Transcription

International Journal of Electronics Communication and Computer EngineeringVolume 7, Issue 5, ISSN (Online): 2249–071XEstimation of Number of Simultaneously SupportedInteractive VoIP and Video Sessions in LTE-AdvancedMarko Porjazoski, Borislav PopovskiAbstract – LTE-Advanced has engaged several techniquesin order to enhance system performance enhancementscompared to LTE. In this paper we evaluate LTE-Advancedcapacity in the means of a number of simultaneouslysupported interactive VoIP and video sessions. Some newfeatures that enable the capacity enhancements that areprovided in LTE-Advanced like spatial multiplexing andCarrier Aggregation are overviewed. In order to calculatethe number of supported users, we determine the number ofresource elements available for user data transmission indownlink and uplink, taking in account all the resourceelements used for redundant information. The maximalnumber of simultaneously supported interactive VoIP andvideo sessions is calculated assuming 64-QAM modulationwith ¾ code rate for all transmissions. Finally the number ofparallel VoIP and video sessions is calculated and comparedfor different LTE-A scenarios.Keywords – LTE-Advanced, Spatial Multiplexing, CarrierAggregation, Capacity, Interactive VoIP, Interactive Video.Usually mobile systems performances are evaluated bycomputer simulations. It is rare to find in literature ananalytical method for calculation of number of supportedusers in mobile networks. So in our work we are trying toestablish a framework for calculation of number ofsimultaneous sessions support in LTE-Advanced network.The paper is organized as follows. In Section II weprovide an overview of LTE and LTE-A radio frame indownlink and uplink, in order to introduce the resourcegrid and estimate the number of resource elements whichare available for user data transmission. We use theseresults in Section III to evaluate and compare themaximum number of simultaneous VoIP sessions andsimultaneous video sessions when using 64QAM with ¾code rate, for different scenarios of MIMO and CA use. InSection IV we propose a method for estimating thenumber of parallel simultaneously supported VoIP andvideo users for previously discussed scenarios. Finally,Section V concludes the paper.I. INTRODUCTIONII. LTE-ADVANCED RADIO FRAME3GPP Release 10, system known as LTE-Advanced(LTE-A) enhances the capabilities of LTE (Long TermEvolution), in order to satisfy the InternationalTelecommunication Union’s requirements for IMTAdvanced[1]. In [2] a list of LTE-A requirements can befound. In order to meet the listed requirements, which arerelated to the system performance, LTE-A introduces newfeatures in physical layer, including enhanced multiantenna support (MIMO), to increase the data rate, CarrierAggregation (CA), to provide wider bandwidth andsupport of heterogeneous networks, to improve thecapacity and coverage [3].In this paper the focus is on improvements that areachieved using MIMO and CA by analyzingsimultaneously supported VoIP and video sessions.One of the new technique that is used in LTE-A toimprove performances is spatial multiplexing [1]. Inspatial multiplexing the transmitter and receiver both usemultiple antennas to establish multiple parallel streams, soas to increase the data rate or the number of supportedusers nearly by the number of antennas used. Thetransmitted data streams can be directed to one set ofusers, introducing the Single User MIMO (SU-MIMO),used to increase the data rate of just one set of users, orthey can be directed to multiple different users leading toMulti User MIMO (MU-MIMO) scenario, which allowsincreasing the overall capacity[4].CA enables bandwidth extension up to 100MHz, whichis the maximum carrier aggregation of five 20 MHzComponent Carriers (CC). However, this is not verycommon, due to the operator’s limited bandwidth [5].Release 10 features are designed to be backward compatible with Release 8, so the physical layer radioframe remains the same as the one used in LTE Release 8.In order to perceive the enhancements gained using multiantenna techniques and CA, we will give an overview ofthe radio frame in LTE downlink (DL) and uplink (UL).LTE physical layer in DL is based on OrthogonalFrequency Division Multiple Access (OFDMA), while inUL it is based on Single Carrier Frequency DivisionMultiple Access (SC-FDMA) [6]. LTE supports bothFrequency Division Duplex (FDD) and Time DivisionDuplex (TDD) schemes, which results in different framestructures in time domain [7].In this paper we will use the FDD scheme. BothOFDMA and SC-FDMA techniques are based on fixedframe-based transmission. In time domain, the radioframes are 10ms long (T frame 10ms ) , and are dividedinto 10 sub-frames each with 1ms duration. Each subframe is composed of two time-slots with 0.5ms duration.The resource allocation is organized in resource blocks(RB), which is the minimum amount of resources that canbe assigned to a user. Users are multiplexed in time andfrequency by allocating different RBs. Each RB is a twodimensional structure containing resource elements (RE).In our work we will use the most common configurationof RB, as follows. In time domain the RB corresponds toone time slot with 0.5ms duration and contains 7 symbols.In frequency domain the RB is divided into 12 arriers*15kHz 180kHz spacing for one RB. ACopyright 2016 IJECCE, All right reserved275

International Journal of Electronics Communication and Computer EngineeringVolume 7, Issue 5, ISSN (Online): 2249–071Xresource element (RE) is defined as one symbol (0.5ms/7)in time domain and 1 subcarrier (15kHz) in frequencydomain [8]. Fig. 1 illustrates the structure of the resourcegrid of LTE.LTE supports different frequency bands: 1.4, 3, 5, 10,15 and 20MHz. Depending on the frequency band used,the number of RB varies from 6 RB per time-slot (for 1.4MHz) to 100 RB per time-slot (for 20MHz).The radio interface capacity is used both for user datatransmission and for redundant information such astransmission of control, synchronization and broadcastingsignals.Fig. 2. RE mapping for the cell-specific reference signals,normal cyclic prefixFig. 1. Resource grid structure [7]All the channels used for transmission of the requiredinformation are mapped onto the structure of REs. Themapping scheme is different in DL and UL.A. Resources Allocation in DLAn illustration of DL configuration for 20MHz FDDwith 15kHz is given in [7].Considering the structure of the RE mapping scheme,we can derive a general equation for calculating theREnumber of REs used for data transmission ( N DATA ) indownlink as:RERERERERERENDATA NTOTAL NPDCCH NREF NSYNC NPBCH(1)REwhere N TOTAL is the total number of REs in one radioframe, NREPDCCHis the number of REs for transmission ofPhysical Downlink Control Channel (PDCCH),REN REFREand N SYNC are the numbers of REs used for reference Eand N PBCH is the number of REs used for transmission ofPhysical Broadcast Channel.The number of resource elements occupied bysynchronization signals and PBCH does not depend on thechannel bandwidth (number of RBs), it is always constant.The synchronization signals are transmitted with 5msperiod, on six central RBs and occupy the first twosymbols in one RB. PBCH is transmitted with 10msperiod, on six central RBs and occupy four symbols in oneRB.The number of REs occupied by PDCCH and referencesignals depends on the used bandwidth. PDCCH signalsare transmitted with 1ms period, on all RBs and occupythree symbols in one RB.The mapping of the reference signals, which are usedfor antenna ports, depends on the number of antenna portsused by the base station and the antenna port number.Fig. 2 shows the resource allocation configuration forone, two and four antenna ports. In case of spatialmultiplexing, while one antenna is transmitting a referencesignal, all others stay silent, in the manner required forspatial multiplexing [1]. That means that the use of spatialmultiplexing downsizes resources for data transmission,even though not all the resources reserved for redundantinformation have to be used.B. Resources Allocation in ULAn illustration of UL configuration for 20MHz FDDwith 15kHz is given in [7].Analog to (1), we can derive an equation for calculatingthe number of RE used for data transmission in UL:REREREREN DATA N TOTAL N PUCCH N REF(2)where NRETOTALis the total number of REs in one radioframe, NREPUCCHis the number of REs for transmission ofREPhysical Uplink Control Channel (PUCCH) and N REF arethe numbers of REs used for Demodulation and SoundingReference Signals transmission respectively.The number of resource elements occupied by PUUCHand Demodulation Reference Signal for PUCCH does notdepend on channel bandwidth (number of RBs), it alwaysremains the unchanged. PUCCH and DemodulationReference Signals for PUCCH are transmitted with 2msperiod, on the first and the last RBs and occupy one RB.The number of REs occupied by DemodulationReference Signals for PUSCH and Sounding ReferenceSignals depends on the bandwidth used. DemodulationReference Signals for PUSCH are transmitted with 0.5msperiod, on all RBs and occupy two symbols in one RB.Sounding Reference Signals are transmitted with 1msperiod, on all RBs and occupy one symbol in one RB.Using equations (1) and (2) we calculated the number ofREs used for user data transmission in DL and UL,respectively. The results for different channel bandwidthsare presented in Table I.Copyright 2016 IJECCE, All right reserved276

International Journal of Electronics Communication and Computer EngineeringVolume 7, Issue 5, ISSN (Online): 2249–071XIII. VOIP AND VIDEO TRANSMISSION OVERLTE-AWe define LTE-A voice and video capacity as numberof simultaneous voice and video calls carried by LTE-A.In [9] the procedure for determining this capacity is given.We will use 64 QAM modulation ( N bits / symbol 6) and¾ code rate (CR).Having obtained the average number of RE per RB fordatadata in both DL and UL - N RE / RB (Table I), we cancalculate the average number of bits carried by one RB.The total number of resource blocks available duringonevoice/videopacketgenerationperiod,voice/ video periodN RBis calculated as:voice / video periodN RB Tvoice / video* NRB / frameTframe(4)Finally, the total number of voice/video calls served byLTE-A can be calculated as:N voice / video voice / video periodN RBvoice / video packetN RB(5)where N voice / video is the number of voice or video userssimultaneously served by LTE.Fig. 3. Resource mapping for video connections, 1.4MHzchannel bandwidth, no MIMOCA requires more complicated signaling. However,there are no differences in the number of reservedresources, in regard to resource allocation. That means thatthe use of CA multiplies the available resources by thenumber of CC used. The bandwidths of the CCs do nothave to be equal, it is possible to use CCs with differentbandwidths and the resources of the additional CCs arejust added to the primary CC.Taking into account that VoIP and video are real timeservices, spatial multiplexing can only influence byincreasing the number of supported users.In order to keep the simplicity, we will demonstrate theway SU-MIMO and MU-MIMO increase the number ofsimultaneously supported users, using 2x21.4MHz.Analog to that, this will apply to systems with moreantennas and wider bandwidths.Table I. Effective resources available for user datatransmission per radio frameChannel band1.4width [MHz]35101520# RB per slot15255075100Total # of RB120per frame300500100015002000Total # of RE10080per Frame25200 4200084000 126000 168000NumberofRE for Data in 7716DL20136 3393668436 102936 137436Average#of RE per RB 64.3for Data in DL67.12NumberofRE for Data 6600in UL18480 3168064680 97680130680Average#of RE per RB 55for Data in UL61.664.6865.34667.872 68.436 68.62463.3665.1268.718Fig. 4. Resource mapping for video connections, 1.4MHzchannel bandwidth, SU-MIMOFig. 3. shows a case where no MIMO is used, whichmeans that the transmitter uses one antenna and oneantenna is used by the receiver. Using(5), we can calculatethat 8 simultaneous 27ms video streams (from 8 users) canbe supported by 2.7 radio frames. Each user’s informationis colored differently.Fig. 4. shows the case when we add one antenna to thetransmitter and one to the receiver i.e. 2x2 SU-MIMO.Every user’s information is divided into two parts and it istransmitted through two parallel antennas. The receiverhas two antennas as well, so it is able to receive streamsfrom the both antennas at a same time.If we consider using MIMO and CA, the number ofresources available for data changes, therefore the numberof simultaneous users increases.Copyright 2016 IJECCE, All right reserved277

International Journal of Electronics Communication and Computer EngineeringVolume 7, Issue 5, ISSN (Online): 2249–071XFig. 5. Resource mapping for video connections, 1.4MHzchannel bandwidth, MU-MIMOAnyway, as previously stated, video and VoIP calls arereal-time services, so the resources that remain availablecannot be used for increasing the data rate, but forsupporting more simultaneous users. In conclusion, SUMIMO multiplies the number of supported users by thenumber of antennas used, or if there are enough availableresources there can be supported a few more users.Fig. 5 shows the case of 2x2 MU-MIMO. Since thereceiver has only one antenna, the data is beingtransmitted through two antennas to two different users(receivers). In conclusion, use of MU-MIMO multipliesthe number of users exactly by the number of antennasused.The connections we are considering are VoIP and videocalls, therefore the communication is symmetrical.Just for illustration and visual comparison, number ofsupported interactive VoIP and video session in LTE-A indownlink direction, presented in Table II are depicted inFig. 6. Even it is intuitively expected that increase ofcarrier aggregation an MIMO order will doubles thesystem capacity, due to additional resources occupied byreference signals MIMO shows slightly smallerimprovement then CA.In Fig. 4 and Fig. 5 it is presented that not all of theavailable resources are used because the number ofavailable resource block is not enough to meet the requitedone for packet transmission of one video connection. Thisled us to the idea to use these available resources for voicepackages, since the number of resources blocks requiredfor a voice connection is lower.IV. SIMULTANEOUS VOIP AND VIDEOTRANSMISSION OVER LTE-AIn order to maximize the utilization of the resources,considering that the VoIP packets are smaller than videopackets, we can easily combine them in the video resourcescheme using appropriate scheduling. In this paper we willcombine both services on the basis of the video radioframe. Fig. 7 shows the proposed scheduling scheme forcombining video and VoIP calls. In order to keep thesimplicity we use 1.4MHz bandwidth in this case, there isno use of MIMO, nor CA.Table II. Maximum number of supported users for 20MHzchannel bandwidthChannel bandwidth20MHzNo MIMO, no CANo MIMO, CA 2CC2x2 MIMO, no CA2x2 MIMO, CA 2CC4x4 MIMO, no CA4x4 MIMO, CA 2845825801164Table II presents the maximum number of VoIP orvideo connections for 20MHz channel bandwidth, fordifferent scenarios. As shown in Table II, the number ofconnections in downlink and uplink is different fordifferent scenarios, so we will use the lower number ofconnections as a constriction.Number of suported VoIP and Video sessions6000VoIP DL5000Fig. 7. Proposed scheduling scheme for combining videoand VoIP callsAn equation which will apply to all the possible casesusing MIMO and CA can be derived. We propose thefollowing equation for calculating the number of VoIPusers that can be supported parallel to arbitrary number ofvideo users:Video DL4000VoIP ULN voiceVideo UL3000totalvideo ( N RB N RB) P voice packet N RB 2000(6)wheretotalslotN RB Tvideo * 2 * N RB* N antenna10000is the total number of RBs ( Nper time slot andFig. 6. Number of supported interactive VoIP and Videosessions in LTE-Advanced downlinkN antennaslotRB(7)is the number of RBsis the number of antennasused);videovideo packetN RB N video * N RBCopyright 2016 IJECCE, All right reserved278(8)

International Journal of Electronics Communication and Computer EngineeringVolume 7, Issue 5, ISSN (Online): 2249–071Xis the number of RBs used for all the video packets( Nvideo is the number of simultaneous video calls whichcan vary from 0 to the maximum number of video usersvideo packetis the number of RBs for video packetand N RBtransmission) andP Tvideo Tvoice (9)is the period for transmitting voice packets over thevideo resource grid.Using the values from Table III in (6), we generatedFig.8 which shows the number of combined video andVoIP connections, for 20MHz basic channel bandwidth.There are six different scenarios presented. As expected,the number of supported users increases adding moreantennas or component carriers.increased, their utilization is maximized, and inconsequence the number of supported users or data rate isincreased.In this paper, we estimated the number of simultaneousinteractive VoIP and video sessions that can be supportedby one LTE-Advanced cell. We overviewed the video andvoice data mapping onto LTE-Advanced resource grid.Taking into account VoIP and video calls, LTEAdvanced increases the number of supported users. It canbe concluded that CA multiplies the number of supportedusers nearly by the number of component carriers used andMIMO multiplies the number of supported users by thenumber of antennas used.In order to maximize the utilization of the resources, weproposed a method for calculating the number of parallelvideo and VoIP simultaneous connections and comparedthe results for different LTE-A scenarios.Table III. Values for calculating NVoIP depending on Nvideofor 20MHz channel bandwidthChannel20MHzbandwidthREFERENCES[1][2]No MIMO, no CA1006371451No MIMO, CA 2CC20063729112x2 MIMO, no CA10063829022x2 MIMO, CA 2CC20063856824x4 MIMO, no CA10084054044x4 MIMO, CA 2CC20084010804[3][4][5][6][7][8]Fig. 8. Number of supported video and VoIP users fordifferent LTA-Advanced configurations[9]We can notice that the case of “no MIMO, CA 2CC”and the one of “2x2 MU MIMO, no CA” almost overlap.The case of “2x2 MU MIMO, CA 2CC” and “4x4 MIMO,no CA” are comparable, but in the case of “2x2 MUMIMO, CA 2CC” more users can be supported. It dependson the network designer which scenario they will use.[10]C. Cox, An introduction to LTE LTE, LTE-Advanced,SAE and 4G Mobile Communications, Wiley, 20123GPP TS 36.913, Requirements for further advancementfor Evolved Universal Terrestrial Radio Access (EUTRA) – (LTE-Advanced), v9.0.0, December 2009.A. Ghosh, R. Ratasuk, Essentials of LTE – LTE-A,Cambridge University Press, 2011.Joao de Quantaniliha Melerio de Araujo Martines, ( 2016,August 10), Impact of MIMO and Carrier Aggregation inLTE-Advanced, thesis to obtain the Master of ScienceDegree in Electronical and Computer s.pdf.M. Iwamura, “Carrier Aggregation Framework in 3GPPLTE-Advanced,” in IEEE Communication magazine,August 2010, pp. 60-67.A. Ghosh, R. Ratasuk, B. Mondal, N. Mangalvedhe andT. Thomas “LTE-Advanced: Next-Generation cations, June 2010, pp. 10-22,F. Rezaei, M. Hempel, H. Sharif, “LTE ,” 2011 IEEE 16th International Workshop onCAMAD, 2011, pp. 102-106.A. Z. Yonis, M. F. L. Abdullah, (2012, June), “PeakThroughput of LTE – Release 10 for Up/Down LinkPhysical Layer,” International journal of information andnetwork security, Vol. 1, No. 2, pp.88-96, IJINS/article/view/452/227.Cisco Systems, Inc., Document ID: 7934, (2016,February), “Voice Over IP Per Call quality/7934-bwidth-consume.html.Alex MacAulay, Boris Felts, Yuval Fisher, “IP Streamingof MPEG-4:Native RTP vs MPEG-2 Transport ivio.com.V. CONCLUSIONCapacity of LTE-Advanced is significantly increasesdue to usage of CA and spatial multiplexing techniques onphysical layer. In such a way available resources areCopyright 2016 IJECCE, All right reserved279

International Journal of Electronics Communication and Computer EngineeringVolume 7, Issue 5, ISSN (Online): 2249–071XAUTHORS' PROFILESMarko Porjazoski received B.Sc., M.Sc. and Ph.D.degrees in electrical engineering from Ss. Cyril andMethodius University, Skopje, Republic of Macedoniain 2000, 2006 and 2012 respectively. His researchinterests include radio resource management in wirelessnetworks, heterogeneous wireless networks and internettechnologies.Currently he works as Assistant Professor at the Faculty of electricalengineering and information technologies, Ss. Cyril and MethodiusUniversity in Skopje, Republic of Macedonia.Dr. Porjazoski is a member of IEEE and IEEE communication societysince 2005 and member of Macedonian society of engineers inElectronic, Telecommunications, Automation and Informatics (ETAI).Borislav Popovski received B.Sc. degree in electricalengineering from Ss. Cyril and Methodius University inSkopje, R. Macedonia in 1987, M.Sc. and Ph.D. degreesin electrical engineering from University of Zagreb in1993 and 1996 respectively. His main research interestsare optical networks, management of multimedianetworks and services and advanced wireless communications.He is currently full professor at the Faculty of electrical engineering andinformation technologies, Ss. Cyril and Methodius University in Skopje,R. Macedonia.Copyright 2016 IJECCE, All right reserved280

Aggregation (CA), to provide wider bandwidth and support of heterogeneous networks, to improve the capacity and coverage [3]. In this paper the focus is on improvements that are achieved using MIMO and CA by analyzing simultane