UNIVERSITY OF CINCINNATI

Size: px
Start display at page:

Download "UNIVERSITY OF CINCINNATI"

Transcription

1 UNIVERSITY OF CINCINNATI DATE: Marc st, 004 I, Bn Xu ereby submt ts as part of te requrements for te degree of: Master of Scence n: Electrcal Engneerng It s enttled: A Blnd Space-Tme Decorrelatng RAKE Recever n a DS- CDMA system n Multpat Cannels, Approved by: Dr. James Caffery, Jr. Dr. oward Fan Dr. Qng An Zeng

2 A Blnd Space-Tme Decorrelatng RAKE Recever n a DS- CDMA System n Multpat Cannels A tess submtted to te Dvson of Researc and Advanced Studes of te Unversty of Cncnnat n partal fulfllment of te requrements for te degree of MASTER OF SCIENCE n te Department of Electrcal & Computer Engneerng and Computer Scence of te College of Engneerng 004 by Bn Xu B.S., Tsngua Unversty, Beng, Cna, 99 Commttee Car: Dr. James Caffery, Jr. Dr. oward Fan Dr. Qng An Zeng

3 Abstract Ts paper addresses te problem of blnd multple access nterference MAI and ntersymbol nterference ISI mtgaton n drect sequence code dvson multpat access DS/CDMA systems. Studes sow tat multuser detecton can be performed wtout te knowledge of users cannel parameters n a frequency-selectve fadng cannel, and tese approaces are of g computatonal complexty. In ts tess, a space-tme decorrelatng RAKE recever s developed n te AGN cannel. Ts metod s blnd because t only needs to know te desred user s sgnature sequence and tmng nformaton. A tme-varyng multpat cannel tat as te desred temporal-spatal correlaton propertes s smulated, and s used to evaluate te performance of te recever. e also develop an adaptve algortm for te proposed recever. Smulatons sow tat te proposed recever s near-far resstant, able to elmnate MAI and ISI effectvely, and as g performance wt low complexty.

4

5 Acknowledgements I am sncerely grateful to Dr. James Caffery, Jr. for beng my advsor and provdng gudance for ts tess work. e ntroduced me to CDMA systems, and s nsgt and motvaton were nvaluable and nsprable. I would also lke to express my apprecaton to my commttee members for ter correctons and comments on ts work. I am grateful to all members of te reless Systems Researc Lab wo are very elpful to me. Fnally, most of all, I tank my wfe for er uncondtonal love and support

6 Table of Contents Table of Contents... Lst of Fgures... 3 Capter Introducton Researc Background Organzaton of Tess... 9 Capter Spread Spectrum and CDMA Sgnal Detecton Rado Propagaton Envronment Pat Loss..... Fadng Slow Fadng Sadowng Fast Fadng Doppler Spread: Tme-Selectve Fadng Delay Spread: Frequency-Selectve Fadng Angle Spread: Space-Selectve Fadng Space-Tme Processng Temporal Processng Spatal Processng Space-tme processng....3 Smulaton Examples... 5 Capter 3 Tme-varyng Multpat Vector Cannel Smulator Tme-varyng Multpat Vector Cannel Second Order Caracterzaton... 35

7 3.3 Vector Cannel Smulator Complex Pat Vector Generator Tme-correlaton Sapng Flter Spatal Transformaton Interpolator Smulaton Examples Capter 4 Space-Tme Decorrelatng RAKE Recever for a DS/CDMA System Introducton System Model Space-Tme Decorrelatng Detector Smulaton Results Capter 5 Adaptve Implementaton LMS-based Algortm Smulaton Results Capter 6 Conclusons and Future ork... 77

8 Lst of Fgures FIGURE. MULTIPAT MODEL IT BASE ANTENNA ARRAY... FIGURE. TEMPORAL TIME-ONLY PROCESSING FIGURE.3 EQUALIZER... 7 FIGURE.4 BEAMFORMER FIGURE.5 SPACE-TIME RECEIVER... FIGURE.6 QPSK MODULATED SIGNAL CONSTELLATION... 8 FIGURE.7 RECEIVED SIGNAL CONSTELLATION AT TE EQUALIZER INPUT FIGURE.8 SIGNAL CONSTELLATION AFTER TIME-ONLY PROCESSING FIGURE.9 SIGNAL CONSTELLATION AFTER SPACE-TIME PROCESSING... 9 FIGURE.0 BEAM PATTERN FOR TE SPACE-TIME PROCESSING FIGURE 3. UNIFORM LINEAR ARRAY IT TREE ELEMENTS... 3 FIGURE 3. TIME-VARYING MULTIPAT VECTOR CANNEL SIMULATOR.. 38 FIGURE.3 TE COMPLEX PAT VECTOR GENERATOR FOR TE IT PAT39 FIGURE 3.4 SPACE-CORRELATION TRANSFORMATION FIGURE 3.5 INTERPOLATOR FIGURE 3.6 LINEAR INTERPOLATION ILLUSTRATION FIGURE 3.7 CUBIC SPLINE INTERPOLATION ILLUSTRATION FIGURE 3.8 DESIRED AND OBTAINED TEMPORAL CORRELATION FIGURE 3.9 CANNEL COEFFICIENT MAGNITUDES VS. TIME FOR TE FIRST ANTENNA

9 FIGURE 3.0 CANNEL COEFFICIENT MAGNITUDES BEFORE AND AFTER INTERPOLATION FOR PAT OF TE FIRST ANTENNA ELEMENT FIGURE. STRUCTURES OF SPACE-TIME RECEIVERS FIGURE 4. RECEIVER STRUCTURE FIGURE 4.3 BER PERFORMANCE FIGURE 4.4 NEAR-FAR RESISTANCE FIGURE 4.5 BER PERFORMANCES AT f d = 50z AND f d = 00z FIGURE 5. LMS ALGORITM STRUCTURE... 7 FIGURE 5. PERFORMANCE OF TE LMS BLIND ADAPTIVE ALGORITM FIGURE 5.3 STEADY-STATE BER PERFORMANCE FIGURE 5.4 PERFORMANCE OF ADAPTIVE ALGORITM FOR EIGENVECTOR ESTIMATION

10 Capter Introducton. Researc Background reless Communcatons for moble telepone and data transmsson s undergong very rapd development. Many of te emergng wreless systems wll ncorporate consderable sgnal-processng ntellgence n order to provde advanced servces suc as multmeda transmsson. In order to make optmal use of avalable bandwdt and to provde maxmal flexblty, many wreless systems operate as multple access systems n wc cannel bandwdt s sared by many users on a random-access bass. As demand for wreless communcatons contnues to grow, trd-generaton cellular communcatons systems are beng standardzed to provde better voce and data servces, and drectsequence Code-dvson multple access DS-CDMA tecnques are beng used as ts bass. In Nort Amerca, te second generaton DS-CDMA standard, IS-95, s te bass for a trd-generaton system CDMA000. In Japan and Europe, a trd-generaton wdeband CDMA CDMA system s also beng developed [], []. An effort s beng made to merge tese systems nto a common, globe trd generaton CDMA standard. le CDMA possesses many ntrnsc advantages over te earler access tecnques suc as tme-dvson multple access TDMA and frequency-dvson multple access FDMA [], te capacty of current practcal CDMA systems s lmted by te multple access nterference MAI caused by code non-ortogonalty due to dverse penomena suc as asyncronous transmsson, mult-pat propagaton, or lmted bandwdt. Moreover, te presence of nter-symbol nterference ISI due to te tme-dspersve nature of wreless cannels s often neglected n low rate CDMA systems, but t becomes a maor problem n wdeband CDMA systems. In addton, a maor tecnologcal urdle 5

11 of CDMA systems s te near-far problem: te bt-error-rate BER of conventonal recever a matced flter for te user of nterest s so senstve to dfferences between te receved energes of te desred user and nterferng users tat relable demodulaton s mpossble unless strngent power control s exercsed. Te optmum multuser recever for asyncronous mult-access Gaussan cannels sows tat te near-far problem s overcome by a more sopstcated recever tat accounts for te presence of oter nterferers n te cannel. Ts recever optmum recever was sown [4] to attan essentally sngle-user performance assumng tat te recever knows te followng: Te sgnature waveform of te desred user; Te sgnature waveform of te nterferng users; 3 Te tmng bt-epoc and carrer pase of te desred user; 4 Te tmng bt-epoc and carrer pase of eac of te nterferng users; 5 Te receved ampltudes of te nterferng users relatve to tat of te desred user. Dfferent tecnques ave been proposed to suppress MAI as well as ISI usng lnear flterng. Te conventonal recever only requres and 3, but t s severely lmted by te near-far problem, and even n te presence of perfect power control, ts BER s orders of magntude far from optmal. Te optmum detector for te asyncronous multpleaccess Gaussan cannel [4] sows near-far resstance and sgnfcant performance mprovement over tat of te conventonal recevers by ontly detectng all of te users sgnals. owever, te optmum recever needs to know,, 3, 4, and 5, and te computatonal complexty of te optmum recever grows exponentally wt te number 6

12 of actve users. Decorrelatng recevers [5] are suffcent n order to aceve optmum resstance aganst te near-far problem, and, at te expense of slgt ncrease over te mnmum BER, te decorrelatng recever avods te exponental complexty n te number of actve users of te optmum multuser detector. But decorrelatng recevers requre,, 3, and 4. Conventonal lnear mnmum mean square error MMSE recevers [5], [6] offer superor performance wt lnear complexty. owever, tey need to know,, 3, 4, and 5. Te adaptve MMSE detectors n [6] and [7] substtute te need to know, 4, and 5 by te need to ave tranng data sequences for every actve user. Te typcal operaton of tose adaptve multuser detectors requres eac transmtter to send a tranng sequence at start-up wc te recever uses for ntal adaptaton. After te tranng pase ends, adaptaton durng actual data transmsson occurs n decsondrected mode. owever, any tme tere s a drastc cange n te nterference envronment, decson-drected adaptaton becomes unrelable, and te data transmsson of te desred user must be suspended and a fres tranng sequence s requred. Tus, te relance on tranng sequence s cumbersome n most CDMA systems, were one of te most mportant advantages s te ablty to ave completely asyncronous and uncoordnated transmssons tat swtc on and off autonomously. So tat mples te need for blnd adaptve recevers, and many papers ave talked about t [0]. owever, most proposed blnd estmaton algortms nvolve computatonally ntensve operatons suc as Sngular Value Decomposton SVD and terefore may be probtve n practce. Tus, several recevers based on te lnear constraned mnmum varance LCMV ave been proposed to decrease te computatonal complexty. Te autors n [0] and [] ntroduce blnd adaptve recevers wc mnmze te mean output energy 7

13 MOE wt te knowledge of te desred user s sgnature sequences and ts tmng, and tey also sow tat te canoncal MOE detector s, n fact, te MMSE detector. Te autors n [6] and [35] ntroduce constraned MOE recevers wc are used n a sngle cannel envronment. Based on a LCMV crteron smlar to tat used n [6] and [35], te autors n [8] develop a decorrelatng recever n a multpat envronment for a syncronous CDMA system by usng a partcular constrant. All of tese recevers are developed for a sngle antenna at te recever. Anoter approac to nterference suppresson n wreless systems s troug space-tme processng usng an antenna array at te recever. Ts approac apples varous dversty tecnques to mprove te detected data qualty so as to ncrease te system capacty. Among tese tecnques, te spatal and temporal combner suc as te beamformer and te RAKE recever as been proposed and ts superor performance as already been verfed n many papers [8], [9]. To empasze te specfed user s sgnal, te temporal combner, suc as te RAKE recever, performs a coerent combnng of te temporally spread multpat components wereas te beamformer combnes te spatally spread components. Te goal of ts tess s to nvestgate and obtan a blnd recever, wc does not requre tranng sequences and requres knowledge of only and 3, tat s, te same knowledge as te conventonal recever, wle ts recever s near-far resstant. In te meantme, ts recever also apples space-tme processng tecnques to mprove ts performance n a frequency-selectve, asyncronous cannel. 8

14 . Organzaton of Tess Te rest of te tess s structured as follows. Capter begns wt a bref ntroducton to rado propagaton because ts s mportant to sgnal detecton. Ten we ntroduce varous CDMA recever structures: tme-only, space-only, and space-tme processng, and compare ter performances n detecton. Capter 3 ntroduces a tme-varyng multpat vector cannel smulator tat wll be used to analyze te performance of te space-tme recever. Te smulator generates cannel coeffcents tat smulate a tmevaryng multpat cannel and as te desred temporal and spatal correlatons. Capter 4 models and proposes te space-tme decorrelatng recever. Performances based on computer smulaton are gven. In Capter 5, an LMS-based adaptve algortm s developed. Capter 6 provdes concluson and future work. 9

15 Capter Spread Spectrum and CDMA Sgnal Detecton Ts capter wll present some of te caracterstcs of spread spectrum systems. Tere are several ways to spread te sgnal spectrum, eac wt ts own partcular set of advantages [5]. Te common measure s called Processng Gan tat reflects te degree of spectral spreadng. e wll focus on drect-sequence spread spectrum systems. Frst, we wll revew te propagaton envronment n wc te rado sgnals are passed. By dong so, we wll understand te problems tat varous recevers ave tred to solve. Ten, we wll brefly dscuss several sgnal detecton metods: tme-only, space-only, and spacetme, and compare ter performance by ter BER beavor.. Rado Propagaton Envronment Rado sgnals generally propagate accordng to tree mecansms: reflecton, dffracton, and scatterng. en sgnals arrve at te recever, te orgnal sgnal arrves from several dfferent pats, called multpat, as Fgure. sows. Every pat as ts own delay, attenuaton, and drect of arrval DOA. Ts secton ntroduces te rado cannel and ts effects. 0

16 Delay Pat Ampltude Fadng Delay L Antenna Array Pat Ampltude L Fadng L Fgure. Multpat Model wt base antenna array... Pat Loss It s well known tat te rado sgnals propagate n te free space n te form of an electromagnetc wave, and te ntensty decays wt te square of te rado pat lengt. Tus te receved power s [] P r = P λ t 4 π d g t g r. were P t and P r are transmtted and receved sgnal power, respectvely, λ s wavelengt, gt and g r are te gans of transmtter antenna and recever antenna, respectvely, and d s te dstance between transmtter and recever. owever, free space propagaton does not apply n a moble rado envronment and te propagaton pat loss depends not only on te dstance and wavelengt, but also on te antenna egts of te moble statons MS and base statons BS, and te local terran caracterstcs suc as buldngs and lls. Te ste specfc nature of rado propagaton makes te teoretcal predcton of pat loss

17 dffcult and tere are no easy solutons. In practce, we can approxmate te receved sgnal power as P r = t r P t g t g r d. were t and r stand for te effectve egts of transmtter antenna and recever antenna, respectvely. Te effectve pat loss follows an nverse fourt power law exponent equal to four tat result n a loss of 40dB/decade []... Fadng In addton to pat loss wc causes sgnal attenuaton, receved sgnal power s also attenuated or vared because of te envronmental dsturbance, called fadng, wc s represented by [3] γ t = γ. γ.3 s r were γ s s called slow fadng and represents long-term tme varatons of te receved sgnal, wereas γ r s called fast fadng, sort-term fadng or multpat fadng. Te slow fadng γ s s te envelope of te sgnal γ t...3 Slow Fadng Sadowng Slow fadng s usually caused by te sadowng effect of mountans or buldngs, and s affected by antenna egt, transmsson frequences and envronments. From [], slow fadng can be descrbed by a log-normal dstrbuton. Its probablty densty functon pdf s

18 log x µ exp p x = πσ x σ 0 x > 0 x < 0.4 were x s a random varable, representng sgnal level, and µ and σ are te mean and standard devaton, respectvely, all n db...4 Fast Fadng In a fast fadng cannel, te cannel mpulse response canges rapdly wtn te symbol duraton. Suppose tat a scattered sgnal as random ampltude wt angle of arrval unformly dstrbuted n [0, π]. Ten te receved sgnal envelope as a Rayleg dstrbuton wt pdf [] exp y / σ y 0 p y = σ. 0 y < 0.5 If tere exsts lne of sgt LOS between transmtter and recever, ten te dstrbuton s Rcean wt pdf [3] exp p y = σ 0 { y + s / σ } ys J σ 0 y 0 y < 0.6 were s s te average power, J p s te modfed pt-order Bessel functon of te frst knd...5 Doppler Spread: Tme-Selectve Fadng en a cannel s tme-varant t s referred to as possessng tme-selectve fadng. Te moble staton ntroduces a Doppler, or frequency, sft nto te receved sgnal. Doppler spreadng results n te sgnal bandwdt beng stretced so tat te receved sgnal s 3

19 bandwdt s dfferent greater or less from tat of te transmtted sgnal. Takng te Doppler power spectrum to be te Fourer transform of te tme autocorrelaton of te cannel mpulse response, te Doppler spread s defned as te support of te Doppler power spectrum [3]. In a smple case tat assumes scatters unformly dstrbuted n angle [0, π] around te moble staton, te baseband power spectrum densty psd of te vertcal electrcal feld as te followng form [3] = 3σ f f c S f f c f m < f < f c + πf m f m / f m.7 were f m v = s te maxmum Doppler sft, λ c s te wavelengt of te carrer, and σ λ c s te total average receved power from all multpat components. e can estmate at wat transmtted sgnal duraton dstorton becomes notceable by referrng to te cannel s coerence tme. Te cannel s coerence tme s approxmately nversely proportonal to te maxmum Doppler sft experenced by te sgnal [3]...6 Delay Spread: Frequency-Selectve Fadng A moble rado cannel causes bot delay and Doppler spreadng. Delay spreadng results n two effects: tme dsperson and frequency-selectve fadng. Tme dsperson s a result of te sgnals takng dfferent tmes to cross te cannel by travelng dfferent propagaton pats. Frequency-selectve fadng occurs because te electrcal lengt of eac propagaton pat can expressed as a functon of frequency. A measure of te transmsson bandwdt at wc dstorton becomes apprecable s often based on te cannel s coerence bandwdt. Te coerence bandwdt ndcates te frequency 4

20 separaton at wc te attenuaton of te ampltudes of two frequency components becomes decorrelated suc tat te envelope correlaton coeffcent reaces a desgnated value. Tat s, wen te transmsson bandwdt s greater tan te coerence bandwdt, te cannel s called frequency-selectve fadng; wen te transmsson bandwdt s less tan coerence bandwdt, te cannel s called non-frequency-selectve, or flat fadng. Coerence bandwdt s approxmately nversely proportonal to te delay spread [3]...7 Angle Spread: Space-Selectve Fadng In te presence of multpat, te sgnal transmtted by eac user arrves at te recever not from one drecton, but from a contnuum. Te angle spread gves te range of angle values for wc sgnfcant energy s receved. At te recever, angle spread refers to te spread of angles of arrval of te multpats at te antenna array; lkewse, at te transmtter, angle spread refers to te spread of departure angles of te multpats. Angle spread causes space-selectve fadng, wc means tat sgnal magntude depends on te spatal locaton of te antenna. Coerence dstance represents te maxmum spatal separaton for wc te cannel responses at two antennas reman strongly correlated. Coerent dstance s used to caracterze te space-selectve fadng. Te larger te angle spread, te sorter te coerence dstance.. Space-Tme Processng In ts secton, we dscuss te caracterstc of temporal, spatal processng tecnques, and ten dscuss temporal-spatal processng. Temporal processng corresponds to 5

21 equalzers tat use a wegted sum of sgnal samples and spatal processng corresponds to smple beamformng tat uses a wegted sum of antenna outputs... Temporal Processng In a frequency selectve cannel, tere are multple replcas tat are resolvable n tme of te transmtted sgnal at te recever, traversng dfferent multpat. Tese multple copes can be combned to mprove te sgnal to nose rato SNR at te recever. Snce te sgnals are comng from dfferent pats, tey encounter ndependent fadng. Ts means tat f one of te pats s undergong a deep fade, t s very unlkely tat te sgnals from te oter pats are also encounterng fadng. As a result te recever stll as a good cance to attan acceptable fdelty. In a CDMA system, te recever can employ multple correlators to separate te multple copes of te sgnal and mtgate fadng. Ts recever, commonly known as a Rake recever [4], s sown n Fgure.. Fnger delay d spreadng waveform correlator Fnger rt delay d spreadng waveform correlator RAKE Combner decson Fnger L dealy dl spreadng waveform correlator Fgure. Temporal tme-only processng. 6

22 Te structure sown conssts of a bank of L RAKE fngers, eac correlatng to a dfferent delay of te receved sgnal. Te fnger outputs are ten combned to form a decson statstc. Te structure s equvalent to more practcal forms n wc te receved sgnal s frst fltered wt a cp pulse sapng matced flter and despreadng s performed usng receved cp samples and te spreadng code sequence. Temporal processng by te RAKE recever lets te CDMA system explot multpat dversty and make t nerently resstant to fadng. If te cannel s descrbed as a tapped-delay-lne TDL model, temporal processng can be mplemented by an equalzer, as sown n Fgure.3. Te desred sgnal st s transmtted troug te cannel, and receved at te recever antenna as xt. Te receved sgnal xt s contnuous, and sampled to obtan te dscrete sgnal xk wc s ten fltered by a lnear flter to get yk. Te equalzer s used to elmnate nter-symbol nterference ISI and mult-access nterference MAI [6]. Cannel SK XK X k- X k-m+ X k-m+ Z - Z - Z - * * M- * M * Σ Y k Fgure.3 Equalzer. 7

23 Assumng tere are Q users and te cannel s nose-free, we obtan te receved dscrete sgnal as x k = Q Σ q= q s k q.8 were =,, L, ] s te dscrete cannel gan coeffcents correspondng to q [ q q ql te multpat components and L s te number of resolvable pats. ere, we model te cannel as a tme-nvarant fnte mpulse response FIR cannel. s q = [ s k, s k, L, s k L + ] q q q T s te data sequence, and te dmenson of te equalzer s wegt vector s Mx, so te output of te equalzer s gven by y k = w x.9 k were x k = [ x k, x k, L, x k M + ] T s gven by Q x = q s q k, k q=.0 q s a Toepltz matrx wt dmenson M x M+L- gven by q q L 0 L 0 0 ql = 0 q q L ql 0 L 0 q, M M O O O O M 0 0 L 0 q q L ql. and T s q k = [ sq k, sq k, L, sq k L M + ] s a vector wt dmenson M+L- x. e rewrte te output of te equalzer n.9 as y k = w S k. 8

24 were = [,,..., Q ] wt dmenson beng M x QM+L- and T T T S k = [ s k, s k, L, sq k] s a vector wt dmenson QM+L- x. If we want to recover te sgnal s k from te receved sgnal xk were s k s te desred sgnal and cancel nterference, te followng zero-forcng ZF requrement must be satsfed n order to elmnate MAI and ISI: w = [,0, L,0]..3 Equaton.3 requres tat must be of full-column rank [5], but t s clear tat as te dmenson of M x QL+M-, and usually M < QL+M-. Ts mples tat ISI and MAI cannot be elmnated smultaneously f only temporal-processng s used. One soluton to ts problem s to oversample te receved sgnal by a factor P so tat as te dmenson of MP x QL+M- were MP > QL+M-. By oversamplng, can reac full-column rank, but te dsadvantages are less effcency of te system and nose enancement... Spatal Processng Te adaptve antenna array can aceve spatal dversty and mtgate multpat fadng. Ts s n addton to te nterference cancellaton attaned from steerng beams towards te desred user and/or steerng nulls n te drecton of nterferers. Te sgnal envelopes observed across te elements of an antenna array sould ave low cross-correlaton n order to aceve dversty gan. As a result, f te sgnal at one of te elements s gong troug a deep fade, t s gly unlkely tat te sgnals at te oter elements are encounterng smlar fades at te same tme. So tere s nearly always good sgnal 9

25 recepton on one of te antenna elements. Terefore combnng te sgnals from varous elements wll ncrease te SNR and te fdelty of te receved sgnal. Spatal processng can be mplemented by a beamformer, as sown n Fgure.4. w * s x k x k w * Σ y k s Q Q x Jk w J * Fgure.4 Beamformer. Assumng tere are Q actve users, and te cannel as no multpat effect, te receved sgnal s gven by x k = Q Σ q= q s k q.4 were q s a J x vector tat represents space-only cannel n te absence of delay spread, and terefore s q k s a scalar. J s te number of antenna array elements. Let s k be te desred sgnal. Te output of te beamformer s gven by y k = w x k.5 were w s te J x wegt vector. Te above equaton can be rewrtten n matrx form y k = w sk.6 0

26 were = [,,..., Q ] J x Q and sk = [s k, s k,..., s Q k] T Q x. Te ZF condton to retreve te frst user and cancel all MAI wll ten be w = [,0, L,0]..7 Te above equaton requres must be a full-column rank matrx. Ts requrement mples J Q,.e., te number of antenna array elements must be greater or equal to te number of actve users. Furtermore, consderng multpat effects, te requrement becomes J Q L q q=.8 were L q s te number of multpats for eac user. en L q and Q are large numbers, te number of antenna array elements s requred to be large, wc s not practcal. But unlke tme-only processng, space-only processng can be qute effectve aganst MAI. Te space-only and tme-only confguratons dscussed above use a ZF structure and dd not balance te effect of nose. As stated n [5], te sngle user matced flter recever s optmzed to fgt te background wte nose exclusvely, wereas te conventonal decorrelatng detector elmnated te multuser nterference dsregardng te background nose, and n te contrast, te MMSE mnmum mean-square error recever can be seen as a compromse soluton tat takes nto account te relatve mportance of eac nterferng user and te background nose. Moreover, te relatve performance of te two structures wll depend eavly on te cannel parameters [7]...3 Space-tme processng A Beamformer-RAKE cascades a beamformer wt RAKE recepton to process te sgnal bot n te spatal and te temporal domans. For eac fnger of te temporal

27 RAKE processor, tere s a beamformer to mprove te fdelty of te sgnal of tat partcular branc. At te front end of te recever s an antenna array. Te sgnals from te array are fed nto a set of spatal combners tat perform beamformng for dfferent multpat and eac wegt vector accentuates te sgnal from a partcular multpat component of te desred user. A temporal combner follows te spatal combner were te contrbuton from dfferent multpat from ter correspondng spatal combner s combned to explot te multpat dversty. Te structure s sown n Fgure.5. Te receved sgnal at eac antenna array element s fed nto an equalzer. Te wegt matrx acts partally as a beamformer. # w * x x x M- x M z - z - z - * * * w w M- w M #J w J * x J x J x JM- x JM z - z - z - * * * w J w JM- w JM Σ yk Fgure.5 Space-tme recever. e use te system model descrbed n Secton... e assume Q s te number of actve users, M s te number of taps n te equalzer, and J s te number of antenna array elements. Te antenna array response sould be taken nto account n te cannel model. tout loss of generalty and to smplfy te analyss, we assume te drecton of

28 3 arrval DOA for all multpats of a user be te same, θ q, tat s, te DOA at te antenna array for te qt user, or at least tey all fall nto te same beam. Te array response s a θ wc s a J x vector and s also known as a steerng vector. e get ],,, [ JxL ql q q q q q T q q θ θ θ θ a a a a L =..9 Substtutng te above nto., we get a modfed cannel matrx = ql q q q q q ql q q q q q ql q q q q q q θ θ θ θ θ θ θ θ θ a a a a a a a a a M L O O L L L O O M M L L Te dmenson of te space-tme cannel matrx for te qt user s JM x L+M-. Te receved sgnal at te antenna array s gven by n s x + + = = = + + Q q q q x M L M L k k k x JM JM M J J s s s x x x x x x x k,,, M M M M. were s s te transmtted nformaton, and we assume te nose n s spatally and temporally wte and Gaussan. From above, as dmenson JM x L+M-, were JM L+M-, so s full column rank and satsfes te ZF condton, and terefore, cancels bot ISI and MAI. In te case of nose, we no longer te consder ZF condton, but nstead dscuss ow to optmze te soluton by usng spatal-temporal tecnques.

29 A popular optmalty crteron s ST-MMSE space-tme mnmum mean-square error. In ST-MMSE, we obtan an estmate of te transmtted sgnal as a space-tme wegted sum of te receved sgnal and seek to mnmze te mean square error between te estmate and te true sgnal at any tme nstant. Te recever structure s sown n Fgure.5. In a space-tme flter, te wegt as te followng form w k L wm k k = M O M. w J k L wjm k. e defne te operator vec. as v vec[ v L v ] = M M. v M Terefore, w k = vec k JMx..3 Tus, we obtan a convenent formulaton for te space-tme flter output: y k = w k x k..4 Te ST-MMSE flter cooses te space-tme flter wegts to aceve te MMSE mnmum mean square error,.e., arg mn { w E w x k s k..5 Te soluton to ts LS least squares problem s gven by [6] w = R z.6 xx 4

30 were R xx = E[xx ] s te auto-correlaton matrx of te receved sgnal at te recever, or * te nput sgnal to te flter. Te vector z = E[xs ] s te cross-correlaton vector between te receved sgnal and te desred sgnal [6]. ST-MMSE combnes te strengts of tme-only and space-only processng, and trades MAI and ISI reducton aganst nose enancement. It wll cancel te MAI prmarly n te spatal dmenson, and ISI n te temporal dmenson..3 Smulaton Examples In ts secton, we compare te performance dfference between te temporal processng and space-tme processng by computer smulaton. Frst we consder temporal processng. Assume te number of users Q=3, te number of multpats L=4, te number of te equalzer taps M=4, and te sgnal-to-nose rato SNR=0dB. Te receved sgnal s gven by x xk xk xk k 3 = s, 0 s, k 0 M 4 s, k k 6 + s + 3s3 + n..7 Te equalzer s output s gven by s y k = w [ ] 3 s. s 3.8 In te deal case, te followng sould be satsfed: [ 3 ] = [ 0 L 0] x L+ M w..9 Te smulaton result s gven below. Te result s a x column vector. ss = w [ ] 3 5

31 ss = [ ] T. From te result above, we can see te frst element s muc larger tan oters. Te rato of -norms of te frst element and te rest s sown below : Ts result s far from te deal case :0, and t proves tat te tme-only processng cannot provde a satsfactory performance. Te Fgure.6 sows te QPSK constellaton of te desred user. Te modulated sgnal s presented by a+b. Fgure.7 sows te receved sgnal at te equalzer nput. Due to te cannel effects, multpat effects, nterference, and nose, te receved sgnals are corrupted and no longer old QPSK caracterstcs. Te recever must process te receved sgnal to elmnate te MAI, ISI and nose n order to recover te transmtted sgnal. Fgure.8 sows te sgnal constellaton at te equalzer output. Te sgnals ave smlar ampltudes to te orgnal transmtted sgnal, but cannot be effectvely recovered. Te performance of tme-only processng s poor. Next, we consder space-tme processng. A unform lnear antenna array wt fve elements and alf-wavelengt spacng s used. Te DOAs are [ ], and 0 s te DOA of te desred sgnal. Te smulaton result s gven below. Te result s a x column vector: 6

32 ss = w [ ] 3 ss = [ ] T. Te rato of -norms of te frst element and te rest s : Te result s very close to :0. Fgure.9 sows te sgnal constellaton after space-tme processng. It sows te recovered sgnals ave smlar ampltudes wt te orgnal transmtted sgnals, and are dstrbuted around eac center of te orgnal QPSK modulated sgnals. Te receved sgnals can be effectvely recovered. Te beam pattern for te above space-tme processng s llustrated n Fgure.0. As we can clearly observe, te antenna gan reaces te largest magntude at te DOA of te desred user,.e., [ 0 ], and as a null at te drectons of te nterferers,.e., [ ]. Terefore, MAI can be effectvely elmnated n te space-tme processng. In ts capter, we presented a descrpton of a wreless cannel and dscussed dfferent pyscal effects suc as pat lass, fadng, delay spread, angle spread, and Doppler spread tat make te receved sgnal muc weaker tan te transmtted sgnal. e also dscussed tree processng tecnques tat are used at te recever: tme-only, space-only and space-tme processng. As sad, tme-only and space-only processng tecnques ave fundamental drawbacks tat make te recever very dffcult to elmnate MAI and ISI smultaneously. In te contrast, space-tme processng ave overcome te drawbacks and 7

33 ave sgnfcant advantages over te oter two processng tecnques, and te recever s performance ave mproved dramatcally Fgure.6 QPSK modulated sgnal constellaton Fgure.7 Receved sgnal constellaton at te equalzer nput. 8

34 Fgure.8 Sgnal constellaton after tme-only processng Fgure.9 Sgnal constellaton after space-tme processng. 9

35 Antenna Gan DOA n Degree Fgure.0 Beam pattern for te space-tme processng. 30

36 Capter 3 Tme-varyng Multpat Vector Cannel Smulator In a wreless cannel, te cange of surroundng structures and te movement of moble users make te cannel tme-varyng. For te purpose of analyzng te performance of communcaton systems, tere s a need for effectvely smulatng te rado cannel. t te employment of an antenna array, te sgnals are receved at te recever by multple antennas, and terefore, te cannel model becomes a vector as opposed to a scalar were only a sngle antenna s employed. In ts capter, we ntroduce a tme-varyng multpat fast fadng vector cannel smulator tat can be used to evaluate te performance of antenna array recever. 3. Tme-varyng Multpat Vector Cannel In [8], a multpat cannel smulator was ntroduced wc s a mult-cannel generalzaton of te scalar cannel presented n [3]. Takng nto account an antenna array, an expanson s made to smulate te new cannels n ts capter. e consder te transmsson from a moble staton to a base staton. Te moble s surrounded by reflectng structures wtout lne of sgt between te moble and base staton and causes multpat effects. Te moble staton s movement also causes Doppler sft. Te faster te moble staton moves, te faster te cannel response canges. In order to present te cannel model, te followng parameters are defned: J number of antenna array elements f c carrer frequency ω c = πf c 3

37 B te transmtted sgnal bandwdt λ c carrer wavelengt v speed of te moble θ drecton of arrval of te receved sgnal; a plane wave s assumed d te dstance between moble staton and base staton τ = d / c, te propagaton delay were c s te velocty of lgt. τ mn and τ max are te mnmum and maxmum propagaton delay, respectvely. D array spacng L number of multpats Te antenna array used ere s a Unform Lnear Array ULA,.e., te spacng between te elements of a lnear array s equal, as llustrated n Fgure 3. Incdent plane wave θ Dsnθ #3 # D # Antenna Fgure 3. Unform Lnear Array wt tree elements. e assume: 3

38 . Te sgnals orgnate far away from te array and te plane wave assocated wt te sgnal advances troug a non-dspersve medum tat only ntroduces a propagaton delay. Under suc crcumstances, te sgnal at any oter element can be represented by a tme-delayed verson of te sgnal at te frst element.. D << c/b,.e., te bandwdt of te mpngng sgnal s muc smaller tan te recprocal of te propagaton tme across te array. Ts s commonly known as narrowband assumpton. Ts assumpton makes t possble to represent te propagaton delay wtn te elements of te array by pase sfts n te sgnal. t te assumptons mentoned above, te receved sgnal at te antenna array s gven by x t = J m= 0 s t a m θ 3. were st s te nformaton and a m θ s te array response. Te above equaton can be wrtten n a vector form as x t = s t a θ 3. were eac element of xt contans te receved sgnal at te correspondng array element and ωcρ θ ωcρ J θ [, e, L, e ] a θ = =, e D π sn θ λc, L, e T D π J sn θ λc T 3.3 were ρ s te propagaton delay relatve to te frst array element. 33

39 Te vector a θ s known as te array response vector or te steerng vector of a ULA, wc s a functon of DOA under suc crcumstances. It can be sown tat te baseband equvalent of J x vector cannel mpulse response s gven by [] L g t; τ = δ τ τ = t 3.4 were t s te observaton tme, τ s te propagaton delay of te t pat, L s te number of tme dfferentable pats wc are composed of a large number of tme nondfferental subpats wt te smlar DOA. Te DOA of eac tme dfferental pat s te mean of angles of all subpats. Te spread of angles of subpats s defned as angle spread. Smlarly, τ s te mean of te delays of te subpats. Te larger te τ, te less correlaton between tose pats. Te parameter s te complex pat vector for te t pat and s gven by [ ] T = L M were te elements of are te cannel coeffcents wc are functons of tme and frequency. If we assume all propagaton pats le n te same plane as te array elevaton angle ψ =0, te t complex pat vector s gven by ωdt ωcτ t = α θ, τ a θ e dθdτ 3.6 τ θ were τ ranges n τ + / B, τ + + / B], θ ranges [ θ /, θ + / ], θ s te [ mn mn mean angle of arrval, s te angle spread, α θ, τ s te ampltude ntensty functon for eac pat, and aθ s te array response. Te Doppler sft and pat delay result n pase varaton, ωdt ωcτ e, were Doppler sft s 34

40 f d = f c v cosψ c 3.7 wt ω = πf and we wll assume ψ=0. d d Usually, all tme dfferentable pats are composed of a large number of nondfferentable subpats. By te central lmt teorem, we fnd te cannel coeffcents to be well approxmated by complex Gaussan varables [6]. Snce Gaussan varables are entrely caracterzed by ter frst and second order statstcs, we can smulate te cannel smulator coeffcents by generatng Gaussan varables tat ave some predefned approprate frst and second order statstcs. Next, we ntroduce te frst and second caracterstcs of te cannel coeffcents. 3. Second Order Caracterzaton Snce te moble s surrounded by local reflectng structures, we ave many ndrect transmsson subpats eac of wc exbts dfferent propertes. t suc a scenaro, we assume:. Subpats correspondng to dfferent delays or angles of arrval ave uncorrelated ampltudes,.e., * ' ' ' ' E[ α θ, τ α θ, τ ] = δ θ θ, τ τ f θ, τ 3.8 were f θ, τ s te ont pdf of θ and τ, and * te complex conugate. Ts s also referred to as te wde sense statonary uncorrelated scatterng SSUS assumpton [0] [36].. For a gven subpat, te delay, angles of arrval are mutually ndependent,.e., 35

41 f θ, τ = f a θ f τ 3.9 b were f θ and f τ are te power densty functon of angles of arrval, θ, and a b transmsson delay τ, respectvely. 3. Te power densty functon of te transmsson delay τ s gven by [0] f b τ = e τ τ τ 3.0 were τ s te mean delay. Integratng f b τ wt respect to te delay doman for te t pat, we get τ F = τ mn mn + + / B + / B f τ dτ b 3. = [ e / Bτ ] e / Bτ were F s smply te power fracton assocated wt te t pat. Because of te random pase assocated wt eac ndvdual tme ndfferentable subpat n a gven tme dfferentable pat, te cannel coeffcents are well modeled by a zero mean complex varable [9]. Under te above assumptons, te cross-correlaton matrx of te complex pat vectors s gven by [] R t, t = E[ t d = δ J ω t F R 0 t ] a, 3. were, represent te t and t pat, δ = 0 wen wc mples te tme dfferentable pats are ndependent from eac oter, t = t t s te tme lag, and J 0. s te Bessel functon of te frst knd and of order 0. R a, s te spatal correlaton matrx for te t pat, and s gven by 36

42 = f θ a θ θ dθ 3.3 R a, a a θ If we assume te power densty functon wt respect to te angles of arrval, f θ, s unform n θ ± /, 3.3 can be wrtten as a θ / R a a +, = a θ θ dθ. θ / 3.4 Tus, te smaller te angle spread, te ger te spatal correlaton between te sgnals at te antenna elements. Te spatal correlaton s also a functon of antenna element spacng D. Te larger D, te lower te spatal correlaton. In macrocellular moble communcatons, snce te base s usually far from te moble and te local reflectors surroundng t, te angle spread nduced by te local reflectors s, terefore, often relatvely small [7], and te spatal correlaton can reman relatvely g even f te antenna elements are spaced by many λ c. 3.3 Vector Cannel Smulator Te vector cannel smulator structure s sown n Fgure 3.. It s sngle-nput multpleoutput SIMO dscrete-tme FIR system wt tme-varyng coeffcents, based on a TDL tapped delay lne model wt taps evenly spaced one sample apart. Te smulator as J brances correspondng to eac antenna array element. Eac branc as L wegts wt eac of wc represents a resolvable pat coeffcent. Te samplng nterval s T c.e., /B. Te power of te nose sgnal s cosen accordng to te requred SNR. Te nput to te smulator s a baseband transmtted sgnal, st, and t s multpled by te correspondng wegt,, wc s te cannel coeffcent correspondng to te t pat 37

43 of te t antenna element. AGN s added at te output and te receved sgnal at te t antenna element s y t. Fnally, we smulate te sgnals receved at te antenna travelng troug te tme-varyng multpat cannel. Next, we descrbe ow to generate te cannel coeffcents. st Tc Tc Tc 00 0 L-0 L-0 AGN y 0 t Tc Tc Tc 0 L- L- AGN y t Tc Tc Tc 0J- J- L-J- L-J- AGN y J- t Fgure 3. Tme-varyng multpat vector cannel smulator. 3.4 Complex Pat Vector Generator e ave known tat te tme dfferentable pats are ndependent from eac oter and te cannel coeffcents can be generated ndependently and approxmated by Gaussan varables. Te cannel coeffcents ave zero mean and second order caracterstcs 38

44 dscussed above, ncludng spatal-temporal correlaton propertes. Terefore, we can apply some knd of space-tme correlaton sapng transformaton to uncorrelated Gaussan wte nose sequences n order to obtan tme-varyng pat vectors tat exbt te approprate spatal-temporal correlaton propertes. Te tme-varyng cannel s mostly due to moble staton moton. Te extent of ow te cannel vares n tme s measured by te Doppler sft. As long as te complex pat vectors are obtaned at a rate ger tan twce te Doppler frequency, ter temporal correlaton structure s preserved. Fgure 3.3 sows te structure of te t pat vector generator. Zero-mean Gaussan Varable Generator _ n 0 m _ n m z z y 0 m y m Space correlaton transformaton 0 m m T/T c T/T c 0 k k _ n J- m z y J- m J- m T/T c J- k Tme-correlaton transformaton flter Interpolators Fgure.3 Te complex pat vector generator for te t pat 39

45 e take te samplng nterval of T were = 3 f, ten feed te samples nto te d tmecorrelaton transformer and space-correlaton transformer, and fnally te smulated T complex pat vectors are nterpolated n tme to te desred samplng rate, /T c = B. Equaton 3. can be rewrtten n te followng way for te case =: R t = J 0 ω t F R a, 3.5 d were F s ndependent of t, and terefore, we defne C = FR a,. 3.6 Ts suggests tat n order to obtan te approprate space-tme correlaton caracterstcs for a gven pat, one can use a tme transformaton followed by a spatal transformaton. Te tme ndex n Fgure 3.3 refers to te samplng nterval T of te complex pat vectors pror to te nterpolator. Te nose generator produces zero-mean complex Gaussan vectors: T n m = [ n0 m n m L n M m ]. 3.7 Te tme-correlaton sapng flter denoted by z s desgned so tat te temporal correlaton of ts output, y m = z n m, s approxmately equal to temporal component n 3.5,.e., J ω. Te space-correlaton transformaton takes care of 0 d t te spatal component of te correlaton,.e., C. Ts transformaton s appled to te vector y m. Fnally te smulated complex pat vectors are nterpolated n tme to te desred samplng rate. 40

46 3.4. Tme-correlaton Sapng Flter A tme correlaton sapng flter for obtanng te desred temporal correlaton s wdely used. Te desred power spectral densty functon s te Fourer transform of te Bessel functon n 3.5, and s gven by [8] s ω = f ω ω d oterwse ω d ω ω For te tme correlaton sapng flter, we must ave e = s ω. Snce s ω as sngulartes at ω = ±ω d, te desgn of ts flter s unrealzable. In practce, te sngulartes at ω = ±ω d are replaced by sarp peak. If FIR s selected to mplement ts flter, te order of te flter must be g to obtan te sarp frequency response. To reduce te computatonal complexty, a low order IIR flter s used. ω e desgn an analog flter tat satsfes e = s ω, and ten apply AD converson. Te samplng rate s T 3 = x f d = 3 f d 3.9.e., te samplng rate s 3/ tmes te Nyqust frequency. Fnally, we get a 4t-order IIR flter [] z + 05z +.53z z z = z +.334z +.343z z 3.0 e wll smulate te result later by comparng te correlaton of te cannel coeffcents after te flter wt Bessel functon. 4

47 3.4. Spatal Transformaton Te spatal transformaton used to obtan te complex pat vector takes te form of a smple matrx operaton to y : = M y 3. Te matrx M must satsfy M M = C,.e., te correlaton matrx of te spatal transformaton matrx must be equal to C. Te development of M s based on Karunen-Loeve expanson for vectoral random process. Snce te C s a correlaton matrx of a dscrete tme random process, t s nonnegatve defnte ermtan matrx [6]. A ermtan matrx s dagonalzable, so we can wrte C = Q Λ Q 3. were Q s te matrx wose columns are ortonormalzed egenvectors of C, denoted by q = 0,..., J- and Λ s te dagonal matrx wose entres correspond egenvalues, λ. From q =, we get q q. Te fact tat C s nonnegatve defnte mples q = tat q Cq 0. By defnton Cq = λq, so we ave q Cq = λ q = λ 0, so te egenvalues of C are real and nonnegatve. e can terefore get C = Q Λ Λ Q 3.3 = M M = [ Q Λ / / / ][ Λ / Q ] were M = Q Λ s te desred spatal transformaton matrx. / Premultplcaton of te spatally uncorrelated sgnal vector y by M = Q Λ gves te / complex pat vector suc tat 4

48 E[ m k] = Q Λ [ Q Λ / d d / E[ y m y J ω T k m C d J ω T k m Q Λ Q 0 0 k] Λ / ] J ω T k m[ Λ 0 Q / Q ] 3.4 wc s te desred result. Fgure 3.4 llustrates te space-correlaton transformaton for te t pat. Eac element of y s frst scaled by te square root of te egenvalue. Te resultant vector s ten multpled by te egenvector matrx Q. Te output of te pat vector generator as te approprate tme-space correlaton propertes. y 0 m 0 m λ 0 y m λ Premultplcaton by Q m y J- m λ J J- m Fgure 3.4 Space-correlaton transformaton Interpolator As mentoned before, te samplng rate to sample te Gaussan varable generator s /T=3f d, and te samples are appled to te tme-space correlaton transformaton. Te frequency assocated wt te pat vector generator s muc smaller tan te one assocated wt cannel flterng /T c = B. e terefore need to obtan te g rate 43

49 cannel coeffcents requred for cannel flterng va nterpolaton. en a sgnal wt a samplng frequency muc ger tan te Nyqust frequency as to be nterpolated, smple metods suc as lnear or cubc nterpolaton can be used wtout compromsng te precson [33]. Snce /T c = B s muc larger tan /T, t s possble to decompose te nterpolaton process to reduce te computatonal requrements, as llustrated n Fgure 3.5. Te system s composed of a lnear nterpolator followed by a cubc nterpolator. Te lnear nterpolator ncreases by a factor I 30 te samplng rate to become muc ger tan te Nyqust frequency, and ten a cubc nterpolator s used to furter ncrease te samplng frequency to ts desred value of /T c. m Samplng perod T Lnear nterpolator Samplng perod T/I Cubc nterpolator k Samplng perod Tc Fgure 3.5 Interpolator. Fgure 3.6 llustrates te lnear nterpolaton: Fgure 3.6A represents te orgnal sgnals, Fgure 3.6B te sgnals after addng I- zeros between samples zero-paddng, and ten te sgnals are fed nto a low-pass flter wt a radal cutoff frequency π / I. Fgure 3.6C represents te sgnals after lnear nterpolaton. Lnear nterpolaton s computatonally complex, but as more precson. en lnear nterpolaton s completed, a cubc nterpolator s used. 44

50 6 A Ampltude B Ampltude Ampltude C Tmesec Fgure 3.6 Lnear nterpolaton llustraton. Fgure 3.7 llustrates te cubc splne nterpolaton, between te sample nterval [ T T ], were te sgnal s represented by a trd-order polynomal: p t + t 3 = a0 + at + at a Terefore, samples can be obtaned by samplng te above functon at te requred sample tme nstant so tat te rate s ncreased. 45

51 xt T- T T+ t Fgure 3.7 Cubc splne nterpolaton llustraton. 3.5 Smulaton Examples e smulate and examne te cannel realzaton va te pat vector generator. Consder a DS/CDMA system wt carrer frequency beng f c = Gz, transmtted sgnal bandwdt s B =.88Mz Tc = / B, te number of antenna array elements s M=3, te mean delay s τ = / B = T, te speed of te moble s ν =30m/s, te Doppler c frequency s f d = f c. ν = 00z, and terefore, te samplng nterval before te c nterpolator s T = = 4096Tc. Te lnear nterpolator decreases te nterval to 8Tc 3 f d I=3, and ten te cubc nterpolaton reduces t from 8T c to T c. By passng a wte Gaussan nose sequence troug te IIR flter z developed prevously, we get fltered outputs wose temporal correlaton s approxmately equal to te temporal component n 3.5, as llustrated n Fgure 3.8 wc plots te desred and obtaned temporal correlaton. Te smaller te tme lag, te more tey are correlated. 46

52 obtaned desred Temporal correlaton Normalzed tme lag Fgure 3.8 Desred and obtaned temporal correlaton. After passng troug te tme correlaton sapng flter, te smulated pat vector needs to be spatally transformed to get an approprate spatal correlaton. Frst of all, assume tere are tree tme dfferentable pats, L = 3, wose DOAs are [ 60,0,50 ], respectvely. Te angle spreads are [0,,5 ], respectvely. Fgure 3.9 llustrates te plot of te magntude of te cannel coeffcents at te frst antenna element. Snce tere s g correlaton of te cannel coeffcents between antenna elements, te cannel coeffcents are very smlar for dfferent antennas, and te plots of te cannel coeffcents versus tme are nearly te same. Te tme nterval of te plot along x-axs s /B, correspondng to a cp n DS-CDMA. As llustrated, te cannel coeffcents are rapdly tme-varyng. 47

53 Fgure 3.9 Cannel coeffcent magntudes vs. tme for te frst antenna. Te nterpolator s used to nterpolate te cannel coeffcents to ncrease te fnal sample rate to te transmtted sgnal bandwdt B. Fgure 3.0 llustrates te plot of te cannel coeffcent magntudes vs. tme for te frst pat of te frst antenna element before and after te nterpolator. As llustrated, te curve s not smoot before te nterpolator, and as sarp corners. After te nterpolator, te curve s very smoot, and very smlar to tat of a typcal wreless cannel [0]. 48

54 Cannel coeffcent magntudes before/after nterpolaton for Pat of te st antenna element after nterpolator before nterpolator 0-5 Gan db tme sec Fgure 3.0 Cannel coeffcent magntudes before and after nterpolaton for Pat of te frst antenna element. In ts capter, a new multpat cannel smulator s ntroduced. Te smulaton results agree wt te dscusson above, and later, we wll use ts tme-varyng multpat vector cannel smulator to generate te desred cannel coeffcents to analyze te recever s performance. 49

Stanford University CS254: Computational Complexity Notes 7 Luca Trevisan January 29, Notes for Lecture 7

Stanford University CS254: Computational Complexity Notes 7 Luca Trevisan January 29, Notes for Lecture 7 Stanford Unversty CS54: Computatonal Complexty Notes 7 Luca Trevsan January 9, 014 Notes for Lecture 7 1 Approxmate Countng wt an N oracle We complete te proof of te followng result: Teorem 1 For every

More information

TLCOM 612 Advanced Telecommunications Engineering II

TLCOM 612 Advanced Telecommunications Engineering II TLCOM 62 Advanced Telecommuncatons Engneerng II Wnter 2 Outlne Presentatons The moble rado sgnal envronment Combned fadng effects and nose Delay spread and Coherence bandwdth Doppler Shft Fast vs. Slow

More information

Multivariate Ratio Estimator of the Population Total under Stratified Random Sampling

Multivariate Ratio Estimator of the Population Total under Stratified Random Sampling Open Journal of Statstcs, 0,, 300-304 ttp://dx.do.org/0.436/ojs.0.3036 Publsed Onlne July 0 (ttp://www.scrp.org/journal/ojs) Multvarate Rato Estmator of te Populaton Total under Stratfed Random Samplng

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:

More information

Throughput Capacities and Optimal Resource Allocation in Multiaccess Fading Channels

Throughput Capacities and Optimal Resource Allocation in Multiaccess Fading Channels Trougput Capactes and Optmal esource Allocaton n ultaccess Fadng Cannels Hao Zou arc 7, 003 Unversty of Notre Dame Abstract oble wreless envronment would ntroduce specal penomena suc as multpat fadng wc

More information

Consider the following passband digital communication system model. c t. modulator. t r a n s m i t t e r. signal decoder.

Consider the following passband digital communication system model. c t. modulator. t r a n s m i t t e r. signal decoder. PASSBAND DIGITAL MODULATION TECHNIQUES Consder the followng passband dgtal communcaton system model. cos( ω + φ ) c t message source m sgnal encoder s modulator s () t communcaton xt () channel t r a n

More information

Introduction to Antennas & Arrays

Introduction to Antennas & Arrays Introducton to Antennas & Arrays Antenna transton regon (structure) between guded eaves (.e. coaxal cable) and free space waves. On transmsson, antenna accepts energy from TL and radates t nto space. J.D.

More information

CHAPTER 4 SPEECH ENHANCEMENT USING MULTI-BAND WIENER FILTER. In real environmental conditions the speech signal may be

CHAPTER 4 SPEECH ENHANCEMENT USING MULTI-BAND WIENER FILTER. In real environmental conditions the speech signal may be 55 CHAPTER 4 SPEECH ENHANCEMENT USING MULTI-BAND WIENER FILTER 4.1 Introducton In real envronmental condtons the speech sgnal may be supermposed by the envronmental nterference. In general, the spectrum

More information

ECE559VV Project Report

ECE559VV Project Report ECE559VV Project Report (Supplementary Notes Loc Xuan Bu I. MAX SUM-RATE SCHEDULING: THE UPLINK CASE We have seen (n the presentaton that, for downlnk (broadcast channels, the strategy maxmzng the sum-rate

More information

The Finite Element Method: A Short Introduction

The Finite Element Method: A Short Introduction Te Fnte Element Metod: A Sort ntroducton Wat s FEM? Te Fnte Element Metod (FEM) ntroduced by engneers n late 50 s and 60 s s a numercal tecnque for solvng problems wc are descrbed by Ordnary Dfferental

More information

Composite Hypotheses testing

Composite Hypotheses testing Composte ypotheses testng In many hypothess testng problems there are many possble dstrbutons that can occur under each of the hypotheses. The output of the source s a set of parameters (ponts n a parameter

More information

Digital Modems. Lecture 2

Digital Modems. Lecture 2 Dgtal Modems Lecture Revew We have shown that both Bayes and eyman/pearson crtera are based on the Lkelhood Rato Test (LRT) Λ ( r ) < > η Λ r s called observaton transformaton or suffcent statstc The crtera

More information

Chapter 6. Wideband channels. Slides for Wireless Communications Edfors, Molisch, Tufvesson

Chapter 6. Wideband channels. Slides for Wireless Communications Edfors, Molisch, Tufvesson Chapter 6 Wdeband channels 128 Delay (tme) dsperson A smple case Transmtted mpulse h h a a a 1 1 2 2 3 3 Receved sgnal (channel mpulse response) 1 a 1 2 a 2 a 3 3 129 Delay (tme) dsperson One reflecton/path,

More information

Rethinking MIMO for Wireless Networks: Linear Throughput Increases with Multiple Receive Antennas

Rethinking MIMO for Wireless Networks: Linear Throughput Increases with Multiple Receive Antennas Retnng MIMO for Wreless etwors: Lnear Trougput Increases wt Multple Receve Antennas ar Jndal Unversty of Mnnesota Unverstat Pompeu Fabra Jont wor wt Jeff Andrews & Steven Weber MIMO n Pont-to-Pont Cannels

More information

The Concept of Beamforming

The Concept of Beamforming ELG513 Smart Antennas S.Loyka he Concept of Beamformng Generc representaton of the array output sgnal, 1 where w y N 1 * = 1 = w x = w x (4.1) complex weghts, control the array pattern; y and x - narrowband

More information

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system Transfer Functons Convenent representaton of a lnear, dynamc model. A transfer functon (TF) relates one nput and one output: x t X s y t system Y s The followng termnology s used: x y nput output forcng

More information

VQ widely used in coding speech, image, and video

VQ widely used in coding speech, image, and video at Scalar quantzers are specal cases of vector quantzers (VQ): they are constraned to look at one sample at a tme (memoryless) VQ does not have such constrant better RD perfomance expected Source codng

More information

Lossy Compression. Compromise accuracy of reconstruction for increased compression.

Lossy Compression. Compromise accuracy of reconstruction for increased compression. Lossy Compresson Compromse accuracy of reconstructon for ncreased compresson. The reconstructon s usually vsbly ndstngushable from the orgnal mage. Typcally, one can get up to 0:1 compresson wth almost

More information

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could

More information

Chapter 11: Simple Linear Regression and Correlation

Chapter 11: Simple Linear Regression and Correlation Chapter 11: Smple Lnear Regresson and Correlaton 11-1 Emprcal Models 11-2 Smple Lnear Regresson 11-3 Propertes of the Least Squares Estmators 11-4 Hypothess Test n Smple Lnear Regresson 11-4.1 Use of t-tests

More information

Digital Signal Processing

Digital Signal Processing Dgtal Sgnal Processng Dscrete-tme System Analyss Manar Mohasen Offce: F8 Emal: manar.subh@ut.ac.r School of IT Engneerng Revew of Precedent Class Contnuous Sgnal The value of the sgnal s avalable over

More information

A Lower Bound on SINR Threshold for Call Admission Control in Multiple-Class CDMA Systems with Imperfect Power-Control

A Lower Bound on SINR Threshold for Call Admission Control in Multiple-Class CDMA Systems with Imperfect Power-Control A ower Bound on SIR Threshold for Call Admsson Control n Multple-Class CDMA Systems w Imperfect ower-control Mohamed H. Ahmed Faculty of Engneerng and Appled Scence Memoral Unversty of ewfoundland St.

More information

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4) I. Classcal Assumptons Econ7 Appled Econometrcs Topc 3: Classcal Model (Studenmund, Chapter 4) We have defned OLS and studed some algebrac propertes of OLS. In ths topc we wll study statstcal propertes

More information

Pulse Coded Modulation

Pulse Coded Modulation Pulse Coded Modulaton PCM (Pulse Coded Modulaton) s a voce codng technque defned by the ITU-T G.711 standard and t s used n dgtal telephony to encode the voce sgnal. The frst step n the analog to dgtal

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

Not-for-Publication Appendix to Optimal Asymptotic Least Aquares Estimation in a Singular Set-up

Not-for-Publication Appendix to Optimal Asymptotic Least Aquares Estimation in a Singular Set-up Not-for-Publcaton Aendx to Otmal Asymtotc Least Aquares Estmaton n a Sngular Set-u Antono Dez de los Ros Bank of Canada dezbankofcanada.ca December 214 A Proof of Proostons A.1 Proof of Prooston 1 Ts roof

More information

Numerical Simulation of One-Dimensional Wave Equation by Non-Polynomial Quintic Spline

Numerical Simulation of One-Dimensional Wave Equation by Non-Polynomial Quintic Spline IOSR Journal of Matematcs (IOSR-JM) e-issn: 78-578, p-issn: 319-765X. Volume 14, Issue 6 Ver. I (Nov - Dec 018), PP 6-30 www.osrournals.org Numercal Smulaton of One-Dmensonal Wave Equaton by Non-Polynomal

More information

Multigrid Methods and Applications in CFD

Multigrid Methods and Applications in CFD Multgrd Metods and Applcatons n CFD Mcael Wurst 0 May 009 Contents Introducton Typcal desgn of CFD solvers 3 Basc metods and ter propertes for solvng lnear systems of equatons 4 Geometrc Multgrd 3 5 Algebrac

More information

Chapter 13: Multiple Regression

Chapter 13: Multiple Regression Chapter 13: Multple Regresson 13.1 Developng the multple-regresson Model The general model can be descrbed as: It smplfes for two ndependent varables: The sample ft parameter b 0, b 1, and b are used to

More information

Multigradient for Neural Networks for Equalizers 1

Multigradient for Neural Networks for Equalizers 1 Multgradent for Neural Netorks for Equalzers 1 Chulhee ee, Jnook Go and Heeyoung Km Department of Electrcal and Electronc Engneerng Yonse Unversty 134 Shnchon-Dong, Seodaemun-Ku, Seoul 1-749, Korea ABSTRACT

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 30 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 2 Remedes for multcollnearty Varous technques have

More information

TR/95 February Splines G. H. BEHFOROOZ* & N. PAPAMICHAEL

TR/95 February Splines G. H. BEHFOROOZ* & N. PAPAMICHAEL TR/9 February 980 End Condtons for Interpolatory Quntc Splnes by G. H. BEHFOROOZ* & N. PAPAMICHAEL *Present address: Dept of Matematcs Unversty of Tabrz Tabrz Iran. W9609 A B S T R A C T Accurate end condtons

More information

Average SIR of the desired user ([7]) β=0.99. β= Average NSE of the desired user (BADD) β=0.95. µ= β=0.9

Average SIR of the desired user ([7]) β=0.99. β= Average NSE of the desired user (BADD) β=0.95. µ= β=0.9 A Blnd Adaptve Decorrelatng Detector for CDMA Systems Sennur Ulukus Roy D. Yates Department of lectrcal and Computer ngneerng Wreless Informaton Networks Laboratory (WINLAB) Rutgers Unversty, PO Box 909,

More information

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests Smulated of the Cramér-von Mses Goodness-of-Ft Tests Steele, M., Chaselng, J. and 3 Hurst, C. School of Mathematcal and Physcal Scences, James Cook Unversty, Australan School of Envronmental Studes, Grffth

More information

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2 Salmon: Lectures on partal dfferental equatons 5. Classfcaton of second-order equatons There are general methods for classfyng hgher-order partal dfferental equatons. One s very general (applyng even to

More information

Problem Set 4: Sketch of Solutions

Problem Set 4: Sketch of Solutions Problem Set 4: Sketc of Solutons Informaton Economcs (Ec 55) George Georgads Due n class or by e-mal to quel@bu.edu at :30, Monday, December 8 Problem. Screenng A monopolst can produce a good n dfferent

More information

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U)

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U) Econ 413 Exam 13 H ANSWERS Settet er nndelt 9 deloppgaver, A,B,C, som alle anbefales å telle lkt for å gøre det ltt lettere å stå. Svar er gtt . Unfortunately, there s a prntng error n the hnt of

More information

Solution for singularly perturbed problems via cubic spline in tension

Solution for singularly perturbed problems via cubic spline in tension ISSN 76-769 England UK Journal of Informaton and Computng Scence Vol. No. 06 pp.6-69 Soluton for sngularly perturbed problems va cubc splne n tenson K. Aruna A. S. V. Rav Kant Flud Dynamcs Dvson Scool

More information

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng

More information

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons

More information

ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM

ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM An elastc wave s a deformaton of the body that travels throughout the body n all drectons. We can examne the deformaton over a perod of tme by fxng our look

More information

A SINR Improvement Algorithm for D2D Communication Underlaying Cellular Networks

A SINR Improvement Algorithm for D2D Communication Underlaying Cellular Networks Advanced Scence and Tecnology Letters Vol.3 (CST 06), pp.78-83 ttp://dx.do.org/0.457/astl.06.3.34 A SINR Improvement Algortm for DD Communcaton Underlayng Cellular Networks Ceng uan, Youua Fu,, Jn Wang

More information

PubH 7405: REGRESSION ANALYSIS. SLR: INFERENCES, Part II

PubH 7405: REGRESSION ANALYSIS. SLR: INFERENCES, Part II PubH 7405: REGRESSION ANALSIS SLR: INFERENCES, Part II We cover te topc of nference n two sessons; te frst sesson focused on nferences concernng te slope and te ntercept; ts s a contnuaton on estmatng

More information

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number

More information

Lecture Notes on Linear Regression

Lecture Notes on Linear Regression Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume

More information

On Pfaff s solution of the Pfaff problem

On Pfaff s solution of the Pfaff problem Zur Pfaff scen Lösung des Pfaff scen Probles Mat. Ann. 7 (880) 53-530. On Pfaff s soluton of te Pfaff proble By A. MAYER n Lepzg Translated by D. H. Delpenc Te way tat Pfaff adopted for te ntegraton of

More information

Tracking with Kalman Filter

Tracking with Kalman Filter Trackng wth Kalman Flter Scott T. Acton Vrgna Image and Vdeo Analyss (VIVA), Charles L. Brown Department of Electrcal and Computer Engneerng Department of Bomedcal Engneerng Unversty of Vrgna, Charlottesvlle,

More information

Global Sensitivity. Tuesday 20 th February, 2018

Global Sensitivity. Tuesday 20 th February, 2018 Global Senstvty Tuesday 2 th February, 28 ) Local Senstvty Most senstvty analyses [] are based on local estmates of senstvty, typcally by expandng the response n a Taylor seres about some specfc values

More information

Structure and Drive Paul A. Jensen Copyright July 20, 2003

Structure and Drive Paul A. Jensen Copyright July 20, 2003 Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.

More information

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS Avalable onlne at http://sck.org J. Math. Comput. Sc. 3 (3), No., 6-3 ISSN: 97-537 COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

More information

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton

More information

Uncertainty as the Overlap of Alternate Conditional Distributions

Uncertainty as the Overlap of Alternate Conditional Distributions Uncertanty as the Overlap of Alternate Condtonal Dstrbutons Olena Babak and Clayton V. Deutsch Centre for Computatonal Geostatstcs Department of Cvl & Envronmental Engneerng Unversty of Alberta An mportant

More information

Numerical Heat and Mass Transfer

Numerical Heat and Mass Transfer Master degree n Mechancal Engneerng Numercal Heat and Mass Transfer 06-Fnte-Dfference Method (One-dmensonal, steady state heat conducton) Fausto Arpno f.arpno@uncas.t Introducton Why we use models and

More information

NUMERICAL DIFFERENTIATION

NUMERICAL DIFFERENTIATION NUMERICAL DIFFERENTIATION 1 Introducton Dfferentaton s a method to compute the rate at whch a dependent output y changes wth respect to the change n the ndependent nput x. Ths rate of change s called the

More information

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan Wnter 2008 CS567 Stochastc Lnear/Integer Programmng Guest Lecturer: Xu, Huan Class 2: More Modelng Examples 1 Capacty Expanson Capacty expanson models optmal choces of the tmng and levels of nvestments

More information

Thermal-Fluids I. Chapter 18 Transient heat conduction. Dr. Primal Fernando Ph: (850)

Thermal-Fluids I. Chapter 18 Transient heat conduction. Dr. Primal Fernando Ph: (850) hermal-fluds I Chapter 18 ransent heat conducton Dr. Prmal Fernando prmal@eng.fsu.edu Ph: (850) 410-6323 1 ransent heat conducton In general, he temperature of a body vares wth tme as well as poston. In

More information

Infinite Length MMSE Decision Feedback Equalization

Infinite Length MMSE Decision Feedback Equalization Infnte Lengt SE ecson Feedbac Equalzaton FE N * * Y F Z ' Z SS ˆ Y Q N b... Infnte-Lengt ecson Feedbac Equalzer as reoval ^ Y Z Feedforward Flter Feedbac Flter - Input to Slcer - Z Y Assung prevous decsons

More information

Chapter 7 Channel Capacity and Coding

Chapter 7 Channel Capacity and Coding Wreless Informaton Transmsson System Lab. Chapter 7 Channel Capacty and Codng Insttute of Communcatons Engneerng atonal Sun Yat-sen Unversty Contents 7. Channel models and channel capacty 7.. Channel models

More information

An Upper Bound on SINR Threshold for Call Admission Control in Multiple-Class CDMA Systems with Imperfect Power-Control

An Upper Bound on SINR Threshold for Call Admission Control in Multiple-Class CDMA Systems with Imperfect Power-Control An Upper Bound on SINR Threshold for Call Admsson Control n Multple-Class CDMA Systems wth Imperfect ower-control Mahmoud El-Sayes MacDonald, Dettwler and Assocates td. (MDA) Toronto, Canada melsayes@hotmal.com

More information

Assessment of Site Amplification Effect from Input Energy Spectra of Strong Ground Motion

Assessment of Site Amplification Effect from Input Energy Spectra of Strong Ground Motion Assessment of Ste Amplfcaton Effect from Input Energy Spectra of Strong Ground Moton M.S. Gong & L.L Xe Key Laboratory of Earthquake Engneerng and Engneerng Vbraton,Insttute of Engneerng Mechancs, CEA,

More information

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers Psychology 282 Lecture #24 Outlne Regresson Dagnostcs: Outlers In an earler lecture we studed the statstcal assumptons underlyng the regresson model, ncludng the followng ponts: Formal statement of assumptons.

More information

Comparison of the Population Variance Estimators. of 2-Parameter Exponential Distribution Based on. Multiple Criteria Decision Making Method

Comparison of the Population Variance Estimators. of 2-Parameter Exponential Distribution Based on. Multiple Criteria Decision Making Method Appled Mathematcal Scences, Vol. 7, 0, no. 47, 07-0 HIARI Ltd, www.m-hkar.com Comparson of the Populaton Varance Estmators of -Parameter Exponental Dstrbuton Based on Multple Crtera Decson Makng Method

More information

FAST CONVERGENCE ADAPTIVE MMSE RECEIVER FOR ASYNCHRONOUS DS-CDMA SYSTEMS

FAST CONVERGENCE ADAPTIVE MMSE RECEIVER FOR ASYNCHRONOUS DS-CDMA SYSTEMS Électronque et transmsson de l nformaton FAST CONVERGENCE ADAPTIVE MMSE RECEIVER FOR ASYNCHRONOUS DS-CDMA SYSTEMS CĂLIN VLĂDEANU, CONSTANTIN PALEOLOGU 1 Key words: DS-CDMA, MMSE adaptve recever, Least

More information

Feb 14: Spatial analysis of data fields

Feb 14: Spatial analysis of data fields Feb 4: Spatal analyss of data felds Mappng rregularly sampled data onto a regular grd Many analyss technques for geophyscal data requre the data be located at regular ntervals n space and/or tme. hs s

More information

1. Inference on Regression Parameters a. Finding Mean, s.d and covariance amongst estimates. 2. Confidence Intervals and Working Hotelling Bands

1. Inference on Regression Parameters a. Finding Mean, s.d and covariance amongst estimates. 2. Confidence Intervals and Working Hotelling Bands Content. Inference on Regresson Parameters a. Fndng Mean, s.d and covarance amongst estmates.. Confdence Intervals and Workng Hotellng Bands 3. Cochran s Theorem 4. General Lnear Testng 5. Measures of

More information

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography CSc 6974 and ECSE 6966 Math. Tech. for Vson, Graphcs and Robotcs Lecture 21, Aprl 17, 2006 Estmatng A Plane Homography Overvew We contnue wth a dscusson of the major ssues, usng estmaton of plane projectve

More information

Some Comments on Accelerating Convergence of Iterative Sequences Using Direct Inversion of the Iterative Subspace (DIIS)

Some Comments on Accelerating Convergence of Iterative Sequences Using Direct Inversion of the Iterative Subspace (DIIS) Some Comments on Acceleratng Convergence of Iteratve Sequences Usng Drect Inverson of the Iteratve Subspace (DIIS) C. Davd Sherrll School of Chemstry and Bochemstry Georga Insttute of Technology May 1998

More information

Research Article Green s Theorem for Sign Data

Research Article Green s Theorem for Sign Data Internatonal Scholarly Research Network ISRN Appled Mathematcs Volume 2012, Artcle ID 539359, 10 pages do:10.5402/2012/539359 Research Artcle Green s Theorem for Sgn Data Lous M. Houston The Unversty of

More information

A Robust Method for Calculating the Correlation Coefficient

A Robust Method for Calculating the Correlation Coefficient A Robust Method for Calculatng the Correlaton Coeffcent E.B. Nven and C. V. Deutsch Relatonshps between prmary and secondary data are frequently quantfed usng the correlaton coeffcent; however, the tradtonal

More information

Suppose that there s a measured wndow of data fff k () ; :::; ff k g of a sze w, measured dscretely wth varable dscretzaton step. It s convenent to pl

Suppose that there s a measured wndow of data fff k () ; :::; ff k g of a sze w, measured dscretely wth varable dscretzaton step. It s convenent to pl RECURSIVE SPLINE INTERPOLATION METHOD FOR REAL TIME ENGINE CONTROL APPLICATIONS A. Stotsky Volvo Car Corporaton Engne Desgn and Development Dept. 97542, HA1N, SE- 405 31 Gothenburg Sweden. Emal: astotsky@volvocars.com

More information

5 The Laplace Equation in a convex polygon

5 The Laplace Equation in a convex polygon 5 Te Laplace Equaton n a convex polygon Te most mportant ellptc PDEs are te Laplace, te modfed Helmoltz and te Helmoltz equatons. Te Laplace equaton s u xx + u yy =. (5.) Te real and magnary parts of an

More information

The Order Relation and Trace Inequalities for. Hermitian Operators

The Order Relation and Trace Inequalities for. Hermitian Operators Internatonal Mathematcal Forum, Vol 3, 08, no, 507-57 HIKARI Ltd, wwwm-hkarcom https://doorg/0988/mf088055 The Order Relaton and Trace Inequaltes for Hermtan Operators Y Huang School of Informaton Scence

More information

Energy-Aware Fault Tolerance in Fixed-Priority Real-Time Embedded Systems*

Energy-Aware Fault Tolerance in Fixed-Priority Real-Time Embedded Systems* Energy-Aware Fault Tolerance n Fxed-Prorty Real-Tme Embedded Systems Yng Zang, Krsnendu Cakrabarty and Vsnu Swamnatan Department of Electrcal & Computer Engneerng Duke Unversty, Duram, NC 778, USA Abstract

More information

COMP4630: λ-calculus

COMP4630: λ-calculus COMP4630: λ-calculus 4. Standardsaton Mcael Norrs Mcael.Norrs@ncta.com.au Canberra Researc Lab., NICTA Semester 2, 2015 Last Tme Confluence Te property tat dvergent evaluatons can rejon one anoter Proof

More information

Error Probability for M Signals

Error Probability for M Signals Chapter 3 rror Probablty for M Sgnals In ths chapter we dscuss the error probablty n decdng whch of M sgnals was transmtted over an arbtrary channel. We assume the sgnals are represented by a set of orthonormal

More information

Research on Modified Root-MUSIC Algorithm of DOA Estimation Based on Covariance Matrix Reconstruction

Research on Modified Root-MUSIC Algorithm of DOA Estimation Based on Covariance Matrix Reconstruction Sensors & ransducers, Vol. 78, Issue 9, September 04, pp. 4-8 Sensors & ransducers 04 by IFSA Publshng, S. L. http://www.sensorsportal.com Research on Modfed Root-MUSIC Algorthm of DOA Estmaton Based on

More information

Chapter 7 Channel Capacity and Coding

Chapter 7 Channel Capacity and Coding Chapter 7 Channel Capacty and Codng Contents 7. Channel models and channel capacty 7.. Channel models Bnary symmetrc channel Dscrete memoryless channels Dscrete-nput, contnuous-output channel Waveform

More information

Report on Image warping

Report on Image warping Report on Image warpng Xuan Ne, Dec. 20, 2004 Ths document summarzed the algorthms of our mage warpng soluton for further study, and there s a detaled descrpton about the mplementaton of these algorthms.

More information

Appendix B: Resampling Algorithms

Appendix B: Resampling Algorithms 407 Appendx B: Resamplng Algorthms A common problem of all partcle flters s the degeneracy of weghts, whch conssts of the unbounded ncrease of the varance of the mportance weghts ω [ ] of the partcles

More information

, rst we solve te PDE's L ad L ad n g g (x) = ; = ; ; ; n () (x) = () Ten, we nd te uncton (x), te lnearzng eedbac and coordnates transormaton are gve

, rst we solve te PDE's L ad L ad n g g (x) = ; = ; ; ; n () (x) = () Ten, we nd te uncton (x), te lnearzng eedbac and coordnates transormaton are gve Freedom n Coordnates Transormaton or Exact Lnearzaton and ts Applcaton to Transent Beavor Improvement Kenj Fujmoto and Tosaru Suge Dvson o Appled Systems Scence, Kyoto Unversty, Uj, Kyoto, Japan suge@robotuassyoto-uacjp

More information

Signal space Review on vector space Linear independence Metric space and norm Inner product

Signal space Review on vector space Linear independence Metric space and norm Inner product Sgnal space.... Revew on vector space.... Lnear ndependence... 3.3 Metrc space and norm... 4.4 Inner product... 5.5 Orthonormal bass... 7.6 Waveform communcaton system... 9.7 Some examples... 6 Sgnal space

More information

Prof. Dr. I. Nasser Phys 630, T Aug-15 One_dimensional_Ising_Model

Prof. Dr. I. Nasser Phys 630, T Aug-15 One_dimensional_Ising_Model EXACT OE-DIMESIOAL ISIG MODEL The one-dmensonal Isng model conssts of a chan of spns, each spn nteractng only wth ts two nearest neghbors. The smple Isng problem n one dmenson can be solved drectly n several

More information

APPENDIX A Some Linear Algebra

APPENDIX A Some Linear Algebra APPENDIX A Some Lnear Algebra The collecton of m, n matrces A.1 Matrces a 1,1,..., a 1,n A = a m,1,..., a m,n wth real elements a,j s denoted by R m,n. If n = 1 then A s called a column vector. Smlarly,

More information

Chapter 3 Describing Data Using Numerical Measures

Chapter 3 Describing Data Using Numerical Measures Chapter 3 Student Lecture Notes 3-1 Chapter 3 Descrbng Data Usng Numercal Measures Fall 2006 Fundamentals of Busness Statstcs 1 Chapter Goals To establsh the usefulness of summary measures of data. The

More information

NON-LINEAR CONVOLUTION: A NEW APPROACH FOR THE AURALIZATION OF DISTORTING SYSTEMS

NON-LINEAR CONVOLUTION: A NEW APPROACH FOR THE AURALIZATION OF DISTORTING SYSTEMS NON-LINEAR CONVOLUTION: A NEW APPROAC FOR TE AURALIZATION OF DISTORTING SYSTEMS Angelo Farna, Alberto Belln and Enrco Armellon Industral Engneerng Dept., Unversty of Parma, Va delle Scenze 8/A Parma, 00

More information

EEE 241: Linear Systems

EEE 241: Linear Systems EEE : Lnear Systems Summary #: Backpropagaton BACKPROPAGATION The perceptron rule as well as the Wdrow Hoff learnng were desgned to tran sngle layer networks. They suffer from the same dsadvantage: they

More information

Statistics for Economics & Business

Statistics for Economics & Business Statstcs for Economcs & Busness Smple Lnear Regresson Learnng Objectves In ths chapter, you learn: How to use regresson analyss to predct the value of a dependent varable based on an ndependent varable

More information

Module 9. Lecture 6. Duality in Assignment Problems

Module 9. Lecture 6. Duality in Assignment Problems Module 9 1 Lecture 6 Dualty n Assgnment Problems In ths lecture we attempt to answer few other mportant questons posed n earler lecture for (AP) and see how some of them can be explaned through the concept

More information

The Geometry of Logit and Probit

The Geometry of Logit and Probit The Geometry of Logt and Probt Ths short note s meant as a supplement to Chapters and 3 of Spatal Models of Parlamentary Votng and the notaton and reference to fgures n the text below s to those two chapters.

More information

Lecture 3: Probability Distributions

Lecture 3: Probability Distributions Lecture 3: Probablty Dstrbutons Random Varables Let us begn by defnng a sample space as a set of outcomes from an experment. We denote ths by S. A random varable s a functon whch maps outcomes nto the

More information

arxiv:cs.cv/ Jun 2000

arxiv:cs.cv/ Jun 2000 Correlaton over Decomposed Sgnals: A Non-Lnear Approach to Fast and Effectve Sequences Comparson Lucano da Fontoura Costa arxv:cs.cv/0006040 28 Jun 2000 Cybernetc Vson Research Group IFSC Unversty of São

More information

Markov Chain Monte Carlo Lecture 6

Markov Chain Monte Carlo Lecture 6 where (x 1,..., x N ) X N, N s called the populaton sze, f(x) f (x) for at least one {1, 2,..., N}, and those dfferent from f(x) are called the tral dstrbutons n terms of mportance samplng. Dfferent ways

More information

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016 U.C. Berkeley CS94: Spectral Methods and Expanders Handout 8 Luca Trevsan February 7, 06 Lecture 8: Spectral Algorthms Wrap-up In whch we talk about even more generalzatons of Cheeger s nequaltes, and

More information

ELG4179: Wireless Communication Fundamentals S.Loyka. Frequency-Selective and Time-Varying Channels

ELG4179: Wireless Communication Fundamentals S.Loyka. Frequency-Selective and Time-Varying Channels Frequeny-Seletve and Tme-Varyng Channels Ampltude flutuatons are not the only effet. Wreless hannel an be frequeny seletve (.e. not flat) and tmevaryng. Frequeny flat/frequeny-seletve hannels Frequeny

More information

Negative Binomial Regression

Negative Binomial Regression STATGRAPHICS Rev. 9/16/2013 Negatve Bnomal Regresson Summary... 1 Data Input... 3 Statstcal Model... 3 Analyss Summary... 4 Analyss Optons... 7 Plot of Ftted Model... 8 Observed Versus Predcted... 10 Predctons...

More information

Convergence of random processes

Convergence of random processes DS-GA 12 Lecture notes 6 Fall 216 Convergence of random processes 1 Introducton In these notes we study convergence of dscrete random processes. Ths allows to characterze phenomena such as the law of large

More information

RELIABILITY ASSESSMENT

RELIABILITY ASSESSMENT CHAPTER Rsk Analyss n Engneerng and Economcs RELIABILITY ASSESSMENT A. J. Clark School of Engneerng Department of Cvl and Envronmental Engneerng 4a CHAPMAN HALL/CRC Rsk Analyss for Engneerng Department

More information

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction ECONOMICS 5* -- NOTE (Summary) ECON 5* -- NOTE The Multple Classcal Lnear Regresson Model (CLRM): Specfcaton and Assumptons. Introducton CLRM stands for the Classcal Lnear Regresson Model. The CLRM s also

More information

Economics 130. Lecture 4 Simple Linear Regression Continued

Economics 130. Lecture 4 Simple Linear Regression Continued Economcs 130 Lecture 4 Contnued Readngs for Week 4 Text, Chapter and 3. We contnue wth addressng our second ssue + add n how we evaluate these relatonshps: Where do we get data to do ths analyss? How do

More information

/ n ) are compared. The logic is: if the two

/ n ) are compared. The logic is: if the two STAT C141, Sprng 2005 Lecture 13 Two sample tests One sample tests: examples of goodness of ft tests, where we are testng whether our data supports predctons. Two sample tests: called as tests of ndependence

More information