Rajbir Kaur 1, Charanjit Kaur 2

Transcription

1 Rajbir Kaur, Charanjit Kaur / International Journal of Engineering Research and Alications (IJERA) ISS: -9 Vol., Issue 5, Setember- October 1, based Channel Estimation Meods for MIMO-OFDM System using QPSK modulation technique Rajbir Kaur 1, Charanjit Kaur 1 (Assistant. Prof., ECE, University College of Engineering, Punjabi University, Patiala, India) (Student, ECE, University College of Engineering, Punjabi University, Patiala, India) ABSRAC Multile-Outut (MIMO) antenna systems rovide high transmission data rate, sectral efficiency and reliability for wireless communication systems. hese arameters can be furer imroved by combating fading, e effect of which can be reduced by roerly estimating e channel at e receiver side. In is aer a new aroach based on time-domain interolation (DI) has been resented. DI is obtained by assing estimated channel to time domain rough Inverse Discrete Fourier ransform (I), zero adding and going back to frequency domain rough Discrete Fourier ransform (). he analysis of e channel estimators has been conducted using a MALAB. he comarison has been carried out between ower of true channel and estimated ower for e given channel using LS, LS- Sline and MMSE for QPSK modulation at SR 3dB. It is investigated at by alying e over e estimated ower of channel, e erformance of e channel estimators becomes better. Keywords: Channel estimation, Discrete Fourier transform, Least square error, Minimum mean square error, MIMO- OFDM, QPSK. I. IRODUCIO he design of a mobile communication channel is very challenging in ese days due to severe multia roagation, arising from multile scattering by buildings and oer structures in e vicinity of a mobile unit [1]. A wideband radio channel is normally frequency selective and time variant. Similar number of coies of a single transmitted signal reaches at e receiver at slightly different times. Various diversity techniques hel to determine e transmitted signal at highly attenuated receiver side. Multile inut multile outut (MIMO) antenna systems are a form of satial diversity. Deloyment of multile antennas bo at transmitter and receiver side, achieves high data rate wiout increasing e total transmission ower or bandwid in e multia rich environment. he major advantage of MIMO system is a significant increase of bo e system s caacity and sectral efficiency. he caacity of a wireless link increases linearly wi e minimum of e number of transmitter or e receiver antennas []. he caacity of communication system increases linearly wi e number of antennas, when erfect knowledge about e channel is available at e receiver. Generally, MIMO detection schemes require erfect channel knowledge but it is never known before. In ractice, e channel estimation rocedure is done by transmitting ilot (training) symbols at are known at e receiver. he quality of e channel estimation affects e system erformance and it deends on e number of ilot symbols being transmitted. raining based channel estimation, blind channel estimation and semi blind channel estimation techniques can be used to obtain e channel state information (CSI). he CSI and some roerties of e transmitted signals are used to carry out e blind channel estimation [3]. As in Blind Channel Estimation no ilot symbols are transmitted so it has advantage of no overhead loss; it is only alicable to slowly time-varying channels due to its need for a long data record. raining symbols or ilot tones at are known a riori to e receiver are multilexed along wi e data stream for training based channel estimation algorims []. Semi-blind channel technique is hybrid of blind and training technique, utilizing ilots and oer natural constraints to erform channel estimation. In is aer channel imulse resonse has been estimated and comared using LS, MMSE and based estimation techniques. he aer is organized as follows. In Section, MIMO system and channel estimation is described. Section 3 discusses training (ilot) based channel estimation. Simulation and results for e erformance of LS, MMSE and based techniques are given in section and Section 5 concludes e aer.. MIMO-OFDM SYSEM AD CAEL ESIMAIO In MIMO system signals are samled in e satial domain at bo ends and combined in such a way at ey eier create effective multile arallel satial data ies, and/or add diversity to imrove e quality of e communication [5]. MIMO-OFDM technology has been researched as e infrastructure for next generation wireless networks. OFDM simlifies e imlementation of MIMO wiout loss of caacity, reduces receiver comlexity, avoids ISI by modulating narrow orogonal carriers and each narrowband carrier is treated as a searate MIMO system wi zero delay-sread in MIMO-OFDM systems. Basically, e MIMO-OFDM transmitter has arallel transmission as which are very similar to e single antenna OFDM system, each branch erforming serial-to-arallel conversion, and ilot insertion, -oint IFF and cyclic extension before e 139 P a g e

2 Rajbir Kaur, Charanjit Kaur / International Journal of Engineering Research and Alications (IJERA) ISS: -9 Vol., Issue 5, Setember- October 1, final X signals are u-converted to RF and transmitted. It is wor noting at e channel encoder and e digital modulation, in some satial multilexing systems, can also be done er branch, where e modulated signals are en sace-time coded using e Alamouti algorim [] before transmitting from multile antennas [7] not necessarily imlemented jointly over all e branches. Subsequently at e receiver, e is removed and -oint FF is erformed er receiver branch. ext, e transmitted * * * A1 A A 1 A A 3... A 1 * * * A A 1 A A 3 A... A he vectors A 1 and A are modulated using e IFF and after adding a as a guard time interval, and are en transmitted by e first and second transmit antennas resectively. a ( n) I{ A ( k)} () A 1 (k) I a 1 (n) b 1 (n) B 1 (k) M-Array Sace ime Coding A (k) I a (n) b (n) B (k) Sace ime Decoding Demaer Maing A (k) I a (n) b R (n) B R (k) Fig.1 MIMO-OFDM System Channel Estimation symbol er X antenna is combined and oututted for e subsequent oerations like digital demodulation and decoding. Finally all e inut binary data are recovered wi certain BER. As a MIMO signaling technique, different signals are transmitted simultaneously over R transmission as and each of ose R received signals is a combination of all e transmitted signals and e distorting noise. It brings in e diversity gain for enhanced system caacity as we desire. Meanwhile comared to e SISO system, it comlicates e system design regarding to channel estimation and symbol detection due to e hugely increased number of channel coefficients. he data stream from each antenna undergoes OFDM Modulation. he Alamouti Sace ime Block Coding (SBC) scheme has full transmit diversity gain and low comlexity decoder, wi e encoding matrix reresented as referred in [] for two transmitting and two received antenna wi number of subcarrier A A A A * 1 * A1 (1) Assuming at guard time interval is more an e exected largest delay sread of a multia channel. he received signal will be e convolution of e channel and e transmitted signal. Assuming at e channel is static during an OFDM block, at e receiver side after removing e, e FF outut as e demodulated received signal can be exressed as B1 1,1 1, 1, A1 Z1 B,1,, A Z (3) B A R g,1 g, Z R, In e above equation Z1, Z, Z denotes itive White Gaussian oise (AWG). he n column of is often referred to as e satial signature of e n transmit antenna across e receive antenna array. he urose of channel estimation is to estimate channel arameters from e received signal. he function at mas e received signal and rior knowledge about e channel and ilot symbols is called e estimator. he effect of e hysical channel on e inut sequence can be characterized using channel estimation rocess. he channel estimate is simly e estimate of e imulse resonse of e system if e channel is assumed to be linear. A good channel estimate is one where some sort of error minimization criteria is satisfied. If e(n) 1 P a g e

3 Rajbir Kaur, Charanjit Kaur / International Journal of Engineering Research and Alications (IJERA) ISS: -9 Vol., Issue 5, Setember- October 1, denotes estimation error (difference between actual Where B e vector of outut signal is after OFDM received signal and estimated signal), channel estimation demodulation as B B, B1,, B 1, algorims are used to minimize e mean squared error E[ e ( n )] while utilizing as little comutational (MSE), resources as ossible in e estimation rocess. 3. RAIIG BASED CAEL ESIMAIO USIG LS AD MMSE ESIMAOR In is work we have considered Block ye and Comb ye ilot arrangements. he ilots are transmitted on all subcarriers in eriodic intervals of OFDM blocks for a slow fading channel, where e channel is constant over a few OFDM symbols and is tye of ilot arrangement is called e block tye arrangement. he ilots are transmitted at all times but wi an even sacing on e subcarriers, called comb tye ilot arrangement for a fast fading channel, where e channel changes between adjacent OFDM symbols. Wi interolation techniques e estimation of channel at data subcarrier can be obtained using channel estimation at ilot subcarriers. For comb-tye ilot based channel estimation, e ilot signals are uniformly inserted into A(k) according to e following equation [9] A k A lm m m,1,..., M 1 where M = o. of subcarriers ()/ o. of ilot ( ) l = ilot carrier index. Frequency resonse of e channel at ilot sub-carriers defines as k k,1,.... he estimate of e channel at ilot sub-carriers based on LS estimation is given by: B k k k,1,..., 1 (5) A k B k and A k m inf. Data m 1,,..., M 1 () A k are outut and inut at e k ilot sub-carrier resectively. LSE and MMSE algorims are used for estimation of channel at ilot frequencies for bo block tye and comb tye ilot arrangement. An interolation technique is necessary in order to estimate e channel imulse resonse at data frequencies using channel information at ilot subcarriers. 3.1 LSE Let A is e diagonal matrix of ilots as A diag A, A1,, A 1, is e number of ilots in one OFDM symbol, ĥ is e imulse resonse of e ilots of one OFDM symbol, and Z is e AWG channel noise. If ere is no ISI, e signal received is written as [1] B AFhˆ Z is transose, F is e Fourier transfer matrix. he urose of LS algorim is to minimize e cost function K wiout noise. K B AFhˆ () Let Ĥ is e estimate imulse resonse of e channel ˆ 1 LS A B (7) or ˆ B B1 B B 1 LS A A1 A A 1 Because of no consideration of noise and ICI, LS algorim is simle, but obviously it suffers from a high MSE. 3. MIIMUM MEA SQUARE ERROR If e channel and AWG are not correlated, MMSE estimate of is given by [11] ˆ 1 S S B () Where SB EB SBB EBB MMSE B BB S A AS A I Z are e cross covariance matrix between and B, and auto-covariance matrix of B resectively. S is autocovariance matrix of. Z is e noise-variance. If S and Z are known to e receiver, CIR could be calculated by MMSE estimator as below ˆ 1 S S B MMSE B BB S A AS A I Aˆ 1 Z LS 1 1 Z LS S S A A ˆ (9) E b At lower value of e erformance of MMSE estimator is much better an LS estimator. MMSE estimator could gain 1-15 db more of erformance an LS.. BASED CAEL ESIMAIO Alication of on LS, MMSE channel estimation can imrove e erformance of estimators by eliminating e effect of noise. In OFDM system, e leng of e channel imulse resonse is usually less an e leng of e cyclic refix L. -based algorim uses is feature to increase e erformance of e LS and MMSE algorims. It transforms e frequency channel estimation into time channel estimation using I, considers e art which is larger an L as noise, and en treats at art as zero in order to eliminate e imact of e noise. 11 P a g e

4 Power[dB] Power[dB] Power[dB] Rajbir Kaur, Charanjit Kaur / International Journal of Engineering Research and Alications (IJERA) ISS: -9 Vol., Issue 5, Setember- October 1, Let Ĥk denote e estimate of channel gain at e k subcarrier, obtained by eier LS or MMSE channel estimation meod. aking e I of e channel estimate k ˆ [ ] 1, k ˆ I ˆ [ k ] h [ n ] z [ n ] h [ n ], n,1,..., 1 where z[n] denotes e noise comonent in e time domain. Eliminate e imact of noise in time domain, and us achieve higher estimation accuracy. ˆ h[ n] z[ n], n,1,,..., L 1 h [ n] (1), oerwise It is observed at simulation results become better if e estimated outut from various estimators is subject to. Simulation results show at MMSE wi erforms better an oer estimations at e cost of comutational comlexity. 1 LS wi and wiout aking e remaining L elements to transform in frequency domain [1-1] ˆ [ k] hˆ ( n) (11) Simulations are carried out for channel estimation using LS-Linear, LS-sline, MMSE meods. he Simulation results show at e erformance of based channel estimator is much better over e LS, MMSE estimator. Fig., 3 and reresents e erformance of above mentioned channel estimator wi and wiout. OFDM system arameters used in e simulation are indicated in e ABLE 1. - rue Channel LS wiout LS wi Fig. Performance of MIMO-OFDM System Using LS- Linear Channel Estimation wi and wiout for QPSK at SR 3 db 1 LS -SPLIE wi and wiout able 1 Simulation Parameters Parameters Secifications FF size 3 SR 3 db Guard interval OFDM symbol leng 3 Symbol duration 1 sacing umber of ilot - rue Channel LS-SPLIE wiout LS-SPLIE wi Fig. 3 Performance of MIMO-OFDM System Using LS- Sline Channel Estimation wi and wiout for QPSK at SR 3 db 1 MMSE wi and wiout Data er OFDM(modulated) symbol - umber of bits er symbol Signal Constellation QPSK rue Channel MMSE wiout MMSE wi Fig. Performance of MIMO-OFDM System Using MMSE Channel Estimation wi and wiout for QPSK at SR 3 db. 1 P a g e

5 Rajbir Kaur, Charanjit Kaur / International Journal of Engineering Research and Alications (IJERA) ISS: -9 Vol., Issue 5, Setember- October 1, COCLUSIO he combination of OFDM wi Multile Inut and Multile Outut has fulfilled e future needs of high transmission rate and reliability. he quality of transmission can be furer imroved by reducing e effect of fading, which can be reduced by roerly estimating e channel at e receiver side. For high SRs e LSE estimator is bo simle and adequate. he MMSE estimator has good erformance but high comlexity. o furer imrove e erformance of LSE and MMSE, based channel estimation is alied. For subcarrier index 1, true channel ower comes out to be 7.9dB. Estimated ower is calculated using LS linear, LS sline and MMSE as.79db, 7.5dB and 7.5 db and erformance is imroved by.57 db,.3 db and..1 db resectively wi alication of technique. REFERECES [1] Andrea Goldsmi, Wireless Communication, Cambridge University Press, 5. [] aris Gacanin and Fumiyuki Adachi, On Channel Estimation for OFDM/DM Using MMSE-FDE in Fast Fading Channel, EURASIP Journal on Wireless Communications and etworking, vol. 9, June 9,. 11. [3] A. Petroulu, R. Zhang, and R. Lin, Blind OFDM channel estimation rough simle linear re-coding, IEEE ransactions on Wireless Communications, vol. 3, no., March, [] Osvaldo Simeone, Yeheskel Bar-ess, Umberto Sagnolini, -Based Channel Estimation for OFDM Systems by racking e Delay-Subsace, IEEE ransactions on Wireless Communications,vol.3, no.1, January, [5]. Jiang and P. A. Wilford, A hierarchical modulation for ugrading digital broadcasting systems, IEEE ransaction on Broadcasting, vol. 51, June 5, -9. [] Siavash M. Alamouti, A Simle ransmit diversity echnique for Wireless Communications, IEEE Journal on Select Areas in Communications, Vol. 1, o., October 199. [7] PA Pei-sheng, ZEG Bao-yu, Channel estimation in sace and frequency domain for MIMO-OFDM systems, ELSEVIER journal of China Universities of Posts and elecommunications, Vol. 1, o. 3, June 9, Pages -. [] Mohammad orabi, Antenna selection for MIMO-OFDM Systems ELSEVIER journal on Signal Processing, Vol.,, Pages [9] Meng-an sieh Channel Estimation for OFDM Systems based on Comb-ye arrangement in frequency selective fading channels, IEEE ransaction on Wireless Communication,vol., no. 1, May 9, [1] Meng-an sieh, Channel Estimation For OFDM Systems Based On Comb-ye Arrangement In Frequency Selective Fading Channels, IEEE ransactions on CE, Feb 199, Vol., o. 1, [11] M. R. McKay and I. B. Collings, Caacity and erformance of MIMO-BICM wi zero-forcing receivers, IEEE rans. Commun., vol. 53, no. 1,. 7 3, Jan. 5. [1] Minn,. and Bhargava, V.K. (1999) An investigation into time-domain aroach for OFDM channel estimation, IEEE rans. on Broadcasting, 5(), 9. [13] Van de Beek, J.J., Edfors, O., Sandell, M. et al. () Analysis of -based channel estimators for OFDM, Personal Wireless Commun., 1(1), [1] Fernandez-Getino Garcia, M.J., Paez-Borrallo, J.M., and Zazo, S. (May 1) -based channel estimation in D-ilot-symbol-aided OFDM wireless systems, IEEE VC 1, vol., P a g e