Performance Analysis of Power Line Communication Using DS-CDMA Technique with Adaptive Laguerre Filters

Transcription

1 Internatonal Conference on Informaton and Electroncs Engneerng IPCSIT vol.6 ( ( IACSIT Press, Sngapore Performance Analyss of Power Lne Communcaton Usng DS-CDMA Technque wth Adaptve Laguerre Flters S. Lewang and C. Banjangaprasert + Faculty of Engneerng, Kng Mongut s Insttute of Technology Ladraang, Bango 5, Thaland E-mal : _am_pon3@hotmal.com, chawal@mtl.ac.th Astract. Ths paper presents the system performance of the power lne communcaton (PLC usng drect-sequence code dvson multple access (DS-CDMA technque wth an adaptve equalzer ased on Laguerre flter. The ojectves are to mprove the performance of system y means of t error rate (BER at the recever, and to elmnate the effect of nose dsturance due to addtve whte Gaussan nose (AWGN, mpulse nose and multpath fadng n the power lne channel. Our smulaton results show that the proposed recever performs lower BER than the conventonal FIR flter recever. Keywords: PLC, DS-CDMA system, Adaptve Laguerre flter.. Introducton As a result of recent developments, the electrcal power supply system s on the way to mgrate from a pure energy dstruton networ to a multpurpose medum delverng energy, voce, and varous data servces. In partcular, Internet access s currently n the focus of the efforts of varous research actvtes []. There are lmts n transfer power lne systems[-5] whch are n part due to some dffcultes found when sendng the data system. Some examples of these dffcultes nclude mpedance value (change all the tme, nterference sgnal n the power lne system, attenuaton that has a hgh value. A smple approach to rough estmaton of the transfer functon of power lne channels was presented n []. A straght lne can nterpolate the attenuaton ncreasng wth hgher frequences, so a smple equaton can e found to calculate the ampltude of the channel transfer functon. As ths approach does not consder multpath propagaton and the resultng notches of the channel transfer functon, more detaled models had to e developed. Multpath propagaton approaches, whch are sutale for descrng the transmsson ehavour of power lne channels, have een proposed n [3] and [4]. Drect sequence code dvson multple access (DS-CDMA[6] s a canddate access technque for a thrd generaton mole system. In DS-CDMA system, sgnal transmsson over tme-varyng multpath mole rado channel produces ntersymol nterference (ISI and other tme-varyng effect from the channel. Adaptve equalzers are generally requred to mtgate the effect of the channel. Laguerre flter[7], sharng the characterstc of fnte mpulse response (FIR flter and nfnte mpulse response (IIR flter, represents a compromse etween computatonal complexty and stalty. The recursve least squares (RLS algorthm s the one of the most popular algorthm for the adaptve flter applcaton [7-]. The man feature of RLS algorthm s ts fast convergence rate, especally for hghly correlated nput sgnal. Therefore, n ths paper nterests the RLS algorthm for the proposed equalzer. In ths paper the performance of the power lne communcaton usng drect-sequence code dvson multple access(ds-cdma technque wth adaptve Laguerre equalzer s presented. The ojectve of ths paper s to ncrease potental and decrease t error rate, whch occurs from nterference of the sgnal n power lne systems. + Correspondng author. Tel.: ; fax: E-mal address: chawal@mtl.ac.th 5

2 . Multpath Model for Power Lne Channel The echo model descres the channel mpulse response as a superposton of N drect pulses representng the superposton of sgnals from N dfferent paths. Each of these mpulses s multpled y a complex factor ρ and delayed y tme τ. The factors ρ represent the product of reflecton and transmsson factors along each echo path. The complex channel transfer functon s gven as [3] N jπ fτ ( ρe = H f = ( Ths model allows realstc reproducton of notches of the channel transfer functon and s therefore well suted to descre ndoor channels where the low-pass characterstc of the channel s not relevant. In [4] descres a unversal and practcally useful form of the complex transfer functon for power lne channels y an adapted echo model that contans an addtonal attenuaton factor. Ths model represents the superposton of sgnals from N dfferent paths, each of whch s ndvdually characterzed y a weghtng factor g and length d. Furthermore, frequency-dependent attenuaton s modeled y the parameters a, a, and. Then, the useful form of the complex transfer functon for power lne channels can e expressed as d N jπ f ( a+ af d v p = H( f = ge e ( Whle the frst exponental functon descres attenuaton, the second one, ncludng the propagaton speed v p, represents the echo scenaro. The parameters for the multpath model can e otaned from measurements of the complex channel transfer functon. The attenuaton parameters a (offset of attenuaton, a (ncrease of attenuaton, and (exponent of attenuaton can e otaned from the magntude of the frequency response. To determne the path parameters d and g, the mpulse response s necessary. The mpulse response gves nformaton aout the tme delay of each path, whch s proportonal to d. The weghtng factors g can e otaned from the ampltude of each mpulse. 3. Drect Sequence Code Dvson Multple Access(DS-CDMA DS-CDMA s the method that allows the data sgnal to drectly modulate wth dgtal sgnal. The data sgnal can e acheved n oth an analog and dgtal sgnals. The data sgnal wll multply the code sgnal drectly and then rng the sgnal to modulate wth the carrer sgnal. Ths paper wll consder the data transfer through power lne channels y the DS-CDMA system, whch has the followng model. a nt ( s a H ( f, r( t rt ( s T dt y s s a s ( t Fg. Transmtter model. Fg. Recever model. In consderaton of the data transfer model n the DS-CDMA system through power lne channel, whch has nose n whte Gaussan nose, and assgn users n the system for user, s a data t at of user (whch has T s tme perod of sgnal, a s sze of acceptale sgnal from recever, s s spread code of user, whch has chp equal to L c, allows the followng formula to e formed : L c s ( = s ( x( t T (3 c = A formula showng the collectve sgnal from the recever can e wrtten as follows: K rt ( = ast ( Ts* ht ( + nt ( (4 = = 5

3 Recevers n DS-CDMA system s duty s separate to each user, where the users recever ncludng despread code of each user and ntegraton crcut can e shown n the Fg.. We oserved that each user s recever usng only the own spread data n separate wanted sgnal then rng the sgnal past to t decson crcut n order to get the desgned data. If y s outcome sgnal from user s recever at T s, we can wrte the followng formula: Fnally, the decson output of the detector can e defned as 4. Adaptve Laguerre Equalzer y T s = r( t s ( t dt (5 = sgn( y (6 Adaptve equalzer s the man requred for the system to acheve hgh performance of the communcaton. Therefore, n ths secton we wll refly descre the structure of the equalzer and followed y the usefulness adaptve algorthm. 4. Adaptve Laguerre equalzer structure The structure of an adaptve Laguerre transversal equalzer[7] s shown n Fg. 3. u( a L( a a u ( t, u( t, u ( t, u M ( t, L( a w ( t w ( t w ( w M ( dˆ ( e( d( Fg. 3 Adaptve Laguerre equalzer. The Laguerre sequences have smple z-transforms, as shown y formula: L ( z α ( αz ( = + The flter at ts front end s represented y the transfer functon α α =,,... (7 α L ( α = (8 αz Ths low-pass flter s followed y a cascade of dentcal all-pass flters of order one, each of whch has the transfer functon z α α =, for α < (9 αz Accordngly, the Laguerre functon may e expressed as L ( ( L(, = Specfcally, when the parameter α s set equal to zero, we have α =,, ( α ( = L (, = z α z for ( The equalzer processes egn wth decson of the optmal pole and the ntal values of the tapsweght of the Laguerre transversal equalzer can e set w (. The nput and output of the Laguerre flter at ^ tme t are denoted y u ( and d ( t. d ( s the desred response of the adaptve Laguerre transversal ^ equalzer, and e ( t = d ( t d ( t s the estmaton error. 4. Recursve Least Squares (RLS algorthm Let us consder the exponentally-weghted least squares prolem descred for the structure n Fg. 3, whch sees w ( that solves 5

4 The squared error can e defne as J( t t = λ e( (3 = e( = d( w( t u( (4 and the update equaton of the adaptve RLS algorthm s gven as follows [7]: T t u ( ( = λ T + u( t u ( (5 = [ t ( u( t ] λ (6 w( = w( t + ( e ( (7 where = δ I and δ s small postve constant, I s dentty matrx and the forgettng factor < λ <, whch s usually chosen n the range of.9 < λ <. 5. Numercal Results In ths secton, the BER performance of the proposed technque for the PLC channel recever s presented. Besdes the measurement data n Fgs. 4-6, smulaton results are derved from the followng parameters : length of transmsson nformaton s equal to ts, the length of the ndvdual path of the PLC channel s m, frequency response of PLC channel s MH an attenuaton parameter 3 a = 9.4 and a = 7.8, exponent of the attenuaton factor =.7 and weghtng factor for path g =.64. We assume the DS-CDMA and adaptve RLS algorthm parameters are as follows : processng gan s equal to 4, and δ s set to e.3. Fg. 4 shows the convergence process of the LMS and RLS algorthms for adaptve FIR equalzer and adaptve Laguerre equalzer. From Fg. 4, t s evdent that the RLS algorthm wth Laguerre equalzer s potentally more suted for power lne communcaton system ecause that fast convergence rate. Fg. 5 dsplays the BER performance comparson etween addtve whte Gaussan nose (AWGN channel and AWGN wth mpulse nose channel. From the smulaton results can e oserved that mpulse nose s a severe degradaton factor for the performance of the PLC system. Fnally, Fg. 6 dsplays the average BER for each SNR. From the smulaton results can e oserved that adaptve equalzer usng Laguerre structure has lowest BER when compared to the conventonal adaptve equalzer for every SNR. Therefore, the proposed method can ncrease the performance of the system. 6. Concluson Ths paper presents the performance of power lne communcaton usng DS-CDMA technque wth adaptve Laguerre structure flters through the power lne channel. From the smulaton results, t can e concluded that the effcency of the power lne communcaton systems can greatly mproved y usng the proposed technques. It s easy to devse algorthms to adapt the weghts of Laguerre flter smlar to the LMS or the RLS developed for the transversal flter. Tap Weght Numer of teratons Fg. 4 Convergence process of adaptve equalzers. 53

5 Fg. 5 BER performance comparson etween AWGN and AWGN wth mpulse nose envronments. 7. References Fg. 6 Average BER performance of adaptve equalzers. [] P. Langfeld and K. Dostert, OFDM system synchronsaton for power lne communcatons, n Proc. 4 th Int. Symp. on Power Lne Communcatons and ts Applcatons, Lmerc, Ireland,, pp.5. [] C. Hensen and W. Schul Tme dependence of the channel characterstcs of low voltage power lnes and ts effects on hardware mplementaton, AEÜ Int l. J. Electroncs and Communcatons, Vol. 54, No., Fe.. [3] H. Phlpps, Modelng of power lne communcaton channels, Proc. 3 rd Int l. Symp. Power Lne Communcatons and ts Applcatons, Lancaster, UK, 999, pp. 4. [4] M. Zmmermann and K. Dostert, A multpath model for the power lne channel, IEEE Trans. Communcatons, Vol. 5, No. 4, Aprl, pp [5] M. Zmmermann and K. Dostert, Analyss and modelng of mpulsve nose n road and power lne communcatons, IEEE Trans. Electromagnetc Compatlty, Vol. 44, No., Fe., pp [6] M.B. Pursley, Drect-sequence spread-spectrum communcatons for multpath channels, IEEE, Mcrowave Theory and Technques, Vol. 5, Issue 3, March, pp [7] S. Hayn, Adaptve Flter Theory, 4 th edton, Prentce-Hall,. [8] R. Wang, N. Jndal, T. Bruns, A. R. S. Baha, and D. C. Cox, Comparng RLS and LMS adaptve equalzers for nonstatonary wreless channels n mole AD-HOC networs, Proceedngs of 3 th IEEE Internatonal Symposum on Personal, Indoor and Mole Rado Communcatons, pp.3-35,. [9] T. Somur, and C. Benjangaprasert, Adaptve equalzaton for DS-CDMA mult-user communcaton system usng Laguerre flters, Internatonal Conf. on Control, Automaton and Systems, 8, pp [] S. He and X. Shan, Adaptve equalzaton for multpath fadng channels usng Laguerre flters, IEEE Pacfc Rm Conf. on Communcatons, Computers and Sgnal Processng, vol., 3, pp