Adaptive MMSE Rake-Equalizer Receiver Design with Channel Estimation for DS-UWB System

Transcription

1 Adaptive MMSE Rake-Equalizer Receiver Desig with Chael Estimatio for DS-UWB System Departmet of Electric Egieerig, Diwa Uiversity No.87-1, Nashi Li, Madou Tow, Taia Couty 7153, Taiwa, R.O.C. Abstract: - I this paper, a robust adaptive MMSE Rake-equalizer receiver scheme is preseted with chael estimatio that has bee trasmitted i Direct Sequece-Ultra Widebad (DS-UWB) system. The DS-UWB has a fie path resolutio by trasmittig iformatio with ultra short pulses. The Rake receiver is kow as a techique that ca effectively combie paths with differet delays ad obtai the path diversity gai. Due to sub-ao secods arrow pulse ad the large trasmissio badwidth i the systems, we therefore use equalizer to overcome the resultig ISI ad a log delay spread i the characterizatio of the UWB chael. The proposed receiver structure reduces itese multi-path destructio ad severe iter-symbol iterferece (ISI) by usig a combied adaptive Rake ad equalizer structure referred to as the MMSE (miimum mea square error) algorithm. Chael iformatio is obtaied through the use of a least mea square (LMS) adaptive algorithm o the traiig sequece of a preamble. The idoor eviromet is represeted by dese multipath chael models proposed by IEEE a. Simulatio results show that the ew scheme employig the eergy of a few paths to obtai better performace tha the traditioal maximum ratio combiig (MRC) Rake receiver. Key-Words: - Least mea squares (LMS), miimum mea square error (MMSE), adaptive receiver, direct sequece ultra widebad (DS-UWB) ratio (SNR) whe the iterferece is modeled as additive white Gaussia oise [6]. But for its complexity ad cost to remai low, the umber of figers that ca be afforded is too small to capture the ample eergy provided by the UWB chael, which etails a large umber of paths (ofte>50) [7]. I the presece of arrowbad iterferece, the bit error rate preferece of a UWB MRC Rake receiver exhibits a error floor. A more effective receiver scheme is the miimum mea square error (MMSE) Rake receiver [8] which achieves a much improved performace for WCDMA system. Meawhile, due to sub-ao secods arrow pulse ad the large trasmissio badwidth i the UWB system, the equalizer has also recetly attracted much attetio [9] as a meas to overcome the resultig iter-symbol iterferece (ISI) ad a log delay spread i the characterizatio of the UWB chael. It is reported that the equalizer ca obtai a excellet performace eve i strog frequecy selective chaels [10] which assume i perfect chael eviromet. Combiatio of spatial diversity combiig ad equalizer is a well established scheme for frequecy selective fadig chaels. However, like ay Rake structure, it eeds the complete chael state iformatio. Furthermore, the performace evaluatio of UWB systems over a multipath fadig chael is always coducted uder the assumptio of perfect chael estimatio [11]. 1. Itroductio Direct Sequece-Ultra Widebad (DS-UWB) offers great potetial i terms of low power, ehaced user capacity, high data rates ad ability to coexist with legacy services. Most of these beefits origiate from the uique characteristics iheret to UWB wireless trasmissios [1]. The UWB systems spread the eergy from several hudred MHz to a few GHz []. The received sigal is composed of a sigificat umber of multipath compoets that have differet delays o the order of aosecods. The UWB has a fie path resolutio by trasmittig iformatio with ultra short pulses. The Rake receiver is kow to be a techique that ca effectively combie paths with differet delays ad obtai the path diversity gai [3]. However, the UWB multipath chael is spread over dozes of symbols i the case of ultra high-speed commuicatios of several hudreds Mbps, which results i a strog frequecy selective chael [4]. Cosequetly, the Rake receiver eeds a large umber of figers ad the computatioal complexity of the Rake receiver becomes high [5]. I additio, the Rake receiver must estimate a large umber of chael impulse resposes, thus computatioal complexity of chael estimatio is very high. The covetioal Rake receiver employs the weight vector to perform the maximal ratio combiig (MRC) which maximizes the output sigal-to-oise ISSN: Issue 1, Volume 8, Jauary 009

2 The iaccurate chael estimatio [1] would lead to a serious degradatio of BER performace. The method was proposed i [13] to estimate chael ad equalizer desig i frequecy domai ad it requires extra FFT calculatios. I this paper, we itroduce a robust receiver desig icorporatig a chael estimatio scheme for DS-UWB over a realistic idoor multipath chael. The proposed receiver reduces itese multi-path destructio ad severe ISI by usig a combied adaptive Rake ad equalizer structure referred to as the adaptive MMSE Rake-equalizer receiver. Relevat receiver parameters are estimated usig the MMSE algorithm. This has motivated studies of multipath combiig receivers that process oly a subset of the available L p resolved multipath compoets. We obtai more robust sigal detectio at the receiver side by the adaptive chael estimatio i order to extract more accurate chael state iformatio. The chael characteristics are first estimated usig the LMS adaptive algorithm o the traiig sequece of the preamble. The proposed receiver is able to employ the eergy of a few paths ad obtai better performace by the proposed receiver desig. The performace of the adaptive MMSE Rake-equalizer structure for DS-UWB system is ivestigated for differet umber of Rake figers ad equalizer taps. The remaider of the paper is orgaized as follows. Sectio II briefly describes the proposed DS-UWB system model icludig trasmitter, UWB idoor chael model ad adaptive receiver structure. I Sectio III, a adaptive MMSE Rake-equalizer is proposed cojuctio with a LMS chael estimator. Sectio IV ivestigates the performace of the proposed method ad compares it with the traditioal approach. Fially, some cocludig remarks are give i Sectio V.. System Model We show the trasmitter ad receiver structure for DS-UWB i Fig. 1. The data symbol sequece modulated by BPSK is first spread usig the spreadig code. The trasmitted sigal of the cosidered DS-UWB system is give by s( t) bw( t Ts) (1) 1 where b { 1} is the th trasmit data symbol, T s is the symbol duratio ad the spreadig waveform w(t) is expressed as M w( t) cmp( t mtc), () m1 where c m deotes the m th chip of the spreadig code of legth M ad T c is chip duratio (T s =MT c ). The chip pulse p(t) represets the moocycle waveform ormalized to have uit eergy ad the spreadig code is defied i the DS-UWB specificatio [14]. I this paper, each DS-UWB is evaluated usig the UWB multipath chael model based o the idoor chael measuremets i the -8GHz frequecy bad accepted by the IEEE SG3a stadard [15]. The UWB idoor chael model adopts a double-expoetial decay itesity profile based o the Saleh-Valeuela model. It provides eough degrees of freedom to match chael measuremets, ad ca be used to match the NLOS ad LOS chael characteristics separately. The multipath model cosists of the followig discrete time impulse respose: where LT K l (3) l0 k0 h( t) ( t T ) is the real multipath gai coefficiet, T l is the delay of the l th cluster, ad is the delay of the k th multipath compoet of the l th cluster arrival time (T l ). L T is the total umber of clusters ad K is the total umber of ray paths relative to the l th cluster. The chael coefficiets ca be expressed as, where p k,l is equally likely to take o p the values of 1, ad is the logormal fadig term. The distributio of cluster arrival time ad the ray arrival time are give by ptl Tl 1 exp Tl Tl 1, l 0 k, l ( k 1), l ( k 1), l p exp, k 0 The chael coefficiets are defied as follows: 0log10( ) Normal(, ) E k, l k, l T / / l k,l e e k, l 0 10 l( 0 ) 10Tl / 10 / l(10) l l(10) 0 respectively, where = the mea power of the first path of the first k,l cluster T l = the arrival time of the first path of the lth cluster; k,l = the delay of the kth path withi the lth cluster relative to the first path arrival time, T l = cluster arrival rate; = ray arrival rate, i.e., the arrival rate of path withi each cluster. = cluster decay factor; ISSN: Issue 1, Volume 8, Jauary 009

3 = ray decay factor; =stadard deviatio of logormal fadig (db). Table 1 lists a example of these key parameters for four differet chael models. The above results assume that the UWB chael model fits the measuremets take i the home eviromet for both LOS ad NLOS cases. I this study, the chael model parameters have matched the UWB chael measuremets with a typical rms delay spread of 15-0s. The chael coefficiets are ormalized to remove the effect of path loss for the UWB chael model. Hece, the received sigal, rt, () at the receiver s iput is give by LT K r( t) b w( t T T ) u( t) 1 l0 k 0 s l (4) where ut () is the additive white Gaussia oise (AWGN) with two sided power spectral desity N /. I geeral, a receiver structure cosists of L 0 correlators followed by a Rake combier. Each correlator correlates the received sigal with the referece waveform at the delay time ad itegrates over T. s 3. Chael Estimatio ad Adaptive MMSE Rake-Equalizer Receiver The proposed receiver structure for DS-UWB system is as show i Fig.. I this sectio, the proposed receiver reduces itese multi-path destructio ad severe ISI by usig a combied adaptive Rake ad equalizer structure referred to as the adaptive MMSE Rake-equalizer receiver. Due to the large trasmissio badwidth, the UWB chael is characterized by a log root-mea square delay spread ad the Rake receiver caot always overcome the resultig ISI [16]. We therefore study equalizatio for DS-UWB systems. The chael characteristics are first estimated usig the LMS adaptive algorithm ad iitial value by the traiig sequece of preamble. The structure of the preamble is show i Table. A adaptive receiver structure ad chael estimatio for DS-UWB system is ivestigated, which ca improve the performace of bit error rate. A. Chael estimatio by usig LMS adaptive algorithm Equatio (4) ca be writte as r( t) b g( t Ts) u( t) (5) 1 where g( t) w( t) h( t) (6) We employ the LMS adaptive algorithm to perform the chael estimatio. We assume the kow traiig sequece b [ b1, b, b P ] icludig p bit ad chael atteuatios i DS-UWB system to be estimated ito a matrix g. A fadig chael with L resolvable paths is cosidered ad the received sigal at time show ito a vector r P. Let us defie matrix g [ g, g,, g ] of size ( P L) p,1 p, p, L where g p,1 (p=1,,,p) is the first colum vector. At the first iteratio, the iitial chael estimatio, g, which cotais chael iformatio, is based oly upo traiig sequeces that were computed. The liear least square chael estimatio [17] ca be estimated by the Gauss-Markov theorem give by: H g b r (7) P where = r ( ), r ( ),, r ( ) r ad P 1 g1,1 g1, g1, L g,1 g, g, L = g gp,1 gp, gp, L P L The adaptive chael estimatio based o LMS algorithm ivolves three basic iterative steps: Step1: Filterig ( ) ( ) ( ) b r ( H ) P g (8) Step: Computig estimatio error ( ) ( ) e b H ( b H ) (9) Step3: Adaptive weight adjustmet ( 1) ( ) ( ) ( ) ( ) ( ) ( ) g g c g c E[ e ] ( ) b ( ) ( ) ( ) ( ) g e ( r ) c p (10) where () represets the time step i the iterative process, is the positive step-size parameter, c ( ) is the gradiet of E[ e ] ad the superscript H ( ) represets Hermitia traspose. Note that e ( ) itself is a very oisy approximatio to E[ e ]. We ca get a oisy approximatio to the gradiet by ( ) fidig the gradiet of E[ e ] (Hoff first published the LMS algorithm based o this clever idea, i 1960.) By meas of the self-optimizig techique [18], we ca derive the optimum step-size parameter ( ) [ c ] opt 1/ r p, which is time-varyig ad iversely proportioal to istataeous power of the chael estimator s iput sigal. Thus adaptive weight adjustmet with the optimum step-size L ISSN: Issue 1, Volume 8, Jauary 009

4 parameter is: ( 1) ( ) ( ) ( ) g g e ( r ) / r (11) p Therefore, g ca be estimated at the iterative process i matrix form as g [ g p,1, g p,,, g p, L ], which cotais chael iformatio (The amplitudes ad delays are icorporated ). B. Adaptive MMSE Rake-Receiver Weight Estimatio The weights of the traditioal Rake receiver are H H H β [ β1, β,, βl] [ g p, L, g p, L1, g p,1]. The proposed adaptive MMSE Rake receiver i Fig. 3 is composed of a Rake combie with L figers to collect sigal eergy ad mitigate ISI ad other iterferece. The adaptive rake structure attempts to utilize mulipath diversity i the same style as the rake receiver o each figer. Miimisig the cost fuctio E { e [ ] } results i each compoet acquirig iformatio o how to improve the fial estimate I [ ] takig all iput samples ito accout. The weight coefficiets of the L taps for the MMSE filter are chose as to miimize the MSE betwee the desired output ad the received output at each figer. The weight vector β cotais all of the filter coefficiets ad the receiver vector r [ r [ ], r [ ], r [ ]] cotais all the filter iput 1 L samples. Assumig the perfect sychroizatio betwee the trasmitter ad the receiver, the lth correlator output rl [ ] (l= 1,, L) for the th desired data symbol. The output I [ ] of the Rake-Receiver system ad update equatios are cosequetly as follows. ( ) ( ) H ( ) I [ ] r ( β ) (1) ( ) ( ) e[ ] I [ ] d[ ] (13) H ( ) ( ) H ( 1) H ( ) ( ) e [ ] ( β ) ( β ) r ( ) r (14) As with the MMSE receiver the filters are oly required to compute a output ad perform coefficiet updates at symbol rate. The iterative scheme is preseted to obtai more accurate performace, which ca efficietly suppress the ( ) ( 1) ifluece of oise compoet. If [ ] e e[ ], the iterative ca be termiated ad the fial decisio is available. The just as we see, the iteratio operatio is a key feature of our proposed system ad much more precise receiver desig ca be achieved. It is ature to assume the more the iteratio time is, the better the performace will be achieved. p However, we will see through the simulatio results that the performace improvemet obtaied by the iteratio will appear saturate after some iteratio. C. MMSE Chael Equalizer Coefficiets Estimatio Due to the tremedous high data rate ad loger multipath delay spread as CM3 ad CM4, the Rake receiver caot always overcome the resultig ISI [19]. We therefore eed to apped equalizer for DS-UWB systems. I the MMSE Rake-equalizer receiver, the equalizer is used to mitigate residual iterferece. I this paper, assumig that the th data bit is beig detected, the MMSE criterio is to miimize E[ b ˆ b[ ] ]. Let the umber of equalizer taps be J ad the coefficiets of the equalizer be { c J,, cj}. The the equalizer output is J bˆ[ ] c ji [ j ]. (15) jj Let us defie a vector C which cotais all of the equalizer Coefficiets ad a vector b which cotai all the [ ] b bˆ [ ] J bˆ[ ] c ji [ j ]. jj To optimize performace, the equalizer coefficiets, c [ c,, c ], are chose to miimize the MSE of J J its output, that is ˆ c arg mi E[ b b[ ] ] (16) c 4. Simulatios Results UWB systems have a badwidth exceedig 1GHz ad satisfy the spectrum maskig requiremet of the FCC, the UWB pulse, or also kow as the Gaussia moocycle waveform i [0], is chose to be the 1 th derivative of the Gaussia fuctio ad it ca be expressed as, tt0 t t w() t t e (17) where is a time costat for cotrollig the width of the pulse ad is chose accordig to the spectral mask requiremet of the FCC ad t 0 is a time offset. We evaluate the performace of DS-UWB with chael estimatio ad Rake receiver by computer simulatio. The modulatio format for DS-UWB is ISSN: Issue 1, Volume 8, Jauary 009

5 assumed to be BPSK with the spreadig code of legth 4. The trasmitter pulse shapig has uit eergy ad the duratio value for pulses trasmitted is s. We assume the perfect chip sychroizatio betwee the trasmitter ad the receiver. The characteristics of the covolutioal ecoder are such that its rate is 1/ ad costrait legth is 6. The raw data rate is set to 8Mbps i the lower operatig bad. The receiver uses a traiig sequece legth of 56 bits. Two differet chael models (CM ad CM4) were simulated usig at least 100 chael realizatios ad the umber of equalizer taps is 1( J 1). Fig. 4 presets the bit error rate of the DS-UWB system based o the IEEE-UWB CM chael model for the perfect kowledge of the chael characteristics with differet umbers of Rake figers by varyig the umber of iteratios. The proposed MMSE Rake-equalizer receiver performs quite well uder the assumptio of a perfect chael estimate ad that the umber of Rake figers is 5 ad 0. As idicated i Fig. 4, it is show that the BER performace is improved by approximately 0.5dB after three iteratios as compared with that of oe. Fig. 5 ad Fig. 6 preset the bit error-rate performaces of the after three iteratio of the proposed receiver ad the MRC Rake receiver with perfect chael estimatio for the cases of CM ad CM4 chaels respectively. It is clearly see that performace improves with icreasig umber of Rake figers sice more eergy is effectively used i the receiver as the umber of Rake figers icreases. It is also see that the MRC Rake receivers greatly uderperforms the proposed iterative receiver uder perfect chael estimatio. I fig. 5, a adaptive MMSE Rake-equalizer receiver with umber of Rake figers L=5 performs as well as a MRC Rake receiver with umber of Rake figers L=0 over CM chael. I fig. 6, a adaptive MMSE Rake-equalizer receiver with umber of Rake figers L=5 outperforms a MRC Rake receiver with umber of Rake figers L=10 over CM4 chael. Fially, Fig. 7 ad Fig. 8 show the bit error rates of the DS-UWB system usig the proposed iterative receiver with LMS chael estimatio ad assumed perfect chael estimatio over CM ad CM4 chaels respectively. All detectors were implemeted usig the adaptive MMSE Rake-equalizer receivers with differet umbers of figers. It shows that the adaptive algorithm implemeted ca give a satisfactory performace. I practical receivers, however, it might ot be viable to implemet too may figers i the Rake receiver. 5. Coclusios I this paper, the performace of a robust MMSE Rake-equalizer receiver scheme i DS-UWB system is ivestigated for differet umber of Rake figers ad equalizer taps. The proposed receiver reduces itese multi-path destructio ad severe ISI by usig a combied adaptive Rake ad equalizer structure referred to as the MMSE Rake-equalizer receiver. This paper studies a DS-UWB system that employs LMS chael estimatio for cacelig chael distortios subsequet to obtaiig the chael iformatio via a traiig sequece. The proposed receiver is able to employ the eergy of a few paths ad obtai better performace by the additio of a chael estimatio scheme. The proposed receiver is a modified versio of the covetioal Rake receiver i which a Rake receiver with a simple equalizer is used to couteract the effects of the delay spread of UWB chaels ad the MMSE criterio is used to obtai the filter weights. Simulatio results show that chael estimatio is ecessary i a DS-UWB system. The proposed adaptive MMSE Rake-equalizer receiver ehaces performace very well with a assumed perfect chael estimate. It is show that icreasig the umber of Rake figers is effective i improvig system performace. We also show that DS-UWB with adaptive MMSE Rake-equalizer has better BER performace tha DS-UWB with MRC Rake receiver, especially whe the umber of figers is large. I particular it is show that the performace of a adaptive MMSE Rake-equalizer receiver of umber of Rake figers equal to 5 (L=5) outperforms a MRC Rake receiver of umber of Rake figers equal to 10 (L=10) over CM4 chaels. Fially, the LMS algorithm does give a satisfactory performace with fast covergece. I practical receivers it might ot be able to implemet too may figers i the Rake receiver. The proposed adaptive MMSE Rake-equalizer receiver is able to employ the eergy of a few paths ad obtai better performace by the suitable receiver desig. It shows that the adaptive algorithm gives a satisfactory performace. Refereces: [1] L. Yag ad G. B. Giaakis, Ultra-widebad commuicatios: a idea whose time has come, IEEE Sigal Process. Mag., vol. 1, o. 6, pp. 6 54, Nov [] M. Z. Wi ad R. A. Scholtz, Ultra-wide badwidth time-hoppig spread-spectrum impulse radio for wireless multiple-access commuicatios, IEEE Tras. o Comm., pp , 000. [3] A. Rajeswara, V. S. Somayazulu ad J. R. ISSN: Issue 1, Volume 8, Jauary 009

6 Foerster, Rake performace for a pulse based UWB system i a realistic UWB idoor chael, Iteratioal Coferece o Commuicatios, vol. 4, pp , May 003. [4] J. Foerster ad Q. Li, "UWB chael modelig cotributio from Itel," IEEE P /79-SG3a., Sept. 00. [5] Nig Xie ad Yuapig Zhou, A Adaptive Noliear Rake Receiver i UWB Wireless Commuicatios, IEEE Iteratioal Coferece o Networkig, Sesig ad Cotrol, pp , 3-5 April, 006. [6] H. Sheg, A. M. Haimovich, A. F. Molisch ad J. Zhag, Optimum Combiig for Time Hoppig Impulse Radio UWB Rake Receivers, IEEE Coferece o Ultra Widebad Systems ad Techologies, pp. 4-8, Nov [7] Farahmad, S., Xiliag Luo ad Giaakis, G.B., Demodulatio ad trackig with dirty templates for UWB impulse radio: algorithms ad performace, IEEE Trasactios o Vehicular Techology, vol. 54, pp , Sept [8] G.W. Rice, D. Garcia-Alis, I. G. Stirlig, S. Weiss ad R. W. Stewart, A Adaptive MMSE Rake Receiver, Asilomar Coferece o SSC, vol. 1, pp , 9 Oct.-1 Nov [9] D. Falcoer, S. L. Ariyavisitakul, A. B. Seeyar ad B. Eidso, Frequecy domai equalizatio for sigle-carrier broadbad wireless systems, IEEE Commu. Mag., pp , 00. [10] M. Eslami ad Xiaodai Dog, Performace of Rake-MMSE-equalizer for UWB commuicatios, IEEE WCNC, vol., pp , March, 005. [11] J.F. Liao, C.L. Tsai ad B.S. Che, Robust adaptive chael estimatio ad multiuser detectio for ultra widebad i a realistic idoor chael, Iteratioal Coferece o Commuicatios, 16-0 May, 005. [1] Y. Ishiyama ad T. Ohtsuki, Performace evaluatio of UWB-IR ad DS-UWB with MMSE-frequecy domai equalizatio (FDE), IEEE Cof. o Globalcom, pp , Dec [13] H. Sato ad T. Ohtsuki, Frequecy domai chael estimatio ad equalizatio for direct sequece ultra widebad (DS-UWB) system, IEE proceedigs commuicatios, vol. 153, pp , Feb., 006. [14] IEEE P80.15 Workig Group for Wireless Persoal Area Networks (WPANs) DS-UWB Physical Layer Submissio to Task Group 3a, IEEE P /0317r4, Jauary, 005. [15] IEEE P80.15 Workig Group for Wireless Persoal Area Networks (WPANs) UWB Chael Modelig Cotributio from Itel, IEEE P /79r0-SG3a, 4 Nov., 005. [16] H. Niu, J. A. Ritcey ad H. Liu, Performace of UWB Rake receivers with imperfect tap weights, IEEE Itl. Cof. o Acoustic, Speech ad Sigal Processig, vol. 4, pp , April 003. [17] Hiroyuki SATO ad Tomoaki OHTSUKI, Computatioal Complexity ad Performace of Rake Receivers with Chael Estimatio for DS-UWB, IEICE Tras. Fudametals, pp , 005. [18] X. Hou, S. Li, D. Liu, C. Yi ad G. Yue, O Two-Dimesioal Adaptive Chael Estimatio i OFDM Systems, IEE Proceedigs Commuicatios, vol. 153, Oct [19] Ye Li, F. M. adreas ad Jiyu Zhag, Chael Estimatio Ad Sigal Detectio for UWB, A Mitsubishi Electric Research Laboratory, Nov [0] R. Dilmaghai, M. Heradez ad R. Koho, Optimum sequece desig ad performace evaluatio of chael estimatio techiques for UWB systems, IEE Proc. Comm., pp 41 45, April 005. ISSN: Issue 1, Volume 8, Jauary 009

7 data { b } spreadig isertio of CP pulse geerator st () chael estimatio rt () removal of CP Rake receiver desspreadig { b } Fig.1 The trasmitter ad receiver structure for DS-UWB Table 1 Multipath chael model parameters example. Model Parameters CM 1 LOS(0-4m) CM NLOS(0-4m) CM 3 NLOS(4-10m) CM 4 Extreme NLOS (1/sec) (1/sec) (db) ut () b DS-UWB st () Trasmitter UWB Idoor Chael + r(t) Adaptive MMSE Rake-Equalizer Receiver bˆ LMS adaptive Chael estimatio Fig.. The proposed receiver structure for DS-UWB system Table Structure of the geeral UWB preamble Acq seq 9s SFD (3bit) Data Field (4bit) Traiig (various legths) PHY Header MAC Header Data ISSN: Issue 1, Volume 8, Jauary 009

8 dt 1st figer r [ ] 1 β 1 d [ ] rt () w( t t L ) w( t t L ) 1 w( t t L ) dt dt d figer r[ ] L th figer r [ ] L β β L I [ ] Adaptive MMSE Rake receiver e [ ] c J c J 1 Ts Ts c b ˆ[ ] J MMSE equalizer Fig. 3. The structures of the Adaptive MMSE Rake-equalizer receiver model. Ts bˆ Fig. 4. Bit error rates of MMSE Rake-equalizer receiver with perfect chael estimatio by varyig the umber of iteratios for DS-UWB over CM chaels Fig. 5. Bit error rates of the proposed receiver ad MRC Rake receiver with perfect chael estimatio at differet umbers of Rake figers for DS-UWB over CM chaels ISSN: Issue 1, Volume 8, Jauary 009

9 Fig. 6. Bit error rates of the proposed receiver ad MRC Rake receiver with perfect chael estimatio at differet umbers of Rake figers for DS-UWB over CM4 chaels Fig. 7. Bit error rate of a proposed receiver usig LMS chael estimatio ad that of a proposed receiver with assumed perfect chael estimatio for differet umbers of Rake figers for DS-UWB over CM chaels ISSN: Issue 1, Volume 8, Jauary 009

10 Fig. 8. Bit error rate of a proposed receiver usig LMS chael estimatio ad that of a proposed receiver with assumed perfect chael estimatio for differet umbers of Rake figers for DS-UWB over CM4 chaels ISSN: Issue 1, Volume 8, Jauary 009