Optimal Relay Placement for Coverage Extension in LTE-A Cellular Systems

Transcription

1 Optimal Relay Placement for Coverage Extension in LTE-A Cellular Systems Suman Khakurel, Mahima Mehta, Ahay Karanikar Information Networks La, Department of Electrical Engineering Inian Institute of Technology Bomay, Mumai, Inia {suman, mahima, Astract Thir Generation Partnership Project (3GPP Long Term Evolution-Avance (LTE-A has consiere the eployment of Relay Noes (RNs for cost-effective throughput enhancement an coverage extension The coverage extension (increase in cell raius epens on the raial position of RNs in the cell This is ecause the location of a RN affects the Signal-to- Interference-plus-Noise Ratio (SINR of the receive signal on the evolve-noeb (enb-rn an RN-User Equipment ( links In this paper, we investigate the prolem of optimal relay placement for coverage extension in relay assiste LTE-A networks Since DownLink (DL an UpLink (UL transmission scenarios in cellular networks are asymmetrical in terms of coverage (ue to iscrepancy in maximum transmit power, we consier oth DL an UL transmission scenarios for optimal relay placement In aition, we analyze the prolem for the case when interference from neighouring cells is taken into account I INTRODUCTION Long Term Evolution-Avance (LTE-A is a technological avancement propose y the Thir Generation Partnership Project (3GPP to meet the requirements of Fourth Generation (4G moile roaan system The unerlying raio technology of LTE-A networks is ase on Orthogonal Frequency Division Multiple Access (OFDMA which has inherent immunity to the averse consequences of frequency selective faing Moreover, Multiple Input Multiple Output (MIMO techniques an coorination among multiple cell sites calle Coorinate MultiPoint (CoMP transmission/reception are regare as the key techniques to meet the requirements of 4G in LTE-A [1] Yet, capacity at the cell ege remains relatively small ue to low Signal-to-Interference-plus-Noise Ratio (SINR at the cell ege compare to inner regions of the cell [2] Therefore, a cost effective solution of eploying Relay Noes (RNs is propose in LTE-A Deploying RNs in a cellular network has two key enefits: cell capacity improvement an cell coverage extension RNs can provie higher cell capacity in a given cell area eacuse of the link iversity Link iversity is achieve ecause of the two possile links etween User Equipment ( an Base station (terme as evolve-noeb (enb in 3GPP-LTE: irect link (etween to enb or the inirect link (etween to enb via RN Alternately, RNs help increase the cell coverage (cell raius for the same cell capacity Increment in cell raius ue This work is supporte y the Inia-UK Avance Technology of Centre of Excellence in Next Generation Networks (IU-ATC project an fune y the Department of Science an Technology (DST, Government of Inia to eployment of RNs reuces infrastructure cost of eploying more enbs The coverage extension ue to RN eployment is achieve since RNs provie etter SINR to the cell ege s compare to enb ue to their proximity to s The increase in cell raius epens on the raial position of RNs in the cell This is eacuse the location of a RN affects the SINR of the receive signal on the enb-rn (ackhaul an RN- (access links Deploying a RN away from cell ege causes low SINR on the access link (ue to increase in path loss This increases the outage proaility (proaility that the receive signal strength is elow acceptale threshols of cell ege s On the other han, eploying a RN near cell ege will result in higher interference to the neighoring cells Therefore, an optimal location for relay placement nees to e etermine for maximizing cellular coverage Limite contriution is availale in the literature which aresses the prolem of optimal RN placement to maximize coverage extension In [3] an [4], the authors analyze the RN placement issue in IEEE 80216j Worlwie Interoperaility for Microwave Access (WiMAX networks from the perspective of increasing system capacity In [5], optimal RN placement issue is aresse from the perspective of cellular coverage extension Authors efine the cellular coverage in terms of proaility of correct ecoing an use an iterative algorithm to evaluate the optimal RN placement for Downlink (DL transmission scenario Since DL an UpLink (UL transmission scenarios in cellular system are asymmetrical in terms of coverage (ue to iscrepancy in maximum transmit power, the optimal relay placement will e ifferent for each of the scenarios consiere iniviually An optimal RN placement location evaluate with only DL into consieration can aversely affect the performance in UL scenario an vice-versa Hence, a joint optimization prolem nees to e formulate with oth DL an UL scenario into consieration in orer to etermine the optimal RN placement in cellular systems In this paper, we exten the work in [5] to formulate a joint optimization prolem We analyze this prolem within the framework of LTE-A stanars [6] However, it can e applie to any cellular system We efine coverage raius of the cell in terms of proaility of correct ecoing at a point, as efine in [5] We etermine the optimal location for RN placement to achieve maximum coverage raius for the scenarios with an without interference We also analyze the variation in optimal

2 location of RN placement for varying ecoing threshols an loa conitions The optimal relay placement location an the extene coverage raius evaluate in this paper can e practically use as a system esign parameter since it is optimize with oth UL an DL scenarios into consieration The rest of the paper is organize as follows In Section II, we escrie the system moel Section III gives the prolem formulation, illustrates the optimal relay placement methoology for scenarios with an without interference Section IV presents the results an inferences an finally, Section V conclues the paper II SYSTEM MODEL We consier a two hop LTE-A cellular system The system moel comprises a reference cell surroune y the first tier of co-channel cells Each cell consists of N numer of RNs place symmetrically aroun enb (at the istance R from enb The maximum transmit power of enb, RN an are enote y P enb, P RN an P an the antenna gains of enb, RN an are enote y G enb, G RN an G respectively The maximum transmit power values are 46, 30 an 24 Bm an antenna gains are 16, 5 an -1 B for enb, RN an respectively as per the LTE-A specifications [6] We consier log normal shaowing ξ on oth the access an ackhaul links, where ξ is a Gaussian ranom variale with mean 0 an stanar eviation an for the access an ackhaul link, respectively We ignore the impact of fast faing as our ojective is to evaluate the value of optimal RN placement from a long term prospect Therefore, the evaluation carrie out in this paper is applicale to oth Frequency Division Duplex (FDD an Time Division Duplex (TDD ase LTE-A systems III OPTIMAL RELAY PLACEMENT FOR COVERAGE EXTENSION We formulate the prolem in Section III-A an egin our analysis in Section III-B for an interference-free scenario where the impact of Inter-Cell Interference (ICI is neglecte Then, in Section III-C, we analyze the prolem of optimal relay placement consiering the impact of ICI from the first tier of co-channel cells A Prolem Formulation The authors in [5] efine coverage raius in terms of proaility of correct ecoing in DL (pc at a point For the DL scenario of a single hop network pc is given y: pc = Pr(SINR enb > T, (1 = Pr( enb +G +ξ 10ηlog I N > T, T +N +I enb G +10ηlog = Q σ where is the istance etween enb an an Q(x = 1 2π exp( x2 2 x The authors suggest that for every point with x pc 05, the expecte value of the receive SINR at is greater than the ecoing threshol T Then, the coverage Symol M,N R an N N sc R R a P enb,p RN,P enb, RN, ξ, pc,pc u pc, pc a pc u, pcu a SINR, SINR a SINR u, SINRu a G,G RN an G enb I,IRN I enb,irn u T N η TABLE I LIST OF NOTATIONS Description Numer of enbs, RNs an s in the network Numer of sucarriers in the network Distance etween RN an enb (ackhaul link Distance etween RN an (access link Maximum transmit power of enb, RN an Maximum transmit power of enb, RN an on each sucarrier Log-Normal Shaowing parameter Stanar Deviation of Shaowing for access an ackhaul link Proaility of correct ecoing for DL an UL Proaility of correct ecoing for ackhaul an access link in DL Proaility of correct ecoing for ackhaul an access link in UL SINR on ackhaul an access link in DL SINR on ackhaul link an access link in UL Antenna gains of, RN an enb Interference at an RN in DL Interference at enb an RN in UL Threshol value of SINR Noise level proaility of sucarrier activity path loss exponent raiusrcov is the istance from enb at which experiences pc = 05, such that all locations of the at a istance > Rcov from the enb experience pc < 05 Now, for a two hop cellular network, the proaility of correct ecoing in DL is given y: pc = pc pc a, (2 = Pr(SINR enb RN > T Pr(SINR RN > T, T +N +I = Q( RN enb G RN +10ηlogR ( T +N +I RN G +10ηlogR a Q wherer anr a are the istances from enb to RN an RN to For this two-hop cellular network, coverage raius Rcov is the maximum istance from enb at which transmission via a RN results in pc = 05 Authors in [3] suggest that the optimal location to eploy a RN must lie on the line joining enb an Thus, Rcov = R + Ra where, R is the optimal RN placement raius in DL an Ra is the RN- istance in DL such that pc = pc pc a = 05 [5] However, UL transmission scenario in a cellular system is asymmetrical compare to DL in terms of maximum transmit power an hence, coverage The maximum transmit power of s in UL is less compare to the maximum transmit power of enb in DL Hence, the optimal relay placement with only DL scenario into consieration may aversely affect the UL performance Deploying a relay at R may not even support two way communication (as the proaility of correct ecoing in UL (pc u is likely to fall elow 05 Therefore, we formulate a joint optimization prolem using DL an UL transmission scenario an fin the optimal location for relay

3 placement For the UL transmission scenario of two hop relay networks, proaility of correct ecoing is given y: pc u = pc u a pcu, (3 = Pr(SINR RN > T Pr(SINR RN enb > T, ( T +N +I u = Q RN G RN +10ηlogRa u T +N +IeNB RN G enb +10ηlogR Q( u It can e seen from (2 an (3 that for a given value of pc, R a is a function of R Thus, there is a trae-off etween the values of R a an R Therefore, we nee to etermine the value of R which maximizes the cell coverage such that the proaility of correct ecoing in oth UL an DL is greater than or equal to 05 (ie, pc u 05 an pc 05 Therefore, the optimal relay placement location (R is given as: R = argmax min((r +Ra u,(r +Ra s t (4 R (0,R max min (pc u pc u a,pc pc a = 05 (5 where R max = min (R umax,r max an,r umax anr max are the maximum possile relay placement istances for UL an DL This implies that the value of pc u is 05 when RN is place at R umax an the value of pc is 05 when RN is place at R max If Rcov is the maximum coverage extension an R is the optimal relay placement Then, R a is given as Rcov R an the numer of relays (N is given y [5]: π N R = (6 sin 1 ( R a R B Analysis in Scenario without Interference For an interference-free scenario, we neglect the interferences receive from neighouring cells Therefore,I =IRN = IRN u = I enb = 0 We etermine the value of R, R cov an N using following steps: 1 Determine the value of R umax an R max An, R max = min (R umax,r max 2 For R = 1, compute Ra u such that pc upc a u = 05 Then, assign Rcov u = R + Ra u 3 For R = 1, compute Ra such that pc pca = 05 Then, assign Rcov = R + Ra 4 R cov (0 = min (Rcov,R u cov 5 Repeat the Steps 2 an 3 R (1,R max ] an form the array R cov 6 Finally compute Rcov = max (R cov an optimal relay placement istance R is the value of R corresponing to Rcov π Also, N R = sin 1 ( R a R C Analysis in Interference Scenario We compute the ICIs in DL an UL separately since the possile interferers in DL an UL are ifferent For DL transmission, I RN at the reference RN an I at the reference are compute as shown in Fig 1 The total interference power receive at the reference RN is the sum of intereference powers from neighouring enbs From Fig 1, it can e seen that there are 6 neighouring enbs Therefore, IRN = Ir i = ( enb +G RN η i (7 where Ir i an i are the interference power an istance from thei th enb to the reference RN uring DL transmission Also, I = Iu i = ( RN +G η i,r N (8 R r=1 Here, I i u is the interference power receive at the reference from RNs of the i th neighoring cell an i,r is the istance etween the reference an the r th RN of the i th neighoring cell For simplicity of analysis, we consier only path loss uring calculation of ICIs enb6 5 enb5 RN8 RN4 6,4 6 7,8 enb7 enb1 enb4 RN enb enb2 enb3 Fig 1 Interference scenario in DL: Dashe lines enote istances i from the enb of i th neighoring cell to the reference RN Soli lines enote istances i,r from r th RN of i th neighoring cell to the reference Similarly, we compute the interference power I enb at the reference enb uring UL as shown in Fig 2 The total interference power receive at the reference enb is the sum of interference powers from RNs of the neighoring cells So, I enb = 7 Ie i = 7 N R ( RN +G enb e η i,r (9 r=1 Here, Ie i is the interference power receive at the reference enb from RNs of the i th neighoring cell an e i,r is the istance etween the reference enb an the r th RN of the i th neighoring cell Also, IRN u is the interference power receive at the reference RN from s of the neighoring cells To calculate IRN u, we assume that s are uniformly istriute in the cell There can e maximum one in the neighoring cell i which interferes with the transmission of the reference If N is the total numer of s in each cell then, IRN u = Iru i = N u=1 (P +G RN r η i,u (10

4 where I i ru is the interference power from s of the i th neighoring cell to the reference RN an r i,u is the istance etween the reference RN an the u th of i th neighoring cell We assume that the sucarrier allocation algorithm is such that each sucarrier has proaility 1/N of eing allote to a in a cell which justifies the factor 1/N in (10 enb6 enb5 RN3 RN4 e 5,4 r 7,1 e 6,3 enb7 RN6 r 7,2 enb1 enb4 enb2 RN enb enb3 Fig 2 Interference scenario in UL: Dashe lines enote istances e i,r from r th RN of i th neighoring cell to the reference enb Soli lines enote istances r i,u from u th of i th neighoring cell to the reference RN In case of interference scenario, SINR at the receiver is not only epenent on transmit power an antenna gains ut also on the numer of RNs in each cell an the istance etween transmitter an receiver Therefore, the value of R cov is require to eterminer a for a given value of R Therefore, we use the moifie version of the iterative algorithm propose in [5] to calculate the optimal relay placement location The algorithm uses the value of R cov calculate from Section III-B to compute the value of Ra u an R a as a function of relay placement istance R Then, we etermine the maximum value of min(r + Ra u,r + Ra for all possile values of R an assign this as the new value of R cov IV RESULTS AND INFERENCES In this section, we provie the numerical results corresponing to the analysis of optimal relay placements in Section III-B an III-C The value of maximum transmit power an antenna gains for enb, RN an are ase on LTE-A specifications [6] The value of Noise level (N is consiere to e 100 Bm, ecoing threshol (T is 42 B, numer of s (N is 120, an numer of sucarriersn sc is 512 Stanar eviation of shaowing on access an ackhaul links are assume to e 6 an3b respectively Also, proaility of sucarrier activity an path loss exponent η are taken to e 1 an 35 respectively We assume that there is equal power ivision amongst sucarrier Therefore, enb = PeNB The sucarrier allocation algorithm is such that each RN gets equal numer of sucarriers Hence, RN = PRN NR Also, Algorithm I: Iterative calculation of R, R cov an N R cov (1 R cov (noint :value of Rcov otaine from Section III-B R cov (0 0 N (1 R = N(noint R :value of N R otaine from Section III-B i 1 while R cov (i R cov (i 1 > ǫ o Comment : For Downlink Scenario for each R (0,R cov] (i R cov {φ} Compute( IRN an I pc = Q T+N+I RN enb GRN+10ηlogR flag = 0 if pc < 05 then set flag = 1 an reak the loop en if pc T+N+I a = Q( RN G +10ηlogR a Solve pc a pc = 05 for R a Appen Ra + R R cov en for if flag = 1 then Appen 0s to R cov to fill the remaining array en if Comment : For Uplink Scenario for each R (0, cov] R u cov {φ} Compute I u RN an I enb pc u = Q T+N+IeNB RN G enb+10ηlogr flag = 0 if pc u < 05 then set flag = 1 an reak the loop en if pc u a = Q T+N+I u RN GRN+10ηlogRu a Solve pc u a pcu = 05 for Ru a Appen Ra u + R R u cov en for if flag = 1 then Appen 0s to R u cov to fill the remaining array en if R cov = min (R cov, R u cov i i+1 cov = max R cov N (i = arg max R cov R = π en while sin 1 ( R(i cov R(i each sucarrier has an equal proaility of eing assigne to a, as mentione in Section III-C So, we compute as P N In Fig 3, we plot the value of extene cell raius (R cov against relay placement raius (R for interferencefree scenario with oth DL an UL into consieration It can e seen from the figure that the optimal raial position for

5 Extene Cell Raius R cov (in m Relay Placement Raius from enb R (in m Extene Cell Raius R cov (in m T = 42 B T = 40 B T = 38 B Numer of Iterations Fig 3 Plot of extene cell raius (R cov versus RN placement istance (R for DL/UL scenario wihtout interference into consieration Fig 5 Plot showing convergence of extene cell raius (R cov for Algorithm-I to evaluate the optimal RN placement location for DL/UL scenario with interference into consieration for varying ecoing threshols relay placement (R is 243 km, R cov is 33 km an the numer of RNs require (N R is 9 (from (6 In orer to fin the optimal relay placement with ICI into consieration, we use Algorithm I The value of Rcov an N R calculate for intereference-free scenario are fe as the initial values in Algorithm I (ie, R cov (noint = 243 km an N (noint = 9 Forǫ=001 an =1, the value ofr cov converges to13 km (Fig 4 The corresponing value of optimal RN placement location is 0775 km an the numer of RNs require is 5 In Fig 4, we also oserve the impact of on R cov = 1 represents a worst case scenario where all the sucarriers in a cell are eing use Therefore, reuction in the value of causes lower interference on the reference cell an hence, increases the cell coverage Fig 5 illustrates the convergence of R cov for various values of ecoing threshol We can oserve from the figure that the cell coverage raius increases with the ecrease in ecoing threshol V CONCLUSIONS In this paper, we have suggeste that the optimal relay placement location evaluate with only DL into consieration may egrae the UL performance This is ecause of the fact that DL an UL transmission scenarios are asymmetrical in terms of maximum transmit power an hence, coverage Therefore, it is essential to consier oth DL an UL transmission scenarios to calculate the optimal RN placement location for cellular coverage extension In this paper, we have propose the algorithms to calculate the optimal RN placement location in scenarios with an without intereference We have also illustrate the effect of ecoing threshol an proaility of sucarrier activity on cell raius Though, we have simulate the algorithms with LTE-A specifications, they can e applie to non-lte-a ase cellular networks as well The optimal relay placement location an the extene coverage raius calculate in this paper can e practically use as a system esign parameter in cellular systems Extene Cell Raius R cov (in m p = 1 act p = 08 act p = 06 act Numer of Iterations REFERENCES [1] 3GPP TR 36814, Further Avancements for E-UTRA Physical Layer Aspects, March 2010 [2] 3GPP TR 36913, Requirements for Further Avancements for Evolve Universal Terrestrial Raio Access (E-UTRA, March 2009 [3] B Lin, P H Ho, L L Xie, an X Shen, Optimal Relay Station Placement in IEEE 80216j Networks, in International Conference on Wireless Communications an Moile Computing, 2007, pp [4] C Y Chang, C T Chang, M H Li, an C H Chang, A Novel Relay Placement Mechanism for Capacity Enhancement in IEEE 80216j WiMAX Networks, in proceeings of IEEE International Conference on Communications, 2009 [5] G Joshi an A Karanikar, Optimal Relay Placement for Cellular Coverage Extension, in NCC, January 2011 [6] 3GPP TSG-RAN WG1 #54is R , Text Proposal for Evaluation Methoology, 2008 Fig 4 Plot showing convergence of extene cell raius (R cov for Algorithm-I to evaluate the optimal RN placement location for DL/UL scenario with interference into consieration for varying