PERFORMANCE ANALYSIS OF DIGITAL BEAMFORMING ALGORITHMS

Transcription

1 IJRET: Intrnational Journal of Rsarch in Enginring and Tchnology ISSN: pissn: PERFORMANCE ANALYSIS OF DIGITAL BEAMFORMING ALGORITMS Sushma K M, Manjula Dvi T Digital Communication and Ntworking and Tlcommunication Dpartmnt Dayanand Sagar Collg of Enginring, Bangalor Abstract Intrfrnc rduction is ndd for bing abl to ffctivly communicat with mobil usrs On mthod that can b usd as a solution to rduc intrfrnc is using smart antnna tchnology Along with intrfrnc rduction thr is incras in capacity and covrag Adaptiv bam forming tchniqus forms a narrow bam and dircts it towards dsird usrs and thrby forms nulls in intrfrnc dirctions This improvs signal to nois ratio Th study documntd in this papr shows th prformanc of variants of th Last Man Squar (LMS) Algorithm basd approachs, in th ralm of bam formation, usd arlir in adaptiv filtring, aiming at rduction in complxity, without considrabl dgradation in prformanc Th proposd algorithms namly Signum- Error Last Man Squar Algorithm along with LMS Algorithm ar tstd for computational complxity in wight vctor updating and obsrvd that th ons proposd prform bttr than th convntional Last Man Squar algorithm with rspct to rducd complxity in bam formation, making thm suitabl for high spd digital communication systms This papr will analyz Last Man squar algorithm and Sampl Matrix Invrs algorithm for th prformanc Th consqunc will dmonstrat th Last Man Squar as a Solution to jammr cancllation whn compard to Sampl Matrix Invrs Kywords: Smart Antnna, Bamforming, Intrfrnc, Adaptiv array, Uniform linar array *** INTRODUCTION In rcnt yars a substantial incras in th dvlopmnt of broadband wirlss accss tchnologis for volving wirlss intrnt srvics and improvd cllular systms has bn obsrvd It is widly forsn that in th futur an normous ris in traffic for mobil and prsonal communication systms shall aris, du to an incras in numbr of usrs and introduction of nw high bit rat data srvics Thr ar crtain ngativ factors lik co-channl intrfrnc and fading in th radiation nvironmnt contributing to th limit in th capacity Smart antnna is a dvlopmnt in this dirction to fulfill th fatur rquirmnts of mobil ntworks A smart antnna is multipl antnnas that prform spatial signal procssing using diffrnt algorithms for dirction of arrival stimation and for bamformation As said, two basic functions of any smart antnna ar: DOA Estimation and Adaptiv Bam Forming (ABF) In digital bamforming, signals ar convrtd from radio frquncis to an intrmdiat frquncy, and convrtd to digital form and thn down convrtd Thn this signal is givn to bamforming algorithms that ar burnt on digital signal procssors In this tchnology digital signal procssing is usd to stimat th dirction in which incoming RF nrgy is incidnt on an array of antnna lmnts Digital Bamforning rcivrs multiply ach usr s signal by complx wight vctors that adjust th xcitation amplituds and phass of th signal from ach antnna lmnt In this papr, Last Man Squar (LMS) adaptiv algorithm is chosn bcaus of its simplicity Jammr rduction can b implmntd using an array with strd bams If w us dirctional bams to communicat with mobils on th downlink, intrfrnc btwn th bas station and its narby co-channl bas stations will b lss than if it usd an Omni dirctional antnna With Adaptiv antnnas a bas station can communicat with two or mor mobils on th sam frquncy using SDMA, whr, many mobils communicat with a singl bas station on th sam frquncy Using highly dirctional bams and/or forming nulls in th dirctions of all but on of th mobils on a frquncy, th bas station crats multipl channls using th sam frquncy, but sparatd in spac Smart or adaptiv antnna systms can improv link quality by combating th ffcts of multipath propagation or constructivly xploiting th diffrnt paths, and incras capacity by rducing intrfrnc and allowing transmission of diffrnt data strams from diffrnt antnnas as wll This papr will addrss an approach to rduc intrfrnc in a noisy nvironmnt as a mans to nhanc intlligibl communications from on mobil usr to anothr Lastly, rsults prsntd will illustrat th utility of LMS algorithm in a congstd nvironmnt as a mans to null out potntial jammrs Volum: 3 Spcial Issu: 3 May-4 NCRIET-4, 889

2 IJRET: Intrnational Journal of Rsarch in Enginring and Tchnology ISSN: pissn: RELATED RESEARC WORK Th latst advancmnts in th fild of smart antnna, undrstanding of Dirction of Arrival and Bamforming algorithms ar xplicitly dscribd in th following paprs Suzana Lamar [] has xplaind about LMS algorithm This algorithm dos stimation of corrlation btwn succssiv rror sampls With this stp siz is controlld Gorg V Tsoulos [] prsnts a gnral ovrviw of smart antnna along with thir application and potntial bnfits for mobil communication systms Sungwon Choi [4] bgins with th dscription of various typs of smart antnna and thn provids information about multipl bnfits involvd such as incrasd transmit powr and capacity It raffirms th fact that utilizing diffrnt charactristics of smart antnnas can lad to svral oprational bnfits for a communication systm Jack Wintrs [5] in his papr givs an ovrviw about th nd for Smart Antnnas (SA) in mobil systms Th papr brifs about th various Bam forming and DOA algorithms along with thir hardwar implmntation Finally th author has concludd th fact that, th implmntation of SA has littl impact on th physical layr thrby improving th Quality of Srvic (QoS) and rducing intrfrnc at th sam tim 3 BEAMFORMER A Bamformr is a st of antnnas arrangd in a linar fashion with outputs that can b lctronically strd Th snsor array consists of isotropic antnna lmnts that tak spatial data from th wavs mittd by signal sourcs Th signal rcivd at ths snsors is snt for computation of wights Bamforming can b dfind as adjusting th snsitivity of antnna in a particular dirction for a givn frquncy r signals ar isolatd according to thir dirctional charactristics and frquncy contnt Thus, a bam formr can b viwd as a spatio-tmporal filtr Th aim of bamforming is to isolat th signal of th dsird usr from intrfrnc and nois Th wights of th filtr ar continuously changd and ar adaptd according to th rcivd signal Such adaptiv filtrs attmpt to filtr out jammr signals, which ar rcivd by th antnnas from dirctions othr than th rquird signal sourc 4 SWITCED BEAMFORMING In Switchd Bamforming th array wights dos not dpnd on th input or output array signals Th complx wights ar slctd to form a bam in particular, prdtrmind dirctions, from an xisting library of wights Switchd bam antnna systms form multipl fixd bams with incrasd snsitivity in spcific dirctions Ths systms find out th signal strngth, chosn from on of svral prdtrmind, fixd bams and switch from on bam to anothr as th mobil usr movs throughout th rang Instad of shaping th dirctional bam pattrn with th mtallic proprtis and physical dsign of a singl lmnt, switchd bam systms combin th outputs of multipl antnnas in such a way as to form dirctional bams with mor spatial slctivity than can b gaind with convntional, on-lmnt approachs 5 DIGITAL BEAMFORMING Adaptiv Bamforming is an advancd approach to bamforming This is a tchniqu in which an array of antnnas is xploitd to achiv maximum rcption in a spcifid dirction whil rjcting signals of th sam/diffrnt frquncy from othr dirctions Th wights ar computd and adaptivly updatd in ral tim basd on signal sampls Doing so, it s asy to track th usr dirctions continuously Th adaptiv procss prmits narrowr bams in look dirction and rducd output in othr dirctions, which rsults in significant improvmnt in Signal to Intrfrnc Nois Ratio (SINR) With this tchnology, ach usr's signal is transmittd and rcivd by th bas station only in th dirction of particular usr Thus thr will b drastic rduction in ovrall intrfrnc in th systm This is improvd by varying th wights applid to ach of th antnnas usd in th array 6 SAMPLE MATRIX INVERSE SMI is usd if th dsird and jamming signals ar known bfor or hav bn stimatd This provids th dirct and fastst solution to comput th optimal wights owvr, if th signals ar not known xactly, thn signal nvironmnt undrgos frqunt changs Thus, th signal procssing unit must continuously updat th wight vctor to mt th nw rquirmnts imposd by th varying conditions Th wight vctor must b updatd without a priori information which, lads to stimation of covarianc matrix R and cross-corrlation vctor r in a finit obsrvation intrval givn by quations R E[ X X Volum: 3 Spcial Issu: 3 May-4 NCRIET-4, 89 r E XS Whr, X is inducd signal matrix, X is hrmitian transpos of X, E is th xpctation oprator and S is rfrnc signal matrix Th quation to comput Lx wight vctor to str bam in look dirction and dirct nulls at intrfring sourc dirctions by using SMI algorithm is givn by Whr, ] w( R r R is invrs of autocorrlation matrix is cross-corrlation r R and

3 IJRET: Intrnational Journal of Rsarch in Enginring and Tchnology ISSN: pissn: Simulation Mthodology of SMI Comput th Lx string vctor for dsird dirction i a( ) i d sin d ( L)sin Comput th LxM array manifold vctor corrsponding to M intrfrnc sourc dirctions i A i d sin id sin,,, d ( L ) sin ( ) sin id ( L ) sin i d L ( M ) id sin M 3 Obtain signal sampls S by sampling continuous tim signal of basband frquncy(for simulation cosin wav sampls ar considrd) 4 Comput Lx cross-corrlation matrix r by using r E XS 5 Comput LxL covarianc matrix R by using R E[ X X ] 6 Th invrs of covarianc matrix R is found 7 Th wight vctor is computd by using quation w( R r 8 Th array factor is computd by using quation L AF w i j d sin( ) ( i) 9 Th valu of in quation varis btwn 9 9 and w (i) is hrmitian transpos wight updat vctor w ( 9 Array factor vrsus angls ar plottd 9 M 7 LEAST MEAN SQUARE Antnna arrays can b usd in dynamic nvironmnts, whr both th dsird and intrfring signals arriv from changing dirctions and with varying powrs Antnna arrays mploy adaptiv wighting algorithms that chang th wights basd on th rcivd signals to incras th prformanc of th array This wighting procss aids in dircting th antnna bam to point in th dirction of th mobil usr, and provids th capability for nulling out potntial jammrs Th LMS algorithm is on of th most widly usd adaptiv algorithms Th primary objctiv for th LMS algorithm is to adaptivly produc wights that dcras th man squard rror btwn a dsird signal and th arrays output Thrby, incrasing th rcption in th dirction of th dsird signal and rducd rcption from th intrfring signal Th LMS algorithm is most commonly usd bcaus of its simplicity and rasonabl prformanc Sinc it is an itrativ algorithm it can b usd in a highly tim varying signal nvironmnt Also, thr ar svral variants of th LMS algorithm that dal with th shortcomings of its basic form 7 Simulation Mthodology of LMS Comput th Lx string vctor for dsird dirction i a( ) i d sin d ( L)sin Comput th LxM array manifold vcto corrsponding to M intrfrnc sourc dirctions id ] A id ( L sin id sin,,, M id sin M ) sin id ( L ) sin id ( L ) sin ( M ) 3 Obtain signal sampls S by sampling continuous tim signal of basband frquncy (For simulation sin wav sampls ar considrd) Comput th autocorrlation matrix 4 Comput th stp siz by using max R Volum: 3 Spcial Issu: 3 May-4 NCRIET-4, 89

4 IJRET: Intrnational Journal of Rsarch in Enginring and Tchnology ISSN: pissn: Comput th following for all signal sampls n N Whr, s s N is th total numbr of signal sampls M ) s( i( a( i ) n ( i x( a( LMS Algorithm T y( w( x( 33 ( s( y( 6 Th array factor is calculatd as Angl(dg) L AF w i j d sin( ) ( i) 9 7 Array factor vrsus angls ar plottd 8 RESULTS Tabl : Inputs to th algorithms for lss lmnts Algorithms Antnna Look Jammr Elmnts Dirction Dirctions SMI 8 45 [,6,7] LMS 8 45 [,6,7] 5 SMI Algorithm Fig- : Polar Plot of LMS for 8 antnna lmnts Tabl : Inputs to th algorithms for mor lmnts Algorithms Antnna Elmnts Look Dirction Jammr Dirctions SMI 45 [,6,7] LMS 45 [,6,7] 5 SMI Algorithm Angl(dg) Fig- 3: Polar Plot of SMI for antnna lmnts Angl(dg) Fig- : Polar Plot of SMI for 8 antnna lmnts 5 LMS Algorithm Angl(dg) Fig- 4: Polar Plot of LMS for antnna lmnts Volum: 3 Spcial Issu: 3 May-4 NCRIET-4, 89

5 IJRET: Intrnational Journal of Rsarch in Enginring and Tchnology ISSN: pissn: DISCUSSION r, w first compard th adaptiv algorithm with th convntional approach for a uniform linar array of 8 and lmnts Th spacing of vry two array lmnts is half wavlngth in ordr to avoid grating lob Digital bam forming (DBF) tchnology is progrssd with th dvlopmnt of adaptiv algorithms and architcturs Modrn bam forming systms Digital bamforming nabls full utilization of th maximum numbr of dgrs of frdom in th arraylast Man Squar (LMS) algorithm is bing chosn to updat complx wights to form th bam in th dsird dirction Adaptiv Bamforming is th procss of rcursivly updating th complx wights for ach of th antnna lmnts usd in th array to achiv maximum rcption in dsird dirction Th Adaptiv DBF rcivr prforms a highly slctiv spatial filtring of arriving signals Divrs adaptiv bamforming algorithms with varying complxitis hav bn widly usd in diffrnt aras such as civilian mobil communications, sonar, radar, radio astronomy tc Most adaptiv bamforming algorithms ar concrnd with th maximization of th signal to nois ratio W hav chosn LMS as an adaptiv algorithm bcaus it is vry simpl and asy to apply sinc it rquirs no calculations or complx masurs This work can also b xtndd for othr adaptiv algorithms dpnding on th nd and application Us of adaptiv antnna in xisting systms not only rducs powr consumption and intrfrnc but also incrass spctral dnsity in wirlss communication Th contribution to a grnr nvironmnt is twofold whil unwantd radiation xposur is rducd th consumption of disl fuls that nrgiz th bas stations and thir transmittrs, ar also rducd Som rcnt survy basd rports by wll known nvironmntal agncis in India [8] indicats vry havy disl consumption by bas stations Th us of smart antnna promiss a drastic rduction of disl consumption in cllular communications CONCLUSIONS Th simulation rsults for 8 and ULA of antnna lmnts Th distanc btwn adjacnt antnna lmnts is lamda by Th primarily usd for this xprimnt was Matlab softwar Th antnna outputs will b pur sinusoidal wav with changd amplituds and phass, and for ach lmnt in th array thr is a tim dlay at which th signal arrivs at ach lmnt Both th SMI and LMS bamforming algorithms wr discussd, and th LMS bam formr dmonstratd its ability to dirct th antnna bam in th dirction of th mobil usr as wll as its ability to null out th jammrs This tchniqu is significant in highly tim varying signal nvironmnts and could srv as a potntial solution to jammr rjction Figur [3,4] clarly shows that SMI fails to form bam in th look dirction without intrfrnc, whr as LMS works wll in bam formation and rducd jamming ffct compard to SMI REFERENCES [] Suzanna LaMar, ugh Nguyn, and Paul Zavidniak, Bamforming Solutions for Intrfrnc Rduction for igh Altitud Airborn CDMA Systms IEEE- [] Gorg V Tsoulo, Smart antnnas for mobil communication systms: Bnfits and Challngs, IEEE procdings of intrnational confrnc on information tchnology, Vol-, Issu-,pp [3] D G Manolakis, Vinay K Ingl and Stphn M Kogon, Statistical and Adaptiv Signal procssing spctral stimation, signal modling, Adaptiv filtring and array procssing, Mc Graw ill [4] Sungwon Choi, ong-min Son, and Tapan K sarkar, Implmntation of a Smart Antnna Systm on a Gnral-Purpos Digital Signal Procssor Utilizing a Linarizd CGM, Digital Signal Procssing Journal, July-997, Vol-7, pp 5-9 [5] Jack Wintrs, Smart Antnnas for Wirlss Systms, procdings of IEEE intrnational confrnc on signal procssing, Jun 5, pp 7-9 [6] Sungsoo, Sungwon and Tapan K sarkar, An Adaptiv Bamforming Algorithm with a Linar Complxity for a Multipath Fading CDMA Channl, IEICE Transaction [7] Disl to Powr Tlcom Towrs - Th Big Problm, authord by VNL Pvt Ltd India, availabl at th URL citd: wrs-th-big-problm/ publishd on 5 Apr, BIOGRAPIES Sushma KM rcivd BE in Elctronics and Communication from PDIT, ospt Pursuing MTch in Digital Communication and Ntworking from DSCE, Banglor Manjula Dvi T currntly working as Associat Profssor in Dayanand Sagar Collg of Enginring, Bangalor Sh has rcivd BE in Elctronics and Communication and ME in Elctronics and Communication and pursuing Phd in Imag Procssing r ara of intrst includs Imag Prosssing, DSP and Communication Volum: 3 Spcial Issu: 3 May-4 NCRIET-4, 893