, Impact Facto :.643 eissn : 349-000 & pissn : 394-4544 Intenational Jounal of Reseach and Applications (Ap-Jun 015 Tansactions) (6): 309-313 Intenational Confeence on Emeging Tends in Electonics & Telecommunications (ICETET-15) E L E C T R O N I C S R E S E A R C A R T I C L E VLSI Implementation of Low Complexity MIMO Detection Algoithms 1 Vakulabhaanam Ramakishna and Tippati Anil Kuma, SMIEEE 1 Depatment of Electonics & Communication Engineeing, 1 SR Engineeing College (Autonomous), Waangal, Telangana., India 1 amakishna_bhaan@yahoo.co.in, tvakuma000@yahoo.co.in ABSTRACT This pape pesents VLSI implementation of low complexity multiple input multiple output (MIMO) detection algoithms which ae Maximum Likelihood and Zeo Focing. Initially, the MIMO system stuctue, the mathematical model and two eceive detection algoithms ae pesented. Then we investigated the chaacteistics and pefomance of typical algoithms and concentated on VLSI implementation of the system achitectue and the two detection algoithms, which can give good pefomance. Finally, simulation esults have been compaed with each othe in tems of complexity and eo pefomance. Keywods: VLSI; MIMO, Detection Algoithms, ZF, ML, FPGA, BER, Low Complexity. 1. INTRODUCTION Multiple Input Multiple Output (MIMO) is used to descibe the multi-antenna wieless communication system, an abstact mathematical model can take advantage of a pluality of tansmitte antennas ae each independently a tansmission signal, and a pluality of antennas at the eceive eceive and ecove the oiginal message. In MIMO communication system without inceasing the tansmission bandwidth channel capacity inceases exponentially [1]. MIMO technology is the futue wieless communication systems to achieve high data ate tansmission, to impove tansmission quality, an impotant way to impove the system capacity. This pape is aanged as follows: Section II gives bief oveview of system model. Section III pesents the detection algoithms like Zeo Focing and Maximum likelihood detection algoithms. Section IV shows the esults of Bit Eo Rates and also hadwae implementation esults and section V is conclusion of the pape. This pape investigates the applicability of FPGA system fo low complexity MIMO Detection algoithms in effective and economical way.. SYSTEM MODEL ee a MIMO with Nt tansmit antennas and N eceive antennas is consideed. A block diagam of system is shown in figue 1. Zeo Focing (ZF) and Maximum Likelihood (ML) detection algoithms ae used to sepaate the spatially multiplexed data units at the eceive []. Implementing the design using FPGA is vey fast with lowe development costs and takes less amount of time. VLSI implementation of the detection algoithms in the MIMO system will tun out to be a key methodology in the futue wieless communication system. T R A N Figue1. Basic MIMO System Model R E C IJRA 015 Volume Issue 6 P a g e 309
Intenational Jounal of Reseach and Applications (Ap-Jun 015 Tansactions) Intenational Confeence on Emeging Tends in Electonics & Telecommunications (ICETET-15) Received signal Z X n (1) Whee =1,, 3... R, R is the numbe of sub caies and eceived signal vecto T Z Z, Z,... Z ], [ 1 Tansmit Signal vecto T X X, X,... X ], [ 1 Nt T [ n, n,.. n is the additive noise vecto and 1 Nt n ] epesents white guassian noise of vaiance σ. indicates the NxNt channel matix. In ode to identify the communicated data it would be best to use ML detecto [3]. A couple of significant poblems exist with planning channel estimatos. Fist one is the pepaation of pilot infomation, which is used as the efeence signal handled by both tansmittes and eceives. The second touble is the stategy of an estimato by means of both low complexity as well as geat channel tacking potential and also both of these issues ae inteconnected. MIMO Wieless communication systems equie the low complexity and high pecision detectos to achieve high data ates with low BER. The channel estimations ae geneally adopted in MIMO systems to achieve the tade-off between complexity and accuacy. In the MIMO channel model, the numbe of pilot signals may be When the pilot signalp1p= [1 0], p p ( p 1 0) () 1 11 1 1 n1 1 S1 S n ( p 1 0) (3) This does not conside the effect of noise. Thee ae 11 1 / p1, 1 1 / p (4) / p, p (5) 1 1 / When the pilot signalp1p= [0 1], Theefoe, using the guide stuctue, the eceiving end can estimate the channel is: 11 1 (6) 1 Pilot-based methods [4] ae extensively used to appoximate the channel popeties and extact the eceived signal. The pilot signal allocated to a specific block, which is diected peiodically in time-domain, this type of pilot pocedue is paticulaly appopiate fo slow-fading adio N channels. Since the taining block compises all pilots, channel intepolation in fequency domain is not equied. So, this kind of pilot scheme is easonably unesponsive to fequency selectivity. Comb-type pilot system is povides bette esistance to fast-fading channels. Since only some sub-caies contain the pilot signal, the channel esponse of non-pilot sub-caies will be estimated by intepolating neighboing pilot sub-channels. Thus the comb-type aangement is sensitive to fequency selectivity when compaing to the blocktype pilot aangement system. In block-type pilot based channel estimation, OFDM channel estimation symbols ae tansmitted peiodically, in which all sub-caies ae used as pilots [5]. If the channel is constant duing the block, thee will be no channel estimation eo since the pilots ae sent at all caies. The estimation can be pefomed by using Zeo Foce, ML, LS, and MMSE. Detection Algoithms Maximum Likelihood Detection (MLD) This is the theoetical optimum detection method can eceive divesity gain full access. By the estimation theoy, fo Y = X+Z, estimation, maximum likelihood estimation method to constuct a cost function p ( Y, X) so that the cost function to obtain the maximum value of is the final estimate ˆ ag max{ p( / Y, X} (7) This pape is focused on the 4 4 MIMO system simulation, constellation map of QPSK, at the eceiving end using the maximum likelihood detection algoithm is to investigate channel estimation algoithm pefomance, so when testing using the tansmitted signal space uses exhausted seach appoach to maximum likelihood detection [6], in ode to fully demonstate channel estimation algoithm BER pefomance. Zeo Focing Detection (ZF) Zeo focing detection algoithm is a channel to the eception signal by the invese matix of the intefeence of othe uses it can be eliminated, but the noise is also multiplied by the invese of the channel matix, in geneal, the channel coefficient matix is less than 1, then it the invese is geate than 1, ie to the noise by a facto geate than 1, the noise must be amplified. The zeo focing detection multiplies the eceived symbol vecto y by an equalization matix G i.e. xˆ G y (8) ZF IJRA 015 Volume Issue 6 P a g e 310
Intenational Jounal of Reseach and Applications (Ap-Jun 015 Tansactions) Intenational Confeence on Emeging Tends in Electonics & Telecommunications (ICETET-15) Zeo Focing means that the mutual intefeence between the layes shall be pefectly suppessed [7]. This is deived fom the Mooe-Penose Pseudo Invese of GZF ( ) (9) but the noise is also multiplied by the invese of the channel matix, in geneal, the channel coefficient matix is less than 1, then its invese is geate than 1, that is a noise by a facto of geate than 1, must be amplified noise. So the eo ate will incease as the numbe of antennas inceases. 3. RESULTS AND DISCUSSION Matlab Simulations Fom Figue it can be seen that the detection of the ML, when the tansmitting and eceiving antennas ae moe, eo ate is lowe, so that the MIMO communication system can ovecome the advese multipath fading effects, to achieve the eliability of signal tansmission, but also can incease the system capacity and hence impoves the spectum efficiency[8]. Figue4. Ideal channel, ZF detection 1x1, x, 4x4 MIMO antenna system eo ate Figue.Pilot estimated channel, ML Detection MIMO System 1x1, x, 4x4 Antenna Pefomance Fom the esults which ae shown in figue 5 and 6 it can be seen, the use of ML detection, the ideal channel and the estimated channel eo ate oveall tend is same, but the ideal channel bit eo ate than the estimated channel bit eo ate is low, indicating that based on the pilot channel estimation in MIMO communication system is feasible. Using ZF detection, ideal channel and estimated channel eo ate oveall tend the same, but the ideal channel bit eo ate than estimated channel bit eo ate is high, it is geneally not ecommended fo pilot-based channel estimation using ZF detection algoithm[9]. Figue3. Pilot sequence estimated channel, ZF Testing unde, MIMO System 1x1, x, 4x4 Antenna pefomance compaisons Fom figue 3 and figue 4 it can be seen in ZF detection, when the tansmitting and eceiving antennas, the moe the highe the bit eo ate, is zeo focing detection algoithm to the eception signal by the invese of the channel matix, the othe uses of its the intefeence can be eliminated, Figue5. Estimate the channel,mld And ZFTesting unde 4x4Antenna MIMOCompaative analysis of the bit eo ate cuve IJRA 015 Volume Issue 6 P a g e 311
Intenational Jounal of Reseach and Applications (Ap-Jun 015 Tansactions) Intenational Confeence on Emeging Tends in Electonics & Telecommunications (ICETET-15) It can be seen fom the esults of figue 5 and figue 6, when using the ideal channel, MLD algoithm eo ate than the ZF algoithm is small, and the same geneal tend, but at high SNR pat MLD algoithm BER than the ZF algoithm is obviously much lowe, indicating that global seach algoithm MLD excellent pefomance. Figue7. ML device utilization summay fo Vitex 4, 4vsx5ff668-1 Deivce Figue6. Ideal channel, MIMO 4x4 antenna system MLD detection and ZF detection pefomance compaison When using the estimated channel, MLD algoithm and the ZF algoithm BER less, whee MLD algoithm is slightly lowe SNR. Fom the computational point of view and un time, when the num = 56, loop = 1000, the use of a total of about MLD algoithm 100s, the use ZF algoithm needs a total time of about 180s. MLD algoithm is much lage than the amount of computation ZF algoithm, the time is also longe than the ZF algoithm. FPGA Implementation The achitectue was designed concentated on geat pefomance and low cost. The achitectue was defined in adwae Desciption Language and synthesized to Xilinx Vitex 4 FPGAs. The synthesis esults demonstate that the achitectue is capable to accomplish pocessing at faste ates and attains the eal-time equiements. The achitectue is consuming fewe esouces when compaed to elated woks [10]. Figue8. ZF device utilization summay 4. CONCLUSION Though this appoach, we know that the multipath MIMO wieless channel with tansmitte and eceive as a whole is optimized to achieve high communication capacity and spectum utilization. This is a nea-optimal joint aispace domain divesity and intefeence cancellation pocessing. This is stated though the establishment of MIMO communication system, and based on pilot estimates the channel, and then using the estimated channel espectively MLD and Zeo-focing signal detection, and finally the application of MATLAB on the system simulation, dawing, obsevation gaphics, simulation esults BER analysis. Fom the above pocess fo the poject, we futhe undestand the pefomance of MIMO system outstanding advantages fo the futue laid the foundation fo leaning 4G communications. IJRA 015 Volume Issue 6 P a g e 31
Intenational Jounal of Reseach and Applications (Ap-Jun 015 Tansactions) Intenational Confeence on Emeging Tends in Electonics & Telecommunications (ICETET-15) REFERENCES 1. Andea Goldsmith, Syed Ali Jafa, Niha Jindal, and iam Vishwanath, Capacity Limits of MIMO Channels Poc.IEEE J. Selected aeas in Comm., Vol.1, No.5, pp.684-70, Jun.003.. G. Tsoulos, MIMO system technology fo wieless communication CRC pess, 006,pp.105-13. 3. Chistoph Windpassinge and Robet F..Fische, Low-Complexity Nea- Maximum-Likelihood Detection and Pecoding fo MIMO Systems using Lattice Reduction, Poc. IEEE Infomation Theoy Wokshop, Pais, Fance, pp. 345 348, Mach 003. 4. Ye. Li, Godon. Stübe, Othogonal Fequency Division Multiplexing fo Wieless Communications, Spinge, 006.pp119-16. 5. S. Colei, M. Egen, A. Pui, "Channel estimation techniques based on pilot aangement in OFDM systems". Poc.IEEE Tans. Boadcasting, vol. 48, pp. 3-9, Sept. 00. 6. K.-K. Wong and A. Paulaj, Efficient nea maximum-likelihood detection fo undedetemined MIMO antenna systems using a geometical appoach, EURASIP J. Wieless Commun. and Netwoking, Oct.007. doi.10.1155/007/8465. 7. D. Shiu and J.M. Kahn. Layeed Space-Time Codes fo Wieless Communications using Multiple Tansmit Antennas. Poc.IEEE Int.Conf.Comm. (ICC 99), Vancouve, B.C. June 6-10 1999. 8. G.V.V.Shama, V.Ganwani, U.B.Desai and S.N.Mechant, Pefomance Analysis of Maximum Likelihood Detection fo Decode and Fowad MIMO Relay Channels in Rayleigh Fading, IEEE Tans.Wieless Comm.,Vol9,No.9,pp880-889, Sept 010. 9. C. Wang et al., On the pefomance of the MIMO zeo-focing eceive in the pesence of channel estimation eo, IEEE Tans. Wieless Comm., vol. 6, pp. 805 810, 007. 10. Le Sun, Wei Yang and u uang, FPGA Implementation of V-BLAST Detection Algoithm In MIMO System, Poc.IEEE Youth Conf.(YC-ICT 09), pp. 134-137,009. IJRA 015 Volume Issue 6 P a g e 313