Available online at ScienceDirect. Procedia Computer Science 93 (2016 )

Transcription

1 Avalable onlne at ScenceDrect Proceda Computer Scence 93 (016 ) th Internatonal Conference On Advances In Computng & Communcatons, ICACC 016, 6-8 September 016, Cochn, Inda Performance Analyss of Adaptve Clppng Technque for Reducton of PAPR n Alamout Coded MIMO-OFDM Systems Sadhana Sngh a, *, Arvnd Kumar b, * a Deprtment of ECE, Natonal Insttute of Technology, Kurukshetra , Inda b Department of ECE, Natonal Insttute of Technology, Kurukshetra , Inda Abstract Lke sngle-nput sngle-output orthogonal frequency dvson multplexng (SISO-OFDM), mult-nput mult-output (MIMO) OFDM systems also suffer from the problem of hgh peak to average power rato (PAPR). To account for ths ssue, ampltude clppng s one of the smplest technques employed to reduce PAPR. However, t leads to addtonal dstorton. In ths paper, adaptve clppng technque s proposed to reduce PAPR of Alamout coded MIMO-OFDM systems. In ths scheme, successve peaks are clpped accordng to an adaptve algorthm. Smulaton results reveal that the proposed clppng technque provdes effectve reducton n PAPR and better bt error rate (BER) performance. BER performance and PAPR reducton capablty of the proposed algorthm n MIMO-OFDM systems are also compared wth SISO-OFDM systems. Power spectral densty (PSD) for conventonally clpped, adaptvely clpped, and normal SISO-OFDM systems s analyzed and plotted. It can be revealed from the PSD plot that the spectral spreadng due to the proposed scheme s less than the conventonal clppng scheme. 016 The Authors. Publshed by by Elsever Elsever B.V. B.V. Ths s an open access artcle under the CC BY-NC-ND lcense ( Peer-revew under responsblty of the Organzng Commttee of ICACC 016. Peer-revew under responsblty of the Organzng Commttee of ICACC 016 Keywords: Clppng;OFDM; peak to average power rato (PAPR);space-tme and space-frequency block code(stbc/sfbc); complementary cumulatve dstrbuton functon (CCDF);power spectral densty (PSD). 1. Introducton Mult-nput mult-output (MIMO) systems n combnaton wth orthogonal frequency dvson multplexng (OFDM) have drawn sgnfcant attenton for the next generaton broadband multmeda applcatons due to ther * Correspondng author. Tel.: , E-mal address: sadhana143s@gmal.com, arvnd_sharma@ntkkr.ac.n The Authors. Publshed by Elsever B.V. Ths s an open access artcle under the CC BY-NC-ND lcense ( Peer-revew under responsblty of the Organzng Commttee of ICACC 016 do: /j.procs

2 610 Sadhana Sngh and Arvnd Kumar / Proceda Computer Scence 93 ( 016 ) potental of provdng hgh data rate, robustness to fadng channels and relable communcaton. MIMO-OFDM has been adopted n varous wreless communcaton standards, such as IEEE 80.11a/g, wreless metropoltan area networks (WMANs), worldwde nteroperablty for mcrowave access (WMAX), and 3GPP Long Term Evoluton (LTE) 1. Also t acts as an ar nterface for varous 4G and 5G standards. However, MIMO-OFDM systems encounters problem of hgh peak to average power rato (PAPR) that results n non-lnear dstorton of the transmtted sgnal at hgh power amplfer. Numerous methods have been proposed for reducton of PAPR n OFDM systems such as clppng 3,6,7, partal transmt sequences (PTS) 4,6,7, selectve mappng (SLM) 5-7, tone reservaton, etc. A number of research papers have also been ntroduced on PAPR reducton n MIMO-OFDM systems 8-1. Among all these PAPR reducton methods, clppng seems to be the smplest scheme. However, t suffers from n-band and out-of-band dstorton that degrades the performance of system. To cope wth ths problem, an adaptve clppng technque 13 s suggested for SISO- OFDM systems. In ths paper, the adaptve clppng algorthm s extended to space tme block coded (STBC) MIMO-OFDM systems to reduce the PAPR. To examne the reducton n PAPR obtaned by the present scheme, complementary cumulatve dstrbuton functon (CCDF) s evaluated. The reducton n the value of CCDF ndcates the reducton n the PAPR. The BER performance and CCDF for MIMO-OFDM as well as sngle-nput sngle-output (SISO) OFDM systems for the conventonal clppng and proposed method are analyzed and compared. Smulaton results reveal that the adaptve clppng technque outperforms the conventonal clppng technque on the bass of BER mprovement and sgnfcant PAPR reducton. To vew the effect of clppng on the power of sde lobes, the power spectral densty (PSD) of standard OFDM system, OFDM wth conventonal clppng and OFDM wth adaptve clppng s plotted. It s clear from the smulaton results, that the spectral spreadng s less n case of proposed clppng scheme as compared to the conventonal clppng scheme. The remander of ths paper s organzed as follows. Bref descrpton of Alamout MIMO-OFDM systems, channel model and PAPR defnton s ncluded n Secton. In Secton 3, Adaptve clppng technque s appled to MIMO-OFDM. Smulaton results are nvestgated n secton 4. The concludng remarks are made n secton 5.. System Descrpton and PAPR n Alamout MIMO-OFDM Systems.1. Alamout MIMO-OFDM Systems The most wdely used transmt dversty technque was frst ntroduced by Alamout 14. K. F. Lee and D. Wllams proposed STBC-OFDM over frequency selectve fadng channels 15. Tx1 Input Data Bts Modulaton Seral to Parallel Convertor STBC Encoder N-IFFT Add CP N-IFFT Add CP Tx h h 1 Rx Output Data Bts Demodulaton Parallel to Seral Convertor STBC Decoder Remove CP N-FFT Fg. 1. Basc schematc of two branch STBC-OFDM System. In SISO-OFDM, an nput data block of length N X [ X(0) X(1) X() X( N1)] T where N s the number of subcarrers and X( k ) are QAM or PSK modulated symbols, s passed through IFFT block, thus generatng dscrete tme doman SISO-OFDM sgnal x [ x(0) x(1) x()... x( N1)] T where xn ( ) s expressed as

3 Sadhana Sngh and Arvnd Kumar / Proceda Computer Scence 93 ( 016 ) N1 1 j kn xn ( ) X( ke ) N, n0,1,, N1 (1) N k0 To study the PAPR, Alamout coded MIMO-OFDM system s consdered wth two transmt antennas and sngle receve antenna. The basc schematc of two branch STBC-OFDM system s shown n Fg.1. Here the nput data block s STBC encoded nto two data vectors. At the output of seral to parallel convertor, consder consecutve symbol vectors n par at one nstant. Out of the par, say frst vector be odd vector X and say the second one to be the even vector X e. Let say Xo s the J th block and Xe s the (J+1) th block of data symbol vector and they are represented by X [ ( ) ( 1) ( 1)] T o X JN X JN X JN N () X [ ( ) ( 1) ( 1)] T e X JN N X JN N X JN N (3) * The vector X s sent durng frst tme nterval and -X durng second tme nterval by the frst transmt antenna. o e * The second transmt antenna transmts X e n frst tme nterval and X n the second tme nterval. Thus the o transmsson matrx contanng STBC encoded OFDM symbol vectors s represented by xo xe G (4) xe xo where denotes the conjugate operaton, xo and x e are the IDFTs of Xo and X e respectvely. A tapped-delay lne channel model s used for dscrete multpath channel 16. The mpulse response of the tme nvarant channel s gven as o L1 j l h() l e ( l) (5) l0 where l fcl, l s the number of multpath, l represents the channel path gans, f c denotes the carrer frequency and l are the delays. Now h 1 and h are the channel mpulse responses between transmt antenna 1 and. Let say H 1 and H are dagonal matrx wth dagonal elements as DFT of h 1 and h respectvely. Also the channel mpulse response remans constant durng the two tme ntervals. The demodulated vectors durng frst and second sgnallng nterval are gven by Y=HX+HX+N e 1 (6) * * Y=-HX+HX+N 1 e o (7) The channel response may be estmated at the recever or assumng t s known at the recever 15. The data symbols are reconstructed usng maxmal rato combnng as * * o 1 1 X =H Y +H Y (8) * * e 1 1 X =H Y -H Y (9).. PAPR n MIMO-OFDM PAPR s expressed as the rato of nstantaneous peak power to average power of OFDM sgnal 7. In SISO-OFDM, PAPR s expressed as

4 61 Sadhana Sngh and Arvnd Kumar / Proceda Computer Scence 93 ( 016 ) PAPR max xn ( ) 0nN1 (10) SISO OFDM E[ x ] Whle n Alamout MIMO-OFDM, PAPR s evaluated for the transmt sgnals from the two transmt antennas and s represented by PAPR max x 0nN1 (11) E[ x ] where x are the dscrete tme STBC-OFDM sgnal samples and =1, denotes the number of antennas. The PAPR of STBC MIMO-OFDM system s defned as 9 PAPR max PAPR (1) MIMOOFDM 1, The PAPR of MIMO-OFDM and SISO-OFDM system s measured. The CCDF s taken as a metrc for evaluatng PAPR for both the systems. 3. PAPR Reducton Schemes 3.1. Conventonal Clppng Technque In Conventonal clppng, pror to passng the sgnal through power amplfer, envelope of the sgnal s clpped to a predefned threshold called clppng level (CL) 7. To reduce the PAPR n Alamout MIMO-OFDM, the clppng operaton s appled on the OFDM sgnal as x f x CL Cx [ ] (13) j CL e f x CL where x s the dscrete tme STBC-OFDM symbol, =1, represents the transmt antenna and x s the phase angle of x ( n ). The value of CL depends on clppng rato (CR) as CR 0log CL 10 Ex [ ] (14) Although clppng s the smplest PAPR reducton technque, but t ntroduces out-of band and n-band dstortons that degrades the BER performance and results n spectral spreadng respectvely. However, the n-band dstorton can be elmnated by oversamplng and out-of band dstorton can be cured by flterng the clpped sgnal. 3.. Adaptve Clppng Technque In conventonal clppng, peaks n the envelope are clpped based on the CR. For small CR values, the PAPR reducton s apprecable but BER degrades severely whereas for large CR values, PAPR as well as BER both degrades. To mprove the PAPR reducton capablty and BER performance of SISO-OFDM systems, an adaptve clppng algorthm was analyzed 13. The scheme provdes better performance n terms of PAPR and BER as compared to conventonal clppng. The same scheme s proposed here for the PAPR reducton n MIMO-OFDM systems. In ths scheme, the PAPR s reduced by clppng the successve peaks of OFDM sgnal n an adaptve manner. The adaptve clppng algorthm for MIMO-OFDM sgnal x s as follows Evaluatng the peak value: In the frst step, the successve peak values of Alamout coded MIMO-OFDM symbols x from both the antennas. Here =1, represents the number of antenna. Consder w1 be the of maxmum peak value of the MIMO- OFDM sgnal x.e.

5 Sadhana Sngh and Arvnd Kumar / Proceda Computer Scence 93 ( 016 ) w1 max[ abs( x)] max[ x ] 1 (15) After that the next peak values of ( ) OFDM sgnal.e. x n are evaluated. Let w be the next maxmum peak n the STBC MIMO- w secondmax[ abs( x)] max[ x ] (16) Smlarly, let w be the thrd maxmum value n the x ( n ).e. 3 w3 thrdmax[ abs( x)] max[ x ] 3 (17) and so on, such that: max[ x ] max[ x ] max[ x ] max[ x ] (18) 1 3 p where the subscrpt p ndcates the number of peak values consdered n ths algorthm. Thus, n ths way successve peak values are computed. Determnng the locaton of peaks: In the second step, the locaton of the successve peaks of x s determned. Let us denote these locatons as w1_ loc, w_ loc, w3_ loc,, wp _ loc (19) Clppng the peaks successvely: After havng the knowledge of the peaks and ther locatons n the MIMO-OFDM sgnal, clppng operaton s carred out by multplyng the frst peak n the MIMO-OFDM sgnal by a reducton factor (RF) whch s bascally a rato of lowest peak value of x consdered (.e. the p th peak) wp to the respectve peak (.e. frst peak n ths case) w 1. Smlarly, the successve peaks of x ( n ) are clpped by multplyng each successve peak by a RF expressed as RF lowest peak value (n consderaton) of x peak of x correspondng to the peak beng clpped n x After ths, the new MIMO-OFDM sgnal s constructed and PAPR of ths sgnal s computed. To test ths algorthm, the MIMO-OFDM system s smulated. To compute RF, ten peak values of MIMO-OFDM sgnal are evaluated and nne peaks are clpped. The modfed MIMO-OFDM symbol, after applyng modfed clppng becomes w10 x( w1 _ loc) x( w1 _ loc)* w1 w10 x( w _ loc) x( w _ loc)* w (0) (1) w10 x( w9 _ loc) x( w9 _ loc)* w9 In ths scheme each peak s not clpped by a constant amount. The peaks are clpped adaptvely dependng upon

6 614 Sadhana Sngh and Arvnd Kumar / Proceda Computer Scence 93 ( 016 ) the values of the reducton factor. The smulaton results show a sgnfcant reducton n PAPR and mprovement n BER performance as compared to the conventonal clppng technque. 4. Smulaton Results In ths secton, smulatons have been done to analyze the PAPR reducton capablty and mprovement n the BER performance of the adaptve clppng technque n Alamout coded MIMO-OFDM systems. Two transmt antennas (Tx) and one receve antenna (Rx) s consdered n the smulaton. The MIMO-OFDM system s smulated usng MATLAB software verson wth N=18 subcarrers, system bandwdth of 0MHz, and 4-QAM and QPSK modulatons. To judge the performance of the proposed scheme, CCDF and BER for SISO-OFDM and MIMO- OFDM s evaluated. 4.1 Complementary cumulatve dstrbuton functon Snce PAPR s a random varable, emprcal CCDF s a tool used to evaluate the PAPR reducton acheved. CCDF ndcates the probablty of sgnal envelope gong beyond a partcular PAPR threshold and s defned as CCDF[PAPR( x(n))]=pr{papr( x(n)) } () N = 1- (1- e ) The CCDF for SISO-OFDM system and Alamout coded MIMO-OFDM system wth conventonal clppng as well as wth adaptve clppng s plotted for 4-QAM and QPSK modulaton. For OFDM system, N=18 subcarrers are used and the CR of 9dB s consdered when conventonal clppng s appled. For MIMO-OFDM system wth adaptve clppng, nne peak values are taken n ths study. It can be observed from Fg. and Fg. 3, that the PAPR reducton capablty of adaptve clppng s better as compared to conventonal clppng. Also PAPR reducton of around 4 db s achevable usng adaptve clppng technque. Fg.. CCDF Comparson for MIMO-OFDM and SISO-OFDM systems wth 4-QAM, N=18 subcarrers and CR=9dB. Fg.3. CCDF Comparson for MIMO-OFDM and SISO-OFDM wth QPSK, N=18 subcarrers and CR=9dB. 4. BER Performance The BER performance comparson of SISO-OFDM and STBC MIMO-OFDM s compared n Fg.4 and Fg.5 for 4-QAM and QPSK modulatons. The COST07 sx path typcal urban Raylegh fadng channel s used n the smulaton. The perfect channel estmaton s assumed 15. It s clear from the BER performance plots that the BER performance of adaptve clppng scheme s comparable

7 Sadhana Sngh and Arvnd Kumar / Proceda Computer Scence 93 ( 016 ) wth the standard OFDM system. However, the BER performance of conventonal clppng scheme slghtly degrades when CR=9dB. At lower CR values, (e.g. CR=3dB), the conventonal clppng results better PAPR reducton than that of adaptve clppng, but t leads to ncrease n the BER as can be observed from Fg.6. Fg.4. BER Comparson of Alamout MIMO-OFDM and SISO- OFDM wth 4-QAM, N=18 subcarrers and CR=9dB under Raylegh fadng channel. Fg.5. BER Comparson of Alamout MIMO-OFDM and SISO- OFDM wth QPSK, N=18 subcarrers and CR=9dB under Raylegh fadng channel. Fg.6. BER Comparson of Alamout MIMO-OFDM and SISO- OFDM wth 4-QAM, N=18 subcarrers, CR=3dB under Raylegh fadng channel. Fg.7. PSD curve comparson of orgnal OFDM sgnal and clpped OFDM sgnal for 80.11a standard. 4.3 Power Spectral Densty (PSD) PSD curves for SISO-OFDM, conventonally clpped SISO-OFDM and adaptvely clpped SISO-OFDM are plotted n Fg.7 for the analyss of n-band and out-of-band spectrum. To study the PSD, two OFDM symbols have been consdered wth specfcatons of IEEE 80.11a standard.e. N=64 subcarrers wth 5 data subcarrers and 1 null subcarrers, 16-QAM constellaton, cyclc prefx of 16 samples and bandwdth of 0 MHz 17. PSD of these symbols s averaged over 1000 OFDM symbols. For conventonal clppng, CR=3dB s consdered and ten peak values are taken n adaptve clppng. It s clear from the smulaton, that the spectral spreadng s less n case of proposed adaptve clppng scheme as compared to the conventonal clppng scheme.

8 616 Sadhana Sngh and Arvnd Kumar / Proceda Computer Scence 93 ( 016 ) Concluson In ths paper, the proposed adaptve clppng technque for PAPR reducton s analyzed for MIMO-OFDM systems. It s clear from the smulaton results, that adaptve clppng technque shows apprecable reducton n PAPR. Although conventonal clppng provdes better PAPR reducton at lower CR values, but on the other hand t results n BER performance degradaton due to n-band and out-band dstortons. The BER performance of the adaptve clppng technque for MIMO-OFDM s comparable to the orgnal MIMO-OFDM system and t s better than the conventonal clppng at CR=9dB. Smulaton results show that a PAPR reducton of around 4 db can be acheved usng adaptve clppng algorthm. The PSD curves also show that the conventonal clppng leads to nband and out-of-band dstortons whereas the PSD of adaptve clppng s comparable to the OFDM sgnal. References 1. H. Yang. A road to future broadband wreless access: MIMO-OFDM based ar nterface. IEEE Commun Mag 005; 43: J. G. Andrews et al. What wll 5G be? IEEE J Sel Areas Commun 014; 3: J. Armstrong. Peak-to-average power reducton for OFDM by repeated clppng and frequency doman flterng. IEE Electronc Lett 00; 38: J. Hou, J. Ge, J. L. Peak-to-average power rato reducton of OFDM sgnals usng PTS scheme wth low computatonal complexty. IEEE Trans Broadcast 011; 57: R. W. Bauml, R. F. H. Fsher, J. B. Huber. Reducng the peak-to-average power rato of multcarrer modulaton by selected mappng. IEE Electronc Lett 1996; 3: S. Sngh, A. Kumar. On the PAPR Reducton Technques of OFDM Systems. AENSI Australan J. of Basc and Appled Scences (AJBAS)015; 9: Y. Rahmatallah, S. Mohan. Peak-to-average power rato reducton n OFDM systems: A survey and taxonomy. IEEE communcatons surveys & tutorals, vol. 15, no. 4, pp , fourth quarter, U.-K. Kwon, D. Km, G.-H. Im. Ampltude clppng and teratve reconstructon of MIMO-OFDM sgnals wth optmum equalzaton. IEEE Trans Wreless Commun 009; 8: S. H. Wang, C. P. L. A low-complexty PAPR reducton scheme for SFBC MIMO-OFDM systems. IEEE Sgnal Process Lett 009, 16: T. Jang, C. N, L. Guan. A Novel Phase Offset SLM Scheme for PAPR Reducton n A1amout MIMO-OFDM Systems Wthout Sde Informaton. IEEE Sgnal Processng Lett 013; 0: S. J. Ku. Low-Complexty PTS-Based Schemes for PAPR Reducton n SFBC MIMO-OFDM Systems. IEEE Trans Broadcast 014; 60: N.Ta spınar, M.Yıldırım. A Novel Parallel Artfcal Bee Colony Algorthm and Its PAPR Reducton Performance Usng SLM Scheme n OFDM and MIMO-OFDM Systems. IEEE Commun Lett 015; 19: S. Sngh, A. Kumar, A Modfed Clppng Algorthm for Reducton of PAPR n OFDM Systems. Proc. IEEE Internatonal Conference on Computatonal Intellgence and Computng Research (ICCIC) 015; p S. M. Alamout. A smple transmtter dversty scheme for wreless communcatons. IEEE J. Select. Areas Commun 1998; 16: K. F. Lee, D. B. Wllams. A space-tme coded transmtter dversty technque for frequency selectve fadng channels. Proc. IEEE Sensor Array and Multchannel Sgnal Processng Workshop 000; p I.Cyrl-Danel. A MATLAB-based Object-Orented approach to multpath fadng Channel Smulaton- a MATLAB central submsson from T. v. Waterschoot, V. L. Nr, J. Duplcy, M. Moonen. Analytcal Expressons for the Power Spectral Densty of CP-OFDM and ZP-OFDM Sgnal. IEEE Sgnal Processng Lett 010; 17: