Synchronization and Compensation in TDM Based Single RF MIMO-OFDM System

Transcription

1 Proceedigs of the Iteratioal Coferece o Electroics ad Software Sciece akamatsu Japa 205 Sychroizatio ad Compesatio i DM Based Sigle F MIMO-OFDM System Chagyoug A Heug-Gyoo yu Departmet of Electroics Egieerig Chugbuk atioal Uiversy Cheogju Korea acy89027@aver.com ecomm@cbu.ac.kr ABSAC I this paper a approach of time divisio multiplexig (DM) for the MIMO-OFDM (orthogoal frequecy divisio multiplexig) trasmter ad receiver is studied ad evaluated to reduce the power cosumptio ad cost dow of the multiple atea F chais. Wh this DM techique our proposed MIMO-OFDM commuicatio system ca reduce the umber of F chais from to which sigificatly reduces the overall cost ad size of the MIMO-OFDM system. However sychroizatio problem like the carrier frequecy offset (CFO) ad sample timig offset (SO) is serious. So we solve the sychroizatio problem to compesate the CFO ad SO. Fially simulatio results demostrate that the sigle F MIMO-OFDM commuicatio system almost ca provide the same diversy gai as the multiple F chai MIMO systems. KEYWODS MIMO DM Sigle F frot-ed SO CFO IODUCIO MIMO (multiple-iput multiple output) systems use multiple atea elemets i trasmter ad receiver. It ca icrease the spectral efficiecy dramatically suppress iterferece ad improve the robustess of trasmissio i wireless systems [-3]. Capacy of MIMO system icreases liearly accordig to the umber of atea elemets. I MIMO system high-rate data trasmissio is achieved by trasmtig parallel data streams simultaeously. All data streams are idepedet of each other. Differet data stream is mapped oto each differet subcarrier. A typical model for special multiplexig is the Bell Labs Layered Space ime (BLAS) [4] that ca achieve high throughput. ecetly is well kow that MIMO approach wh orthogoal frequecy divisio multiplexig (OFDM). his combiatio method ca achieve spatial diversy ad space-divisio multiplexig gai [5-8]. However MIMO has a crical drawback. he mai problem of MIMO system is highcomplexy high-cost ad high cosumptio by a umber of F-chai. Each F-chai icludes a digal to aalog covertor (DAC) a upcovertor filters ad HPA i trasmter side. Also i receiver side Each F-chai icludes a low-oise amplifier (LA) filters dowcovertors ad ADC. Because of a umber of F-chais MIMO system has the drawback [9]. I order to overcome the drawback oe solutio is to use a sigle F-chai i MIMO system istead of multiple parallel F-chais. hus by usig a F sectio the system ca have a lower complexy ad cost a simple F desig a compact size ad also reduce the power cosumptio [9]. o realize a sigle F frot-ed path orthogoal trasmissio of multiple F streams over a sigle frot-ed must be recogized [0]. May proposals utilize typical atea arrays combied wh techiques for adaptive selectio of subset of the available ateas thus reducig the umber of required F chais []. ISB: SDIWC 58

2 Proceedigs of the Iteratioal Coferece o Electroics ad Software Sciece akamatsu Japa 205 I this paper we preset a sigle F MIMO- OFDM system which use the time divisio multiplexig (DM) for atea selectio. Wh this DM techique our proposed MIMO-OFDM commuicatio system ca reduce the umber of F chais from to which sigificatly reduces the overall cost ad size of the MIMO-OFDM system. However sychroizatio problem like the carrier frequecy offset (CFO) ad sample timig offset (SO) is serious. So we solve the sychroizatio problem to compesate the CFO ad SO. Fially simulatio results demostrate that the sigle F MIMO-OFDM commuicatio system almost ca provide the same diversy gai as the multiple F chai MIMO systems. 2 SYSEM MODEL We assume the MIMO-OFDM system wh trasm ad receive ateas the high t t rate symbols to be trasmted are first grouped ito blocks of data symbols at the trasmter. d he OFDM symbols at the ( i t )th trasmter is represeted by the x ( ) [ x (0 ) x ( )... x ( )] i i i i d t t t t where i t ad deote the idexes of the trasmter ad the OFDM symbol respectively. he each OFDM symbol passes a iverse discrete Fourier trasform (IDF) block at each trasmter. At the IDF output a cycle prefix (CP) of legth is iserted at the begiig of g each symbol to avoid ISI ad s legth is assumed to be loger tha the chael delay legth. he resultig th time-domai OFDM symbol at the ( i t )th trasmter is represeted by s ( ) [ s (0 ) s ( )... s ( )] is [5] d g s ( m ) x ( l ) e d j2 l( mg)/ d () d l0 where m d he samples si t ( m ) are set through a frequecy-selective fadig chael which ca g be give as a equivalet discrete time liear fie duratio chael impulse respose give by a sequece of chael matrices Cl () for l 0... h where h is the time spa of the MIMO chael ad Cl () is a r matrix whose ( ii r t) th elemet c () l represets ir the chael coefficiet form the ( i t ) th trasm atea to the ( ir ) th receive atea at delay l 2. Assumig the received OFDM symbols are free of ISI after receiver removes the CP from each symbol. he resultig frequecy-domai data sample at the k th subcarrier ad ( ir ) th receiver i the th OFDM symbol is give by [5] y ( k ) h ( k) x ( k ) ( k ) (2) ir ir ir ii 0 where h ( k ) is the chael trasfer fuctio ir at the k th subcarrier from the ( i t ) th receiver ad the i r ( k ) is addive whe Gaussia 2 oise wh variace f. For a compact otatio the received th OFDM symbol ca be[6] y( k ) H( k) x( k ) ( k ) (3) where h00 ( k) h0 ( k) h0 ( ) t k h0 ( k) h ( k) h ( ) t k Hk ( ) h 0 ( k) h ( ) ( ) r k h r r k t y( k ) [ y ( k ) y ( k )... y ( k )] 0 x( k ) [ x ( k ) x ( k )... x ( k )] 0 ( k ) [ ( k ) ( k )... ( k )]. 0 Furthermore the received th OFDM symbol i time domai is give as [6] where h z( m ) C( l) s( m l ) u( m ) (4) l0 s( m ) [ s ( m ) s ( m )... s ( m )] 0 t z( m ) [ z ( m ) z ( m )... z ( m )] ad 0 t u( m ) are the trasmted received ad oise ISB: SDIWC 59

3 Proceedigs of the Iteratioal Coferece o Electroics ad Software Sciece akamatsu Japa 205 vectors respectively for the m th sample of the th OFDM symbol. I this paper we preset the zero forcig (ZF) applied to the chael give i Eq.(2). Cosider the MIMO chael give i (2) where the data sub-streams are mixed by the chael matrix. he ZF equalizers ca be applied to decouple the sub-streams. he ZF ad MMSE matrices are [7-8] W H H H * * ZF ( ) (5) After multiplyig the received sigal vector yi r by W ZF we obtai decoupled substreams wh outputs Ss[8] where S (6) ZF * [( HH) ] deotes the th diagoal elemet. 3 SIGLE F MIMO-OFDM SYSEM BASED O DM I the covetioal MIMO system the receiver uses multiple F frot-eds. Oe solutio to reduce the extra hardware cost ad F circu imperfectio is by usig a sigle F frot-ed i a MIMO-OFDM system where a sigle F path is used istead of multiple parallel F paths. I this paper we desig the sigle frot-ed MIMO-OFDM system usig the DM techique which ca be realized as a sigle F chai by timig multiplexig the ateas sigals. he approach to achieve sigle F frot-ed chai is realizable by usig multiple-throw F swch. Figure shows a geeral block diagram of sigle F MIMO trasmter based o DM. But istead of usig multiple F frot-eds a sigle F chai is used to up-covert the F sigals for trasmissio through the ateas. he samples of the first atea i the basebad processig sectio are obtaied from the multiplexer ad the by usig a low-pass filter (LPF). he samples from differet ateas are orthogoal i each symbol duratio ad pass thorough the ateas ate the differet time itervals. Figure 2 shows a geeral block diagram of sigle F MIMO receiver based o DM. his receiver ca be realized as a sigle F MIMO system by time multiplexig the atea s sigals usig a F swch. he received sigals are de-multiplexed after F processig. A sample received from the first atea after passig the bad pass filter. Basebad Processig MUX imig cotrol Mixer OSC HPA Figure. Sigle F MIMO trasmter based o DM. LA imig imig cotrol cotrol Mixer OSC DMU X Basebad Processig Figure 2. Sigle F MIMO receiver based o DM. able. Complexy compariso betwee covetioal ad proposed system model. FF DACs Up-covertor HPA LA Dow-covertor ADC IFF Covetioal Proposed MIMO scheme MIMO scheme able shows complexy compariso betwee covetioal OFDM-MIMO ad the proposed system model. he proposed system ca reduce the umber of FF DAC Up-covertor ad HPA i trasmter side. Also the proposed system ca reduce the umber of LA Dow- ISB: SDIWC 60

4 Proceedigs of the Iteratioal Coferece o Electroics ad Software Sciece akamatsu Japa 205 covertor ADC ad IFF i receiver side. hat is the proposed system uses a lower power i compariso wh covetioal MIMO-OFDM system. Eve though amout of power coservatio is chaged depedig o systems a system usig the proposed scheme has lowpower cosumptio i compariso wh the covetioal scheme usig multiple F-chai. 4 SYCHOIZAIO OF SIGLE F MIMO-OFDM Oe Because of the facts that eve a very small frequecy offset i a MIMO-OFDM system ca destroy the orthogoaly betwee the sub-carriers ad thus the performace of the system sigificatly degrades. Oe of the most importat tasks is carrier frequecy sychroizatio. A frequecy offset may be itroduced due a differece betwee the trasmter ad receiver clock or due to the Doppler effect. he carrier frequecy makes the iter-carrier iterferece (ICI) because of destroyig the orthogoaly. Sigal xt () is chaged like [2] t i2 x( t ') x( t) e (7) rasmted sigal xt () ad chaged sigal xt ( ') have same sample umber. Whe the sigal is affected due to carrier frequecy offset the received sigal is like[2] t' k' i 2 (8) Y ( k ') x( t ') e t ' 0 Equatio (8) shows the carrier frequecy effect i time domai ad i frequecy domai ca be give as [2] ( k ) i2 i2 i ( ) r ir i t ir 0 k0 Y k H k X k e e Z ( k) i2 i2 i2 p p p i r ir 0 k0 H X e H k X k e e Z (9) If a sigal is affected by carrier frequecy offset based o MIMO-OFDM system the the sigals ca be multiplied as three steps. Oe is IFF processig secod is phase rotatio ad last is FF processig. ormalized offset is give as [2] f d (0) carrier spacig ormalized offset is divided by carrier spacig. I this system we compesate those problems wh block type pilot ad assume that the receiver speed is costat. Carrier frequecy offset is compesated wh block type pilot. he phase rotatio is estimated by doig the iterpolatio betwee pilots because the receiver speed is ot dramatically chaged. Phase rotatio is estimated usig received pilot sigals [2] P( i) Block _ pilot( i ) () he P(i) is average of the pilot where is the idex of the pilot. A data that has the umber of sub-carriers of before IFF is set a sigal that has the umber of samples of after IFF i OFDM system. A sigal has samples i a symbol. If the symbol is delayed ca affect ext symbol so ISI occurs. But the ISI ca be moved by cyclic prefix. Equatio (2) represets phase error ad ISI for the received sigal give as [3] ( k) i2 i2 i ( ) r ir i t ir 0 k0 Y k H k X k e e Z ( ) 2 t k i i2 i2 p p i r ir k 0 0 k 0 H X e H k X k e e Z (2) he sigal is multiplied as three steps. First is IFF processig secod is phase rotatio last is FF processig. If there is ISI secod em will affects the received sigal. But due to CP ISI is ot geerated. Sample timig offset ca be give as [4] P p p k0 i2 Y H X e (3) Phase error of sample timig offset has differet phase degree per sub-carrier because of orthogoaly of OFDM. Whe timig offset occurs as oe sample a sub-carrier has 80 phase degree. I this paper P sychroizatio ISB: SDIWC 6

5 Proceedigs of the Iteratioal Coferece o Electroics ad Software Sciece akamatsu Japa 205 sigals are used to estimate the offset give as [4] G arg mi Sycreceived i Syc pilot i i * 2 (4) Equatio (4) is miimum search algorhm. he sychroizatio sigal is made by P sequece. 5 SIMULAIOS AD DISCUSSIO able 2. OFDM-MIMO System. Figure 4 shows the BE performace of sigle F MIMO-OFDM system wh differet SO values. Here we give 5 case for the system to show the performace wh compesatio ad whout compesatio. 5 differet sample timig offsets are give as sample 2 samples 3 samples 4 samples ad 6 samples. As we ca see i the five differet cases after compesatig the SO amouts the BE curves are almost close to the ideal oe which ca cofirm our SO compesatio method is well doe i this processio. Parameter Value FF size 28 Subcarrier 64 Badwidth 20M Modulatio QPSK Chael Model ayleigh Fadig Chael Sample timig offset /2/3/4/6 samples Carrier frequecy offset 0./0.3/0.5 Figure 4. BE performace of sigle F MIMO-OFDM system wh differet SO values. 6 COCLUSIOS Figure 3. BE performace whout SO ad wh compesatio. Figure 3 shows the BE performace of the sigle F MIMO-OFDM whout samplig time offset. Whe there is o SO ad CFO i the system which meas a complete sychroizatio the blue lie which idicates the whout compesatio ad red lie which idicates wh compesatio is the same lie. I this paper we propose a DM based OFDM-MIMO system for the desig of sigle F MIMO schemes iteded to reduce the cost size ad the power cosumptio of the multiple F frot-eds MIMO systems. By usig the DM techique our proposed orthogoal frequecy divisio multiplexig OFDM-MIMO system reduces the umber of F chais from to which sigificatly reduces the overall cost ad size. After desigig the sigle F MIMO-OFDM system we solve the sychroizatio problem by cosiderig both carrier frequecy offset ad sample timig offset. Simulatio results demostrate the simple MIMO receiver almost ca provide the ISB: SDIWC 62

6 Proceedigs of the Iteratioal Coferece o Electroics ad Software Sciece akamatsu Japa 205 same diversy gai as the diversy gai of the multiple F chai MIMO systems. ACKOWLEDGEME his research was supported by Basic Sciece esearch Program through the atioal esearch Foudatio of Korea (F) fuded by the Miistry of Educatio Sciece ad echology (o.203a2a2a ) ad this research was supported by a grat (4P- B ) from the ailway echology esearch Project fuded by the Miistry of Lad Ifrastructure ad rasport (MOLI) of the Korea govermet ad by the Korea Agecy for Ifrastructure echology Advacemet (KAIA). [] A. Kalis A. Kaatas ad C. Papadias A ovel approach to MIMO trasmissio usig a sigle F frot ed IEEE J. Sel. Areas Commu. vol. 26 o. 6 pp Aug [2] Q. Wag C. Yu H. Gao "Semiglobal sychroizatio of multiple geeric liear agets wh iput saturatio" Iteratioal Joural of obust ad oliear Cotrolvol. 24 Issue 8 pp Dec [3] Y. Shag "Cotiuous-time average cosesus uder dyamically chagig topologies ad multiple time-varyig delays" Applied Mathematics ad Computatio vol. 244 pp Oct [4] E. Peymai. I. Fosse "Leader-follower formatio of marie craft usig costrait forces ad lagrage multipliers" Decisio ad Cotrol (CDC) 202 IEEE 5st Aual Coferece o pp Dec EFEECES [] D. Gesbert M. Shafi D. sha Shiu P. J. Smh ad A. aguib "From theory to practice: a overview of MIMO space-time coded wireless systems" IEEE J. Selected Areas Comm. vol. 2 pp [2] A. Paulraj D. Gore ad. abar Multiple atea systems. Cambridge U.K.: Cambridge Uiversy Press [3] A. F. Molisch Wireless Commuicatios. Wiley [4] G. D. Golde C. J. Foschii ad P. W. Woliasky Detectio algorhm ad iial laboratory results usig V-BLAS space-time commuicatio archecture Electroics Letters vol. 35 pp [5] D. Agrawal V. arokh A. aguib ad. Seshadri Space-time coded OFDM for high data-rate wireless commuicatio over widebad chaels i Proc. IEEE It. Cof. Veh. echol. Ottawa Caada pp May 998. [6] Bog-Ju Kim; Sug Kyu Kim; Sug-Wook Yoo; Heug-Gyoo yu; Seug Hwa Lee; Dog-Ji Shi "Desig ad evaluatio of sigle F OFDMMIMO based o time divisio multiplexig" Commuicatios ad Iformatio echologies (ISCI) 204 4th Iteratioal Symposium o pp Sept [7] A. V. Zelst ad. C. W. Schek Implemetatio of MIMO OFDM based wireless LA system IEEE ras. Sigal Process. vol. 52 o.2 pp Feb [8] G. L. Stüber J.. Barry S. W. Mclaughli Y. (G). Li M. A. Igram ad. G. Pratt Broadbad MIMO-OFDM wireless commuicatios Proc. IEEE vol. 92 o. 2 pp Feb [9] A. F. Molisch M. Z. Wi Y. S. Choi ad J. H. Witers Capacy of MIMO systems wh atea selectio IEEE ras. Wireless Commu. vol. 4 o. 4 pp Jul [0] P. heofilakos ad A. G. Kaatas Capacy performace of adaptive receive atea subarray formatio for MIMO systems EUASIP J. Wireless Commu. etw. vol pp. 2 Oct ISB: SDIWC 63