig. commuicatio will cause the correspodig iterferece to the cellular users because of reusig the resources of the cellular system, therefore, the users of the cellular system ad eed to coordiate with each other to reduce the iterferece ad esure ormal commuicatio through the system []. There are may studies about resource allocatio ad iterferece maagemet. I [3] models of resource allocatio ito a iteger oliear optimizatio problem are preseted, that optimize the whole throughput of the cellular etwor ad through a heuristic algorithm with the costrait of a miimum Sigal to Iterferece ad Noise Ratio (SINR) requiremet. I [4] the authors propose a iterferece-aware resource allocatio for commuicatio, usig a sequetial secod price auctio algorithm. I [5] the system throughput is ehaced by choosig user s commuicatio mode i a reasoable maer to reduce the iterferece caused by users i the cellular etwor whe reusig the cellular resources. I [6] the authors propose a method based o user s QoS requiremets, which allocates the resource flexibly ad efficietly rather tha reusig a fixed cellular user s resource. I [7] a power cotrol scheme is proposed to reduce the iterferece of li to the cellular li whe the SINR of the cellular li is decliig. Referece [8] deduces the optimal power allocatio i the sceario of a cellular user ad a pair of users, ad icreases the total throughput of the system by esurig the priority of the cellular user. I [9] the authors propose a resource allocatio algorithm based o Sigal to Noise Ratio (SNR) equalizatio, which establishes covex optimizatio to maximize the total system throughput by optimizig the received iterferece power ad balacig SNR of all users. I [10] a sharig resource allocatio algorithm is suggested based o the dyamic power cotrol, which adaptively adjusts the trasmissio power of equipmet i each Resource Bloc (RB) to miimize the iterferece to the cellular etwor. I [11] the authors offer a dyamic power cotrol scheme based o the chael coditios of the cellular ad users, which adjusts the trasmissio power of user with a predefied coverage threshold to efficietly reduce the iterferece of to the cellular system. This paper proposes a joit resource allocatio scheme based o user s QoS requiremets ad a power cotrol algorithm based o covex optimizatio for 9 Uautheticated owload ate 7/19/18 5:40 PM
commuicatio uder a dowli cellular etwor to improve the spectrum efficiecy. I this wor the iterferece that users brig to the cellular users is uder the threshold which the cellular users ca bear. Meawhile, the QoS requiremets of ad cellular users are guarateed respectively. This paper is orgaized as follows: the bacgroud ad the related wor of commuicatio techology is itroduced i Sectio 1; the system model ad the proposed joit resource allocatio scheme are described i details i Sectio ; simulatio results are show i Sectio 3. Fially, cocludig remars are give i Sectio 4.. System model ad the proposed scheme.1. System model This paper focuses o the sceario of multiple User Equipmets (UEs) ad a Evolved Node B (enodeb) located i the ceter of the sigle T-LTE etwor, i which a cellular li ad several lis are multiplexed o the same dowli Orthogoal Frequecy ivisio Multiplexig (OFM) RB. The two types of UEs are Cellular User Equipmet (CUE) ad User Equipmet (UE). UEs exist i pairs with a trasmitter ad a receiver. The system architecture is illustrated i Fig. 1. There are M CUEs with a set of M CUE = {CUE m m = 1,,, M} ad N pairs with a set of N = { = 1,,, N}, i which the -th T trasmitter ca be preseted as ad the -th receiver ca be preseted as. R Fig. 1. System model I order to simplify the model, some assumptios are accepted: 1. The system badwidth is divided ito equal RB, ad the umber of RBs is equal to the umber of CUEs.. Each CUE taes up oly oe RB, ad CUE m is expressed as the m-th CUE occupyig the m-th RB. 3. All the RBs are assiged to each CUE i the system to esure the CUE commuicatio requiremets preferetially due to commuicatio uder the cellular etwor. 93 Uautheticated owload ate 7/19/18 5:40 PM
4. UEs are scheduled to reuse the resources accordig to the commuicatio requiremets i each Trasmissio Time Iterval (TTI) i the system scheduler. The commuicatio situatio show i Fig. 1 shows a simple model. There is oe CUE 1 ad three pairs with two pairs ( pair1, pair) reusig the CUE 1 s RB. CUE 1 receives a sigal from enodeb ad iterferece from trasmitters UE 1 ad UE 3, as well as i the dowli commuicatio. receiver UE receives ot oly the sigal from trasmitters UE 1 i pair1, but also the iterferece from the enodeb ad aother trasmitter UE 3 i pair. The sigal ad iterferece received by receiver UE 4 is similar to UE... The proposed scheme The aim of the proposed scheme is to satisfy the user s commuicatio requiremets whe users reuse the sharig resource ad cotrol the iterferece that users brought to the CUEs i order to improve the performace of the whole system. I the scheme, the CUEs are primary users with absolute priority of the resources i the cellular system, ad users are secodary users uder the cellular etwor. 94 Fig.. Flowchart of the joit resource allocatio scheme Uautheticated owload ate 7/19/18 5:40 PM
The flowchart of the joit resource allocatio scheme is show i Fig., ad the mai steps are as follows: Step 1. Allocate all RBs to the CUEs. Step. The enodeb obtais QoS requiremets of all UEs of CUEs ad UEs by decodig the Soudig Referece Sigal (SRS) respectively. Step 3. Calculate the maximum iterferece threshold I 1,max, I,max,, I m,max, each CUE ca withstad accordig to its commuicatio requiremets, ad obtai the maximum value I = max(i 1,max, I,max,, I m,max ) i each dowli slot i each frame. Step 4. Calculate the cumulative iterferece I the -th trasmitter brought about whe reusig the m-th CUE m s RB, ad calculate the SINR γ d of R -th receiver ; T Step 5. If the cumulative iterferece I brought about whe reusig the m-th CUE m s RB is smaller tha I, go to the ext step, otherwise go to Step 10. Step 6. The -th users reuse the m-th CUE m s RB. * T Step 7. Calculate the optimal trasmittig power p of whe it reuses the m-th CUE m s RB. R Step 8. If the SINR γ of -th receiver is greater tha the predefied d d SINR threshold γ whe reusig the m-th CUE sh m s RB, go to the ext step, otherwise go to Step 10. Step 9. I m,max = I m,max I. Step 10. If all UEs are assiged resources, ed the procedure i this TTI, otherwise retur to Step 3 to cotiue for the rest of the UEs. I order to satisfy the commuicatio requiremets of the m-th CUE m whe users reuse the resource, the dowli SINR of CUE m must coform to (1) pg σ B m,bc c γ m sh. + Im I the above equatio, p B is the trasmissio power of enodeb, g m,bc is the chael gai betwee enodeb ad the m-th CUE m, σ is the average power of oise, c I m is the iterferece that the m-th CUE m ca withstad, ad γ m sh is the miimum SINR threshold of the m-th CUE m. Taig Formula (1), we have: pb gm,bc () I = σ. m,max c γ m sh The cumulative iterferece I which the -th trasmitter brigs about whe reusig the m-th CUE m s RB must coform to (3) I = p g,c Im,max. N T 95 Uautheticated owload ate 7/19/18 5:40 PM
I (3), p is the trasmissio power of the -th trasmitter, g,c is the chael gai betwee -th trasmitter ad the m-th CUE m, N is the set of pairs reusig the same resource of the cellular etwor. I order to guaratee the commuicatio requiremet of the -th receiver whe reusig the resource, the dowli SINR of pg (4) pg + σ + B,B, N, p g must coform to R, γ d sh. Here (4) g, is the chael gai betwee the -th trasmitter ad d receiver, g,b is the chael gai betwee enodeb ad the -th receiver, γ sh is the miimum SINR threshold of the -th receiver, ad p g, is N, the cumulative iterferece that trasmitters except for the -th trasmitter have brought about whe reusig the m-th CUE m s RB. The optimal trasmissio power of the -th trasmitter must be adjusted accordig to the performace of the whole system whe reusig the m-th CUE m s * RB. I order to determie the optimal trasmissio power p, a ew utility fuctio U( p ) is defied with the variable p as follows: (5) B pg, U( p ) = log 1 p g N + α pg B,B + σ + pg, N, MAX subject to 0 < p < pue,,c MAX where B is the system badwidth, p is the maximum trasmissio power of the UE trasmitter, B pg, log 1+ is the system reveue N pg B,B σ p g + +, N, obtaied by commuicatio, pg is the system cost due to the iterferece,c that commuicatio brigs about, ad α is the cost factor referred to the proportioal ifluece to the system. There is o couplig relatioship betwee each pair of, so that the trasmissio power of the trasmitter ca be optimized respectively. The optimizatio problem is a Mixed Iteger No-Liear Program (MINLP) give by (5). Calculatig the secod derivative of (5), we have (6) U( p ) B g = α g p Nl p g + p g + + p g,, B,B σ, N,,C, 96 Uautheticated owload ate 7/19/18 5:40 PM
(7) U( p ) B g, = < 0. ( p ) Nl p g, + pb g,b + σ + p g, N, U( p ) is the cocave fuctio of p due to the fact that (7) is less tha zero, ad the costrait i (5) is a covex set; therefore (5) presets covex optimizatio. The the Lagrage fuctio of the optimizatio problem be costructed based o the covex optimizatio theory [1] as follows: (8) Lp (, λ) = Up ( ) + λ( p p ), MAX UE where λ is the Lagrage multiplier. The optimal trasmissio power of the trasmitter ca be obtaied accordig to the Karush-Kuh-Tucer (KKT) coditio based o covex optimizatio. So the optimal value of p is as follows, solvig L( p the equatio, λ) = 0: p (9) p * pg B B = Nl ( αg λ),c,b + σ + N, g, Accordig to the rage of the trasmissio power of the trasmitter, the rage of α is as follows: Bg, Nlg σ + p g + g p + p g Bg α Nlg,C σ + pb g,b + p g, N, MAX,C B,B, UE, N,, < + p g,,c. λ + g,c λ. g 3. Aalysis by simulatio The simulatio sceario is show i Fig. 3. The ode with three sectors is enodeb, ad the UEs are oly distributed i oe sector of the enodeb, located i the ceter due to the sigle cell studied i the system. There are 0 CUEs with radom distributio ad 6 pairs of UEs distributed i the cell, i which CUE is represeted as a poit ad the trasmitter is represeted as a poit with a atea. 97 Uautheticated owload ate 7/19/18 5:40 PM
enodeb ad UE positios 500 400 9 3 7 18 6 16 300 8 17 y pos [m] 00 100 0-100 -00 6 4 5 15 4 10 13 713 5 3 9 11 8 1 6 5 14 1 19 0 7-300 1 10 98-400 3 1 1 4-500 11-600 -400-00 0 00 400 600 x pos [m] Fig. 3. Simulatio sceario The module of is desiged based o the Log Term Evolutio (LTE) system level simulatio platform developed by Viea Uiversity of Techology [13], ad the mai parameters i the simulatio are show i Table 1. Table 1. The mai parameters i the simulatio Parameter Value Frequecy.14 GHz Badwidth 1.4 MHz Number of RB 0 Trasmissio power of CUE 3 db.m Maximum distace betwee a 5 m pair Maximum trasmissio power of 3 db.m UE Trasmissio power of enodeb 46 db.m Coverage of enodeb Number of CUE 0 Number of pairs of 6 esity of oise Simulatio time Types of the traffic Schedulig algorithm of CUE Chael model 500 m 174 (db.m)/hz 1000 TTI full buffer Proportio Fair (PF) suburba I order to validate the performace, we compare the proposed scheme with the schemes of Radom, Max-CQI ad Iterf-Mi. The scheme of Radom refers to assigig the resource to UE radomly. Max-CQI refers to assigig the resource with a maximum Chael Quality Idicator (CQI) to UE. Iterf-Mi refers to Uautheticated owload ate 7/19/18 5:40 PM
assigig the resource to UE with the miimum iterferece to CUE. The trasmissio power of the trasmitter is adjusted by the utility fuctio i the proposed oe, while the others have full power. Compariso of the Cumulative istributio Fuctio (CF) with the average throughput of UE is show i Fig. 4. It is see i the figure that the performace of the suggested scheme is the best oe. The average throughput of UE based o the Proposed sheme is stable from 0. Mb/s up to 0.5 Mb/s, while that of the other three schemes are relatively dispersed, ad the overall throughput of the Proposed scheme is the greatest oe. 1 0.9 0.8 0.7 Radom Iterf-Mi Max-CQI Proposed 0.6 CF 0.5 0.4 0.3 0. 0.1 0 0 0.05 0.1 0.15 0. 0.5 Average throughput of /Mbps Fig. 4. Compariso of CF with the average throughput of UE Compariso of the CF with the average throughput of CUE is show i Fig. 5 which shows that the differece betwee the maximum ad miimum throughput of CUE is greater compared to Fig. 3. Sice CUE is distributed radomly i the simulatio sceario, a CUE close to enodeb will obtai a greater throughput. Cotrary to this, the throughput is smaller if the CUE is far away from enodeb. I additio, the miimum ad maximum throughput of the Proposed scheme are improved respectively for CUE compared to the other three schemes, ad the overall throughput of the Proposed oe is the greatest. This is due to the fact that the performace of CUE ca be guarateed whe the resource is reused ad the iterferece is cotrolled by adjustig the trasmissio power of UE through the utility fuctio. 1 0.9 0.8 0.7 0.6 Radom Iterf-Mi Max-CQI Proposed CF 0.5 0.4 0.3 0. 0.1 0 0 0.5 1 1.5.5 3 3.5 4 Average throughput of CUE/Mbps Fig. 5. Compariso of CF with the average throughput of CUE 99 Uautheticated owload ate 7/19/18 5:40 PM
Compariso of CF with the average throughput of all UEs is illustrated i Fig. 6, ad the compariso of the average throughput with differet types of UE is displayed i Fig. 7 by a histogram. 1 0.9 0.8 0.7 0.6 Radom Iterf-Mi Max-CQI Proposed CF 0.5 0.4 0.3 0. 0.1 0 0 0.5 1 1.5.5 3 3.5 4 Average throughput of all UEs/Mbps Fig. 6. Compariso of CF with the average throughput of all UEs The average throughput of the Proposed scheme is the best for CUE, UE ad all UEs respectively from Figs 6 ad 7, because the resource is reused as much as possible to ehace the total throughput without affectig the commuicatio requiremet of CUE. Meawhile, the trasmissio power of UE is optimized by balacig the reveue (system throughput) ad the cost (iterferece itroduced by ). The average throughput of the Radom scheme is the smallest oe, because the resource is reused by UE radomly without cosiderig the chael coditio ad the iterferece is possibly produced by multiple UEs reusig the same resource. The average throughput of Iterf-Mi is greater tha Radom, because the resource reused by UE is the resource with miimum iterferece to CUE. The average throughput of Max-CQI is slightly greater tha Iterf-Mi because the chael coditio is reusig the resource of the maximum CQI. Average throughput/mbps 0.7 0.6 0.5 0.4 0.3 0. Radom Max-CQI Iterf-Mi Proposed 0.1 0 CUEs UEs all UEs Fig. 7. Compariso of the average throughput with differet types of UE 100 Uautheticated owload ate 7/19/18 5:40 PM
Ivolvig commuicatio i the system leads to iterferece to the CUEs, ad some commuicatios of CUEs may be bloced because they caot meet the miimum requiremets. I order to compare the commuicatio effect of CUEs with differet algorithms, the CUE Satisfactio egree (CS) is defied as follows: Numbers of CUEs with commuicatio successfully CS=. Total umbers of CUEs I Fig. 8 the powers of the UEs ca be adjusted with the Proposed scheme to reduce the iterfereces to the CUEs, so that CS with the Proposed oe is comparable with Iterf-mi. The iflueces of commuicatio to the CUEs are ot cosidered with Radom ad Max-CQI, so that the CSs are smaller tha the Proposed oe. Max-CQI reuses the maximum CQI without cosiderig the iterfereces to the CUEs, so that CS is the smallest oe. CUE Satisfactio egree 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0. 0.1 0 Radom Max-CQI Iterf-mi Proposed Fig. 8. Compariso of CUE Satisfactio egree with differet schemes 4. Coclusio This paper proposes a joit resource allocatio scheme based o the user s QoS requiremet by aalyzig the iterferece i commuicatio uder a cellular etwor. The resource allocatio is based o a greedy algorithm ad the trasmissio power of is optimized by the covex optimizatio theory to balace the ehaced throughput ad the iterferece brought by commuicatio. The simulatio shows that ot oly the user s QoS requiremet ca be guarateed, but also the system throughput is improved with the users balace. However, this wor oly cosiders a sigle cell model without cosiderig the iterferece betwee multiple cells. For a further study, we are iterested i studyig the resource allocatio scheme i multiple cells. Acowledgmets: This wor was supported i part by a Grat from the Project of Educatio ad Scietific Research of Youg Teacher of Fujia (Grat No JA1433, Grat No JA1444), Natioal Natural Sciece Foudatio of Chia (Grat No 610013) ad Natural Sciece Foudatio of Fujia (Grat No 015J01670). 101 Uautheticated owload ate 7/19/18 5:40 PM
Refereces 1. Z h i, G., F. Z e g, H. L i u et al. Realizatio of the Commuicatio betwee evices of BANET/IP ad Real-Time atabase. Review of Computer Egieer Studies, Vol. 1, 014, No 1, pp. 1-6.. M u m t a z, S., L. L. Y a g, C. W a g et al. Smart-evice-to-Smart-evice Commuicatios: Part I. IEEE Commuicatios Magazie, Vol. 5, 014, No 4, pp. 18-19. 3. X u, S., H. W a g, T. C h e et al. Effective Iterferece Cacellatio Scheme for evice-to- evice Commuicatio uder Cellular Networs. I: Proc. of IEEE 7-d Vehicular Techology Coferece Fall (VTC 010-Fall), Ottawa, 010, pp. 1-5. 4. X u, C., L. S o g, Z. H a et al. Iterferece-Aware Resource Allocatio for evice-to-evice Commuicatios as a Uderlay Usig Sequetial Secod Price Auctio. I: IEEE Iteratioal Coferece o Commuicatios (ICC 01), Ottawa, 01, pp. 445-449. 5. N i e, G., B. L u, G. Y a g. A Method of Commuicatio Iterferece Suppressio Based o Reusig the owli Wireless Resource of Cellular Networ. Telecommuicatios Networ Techology, 01, No 8, pp. 54-59. 6. Z h u, X., S. W e, G. C a o et al. QoS-Based Resource Allocatio Scheme for evice-to-evice () Radio uder Cellular Networs. I: 19th Iteratioal Coferece o Telecommuicatios (ICT 01), Jouieh, 01, pp.1-6. 7. Yu, C. H., O. Tiroe, K. oppler et al. O the Performace of evice-to-evice Uderlay Commuicatio with Simple Power Cotrol. I: IEEE 69th Vehicular Techology Coferece Sprig (VTC 009-Sprig), Barceloa, 009, pp. 1-5. 8. Yu, C. H., O. Tiroe, K. oppler et al. Power Optimizatio of evice-to-evice Commuicatio uder Cellular Commuicatio. I: IEEE Iteratioal Coferece o Commuicatios (ICC 009), resde, 009, pp.1-5. 9. G a o, W., H. Y i, Y. H u et al. Resource Allocatio Algorithm Based o SNR Equalizatio i Commuicatio. Computer Egieerig, Vol. 38, 01, No 10, pp. 5-8. 10. R o g, T. The Study o Commuicatio Techologies. M. S. Thesis, Najig Uiversity of Posts ad Telecommuicatios, Najig, 013. 11. Gu, J., S. J. Bae, B. G. Choi et al. yamic Power Cotrol Mechaism for Iterferece Coordiatio of evice-to-evice Commuicatio i Cellular Networs. I: 3rd Iteratioal Coferece o Ubiquitous ad Future Networs (ICUFN 011), alia, 011, pp. 71-75. 1. S. Boyd, L. Vadeberghe, Eds. Covex Optimizatio. Cambridge, Cambridge Uiversity Press, 004. 13. Istitute of Telecommuicatios Viea Uiversity of Techology. Viea LTE Simulators System Level Simulator ocumetatio v1.6r885. I: Istitute of Telecommuicatios, Viea Uiversity of Techology, 01, pp. 4-7. http://www.t.tuwie.ac.at/ltesimulator/. 10 Uautheticated owload ate 7/19/18 5:40 PM