Train accurate localization using ultra wide band radio and time reversal

Transcription

1 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Volume 3, Issue 1, June 014 Tran accurate localzaton usng ultra wde band rado and tme reversal B. FALL, F. Elbahhar, M. Heddebaut and A. Rvenq Unversty of Valencennes (UVHC), French nsttute of scence and technology for transport, development and networks (IFSTTAR)-France Abstract In guded transport, the precse localzaton of trans s vtal for nonal operaton of the transport system. In a propagaton envronment such as a ralway lne, an effectve localzaton sensor s complex to desgn snce t must operate n the presence of many fxed and moble obstacles consttuted by the nfrastructure and the trans. In order to desgn a sensor delverng hgh localzaton performance, we propose the use of so-called spectral dversty technques also found under the name of ultra wde band rado. In ths work, ths rado technque s assocated to the tme reversal technque, takng advantage of the complex ralway propagaton envronment. The objectve s to obtan a relable and robust localzaton of ral vehcles by focusng rado sgnals from a ground balse to the track and, therefore, to the passng tran recevng antennas. A theoretcal model, smulatons and an expermental valdaton were developed on the propertes of energy focusng of tme reversal takng nto account very dfferent types of envronment. Several parameters related to antenna confguratons were nvestgated. The contrbutons of tme reversal to the accuracy of the postonng system are measured by comparng an ultra wde band postonng system alone and then, combnng t wth tme reversal. Index Terms Tran Localzaton; Balse; Ultra Wde Band Rado; Tme Reversal; Ralway Infrastructure; Temporal Focusng; Spatal Focusng. I. INTRODUCTION In recent years, operators and users of guded transport systems have expressed a strong demand to provde safe and effcent transport, provdng an ncreasng qualty of servce. Among all the techncal requrements whch are consequently generated, t s essental to ensure an adequate exchange of nformaton between vehcles and nfrastructure, regardless of the type of applcaton or the propagaton envronment. In such envronments, t s also necessary to allow accurate localzaton of trans. The rapd development of transport n terms of speed, complexty and dversty of envronments generate dffcultes to develop such equpment and studes have been performed regardng the optzaton of the ground to tran communcaton [1]. Usually, ground to tran rado communcaton explots access ponts nstalled along the track, exchangng data wth moble equpment nstalled on the trans. Ether propretary rado modems or rado modems derved from exstng standards are used. Currently, all these rado modems operate snusodal sources of sgnals occupyng rado channels over a lted bandwdth. For the tran localzaton process, drfts of the tran odometer, usually composed of a wheel turn counter and a Doppler radar that contnuously calculate the poston and velocty data, are perodcally compensated by ground balses nstalled between the rals. Balses are workng as klometer-markers and transt ther absolute localzaton to passng trans. The requested localzaton accuracy s mportant and should allow, for example n automated urban subways, vehcles to repeatedly stop n front of staton doors, thus, necesstatng a few centmeters localzaton accuracy. Ths paper proposes a new approach for ralway track-to-tran short range communcaton smultaneously provdng accurate localzaton nformaton. Breakng wth the recalled conventonal approaches, an assocaton between Ultra Wde Band (UWB) rado and Tme Reversal (TR) technques consttutes the heart of the work outlned n ths paper. The prncple of UWB communcaton s based on the esson of sgnals at low power and extremely broad spectra to obtan hgh flow rates and also facltate flght tme, hghly accurate measurements []. The TR technque consttutes a focusng technque of the energy radated by an antenna or multple transttng antennas to one or more recevng antennas by the nserton, at the transtter, of a channel matched flter. The remander of ths paper wll now descrbed these technques, ther assocaton appled to the balse, and the results obtaned n the context of ths ralway perspectve. It s organzed as follows. Secton II descrbes the proposed ralway balse and ntroduces the UWB and TR technques. Usng dfferent confguratons, secton III develops a theoretcal and smulaton study of the UWB and TR coupled system. Secton IV presents the used expermental TR-UWB up and the assocated expermental results. Fnally, conclusons and perspectves are provded n secton V. A. New balse proposal II. SYSTEM DESCRIPTION Conventonal balses are located between the rals. They have the form of a rectangular parallelepped, as shown n Fg. 1. The tran, passng over the balse, can brefly exchange nformaton wth the ground, reads ts absolute localzaton from ths track klometer marker and, therefore, compensates for the drft of ts proproceptve localzaton sensors. 34

2 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Volume 3, Issue 1, June 014 range Transsson rates Localzaton accuracy expected 560 kbps Potentally up to several hundred Mbps 0 cm < 10 cm Fg 1. TR-UWB proposed new ralway balse In many ralway systems, these balses consttute the only equpment remanng on the track, between the rals, and t could prove worthwhle to remove ths last equpment n order to facltate track mantenance, for example ral replacements. As presented n Fg. 1, n our proposal, the conventonal balse stuated between the rals s removed and replaced by the new balse, nstalled on a pole, on the sde of the track, and a few meters away. Ths new balse focuses the radofrequency energy cong from the pole transtters to an area stuated over the rals, rght over the removed conventonal balse locaton. Therefore, ths new balse does not nterfere anymore wth track mantenance operatons, but stll develop a maxmum of rado frequency sgnal at ths partcular locaton over the rals. Several transtters are coupled on the pole; three can be seen n Fg. 1 to get a multple source transtter. Ths nsures transtter redundancy as well as, when correctly confgured, space focusng. One sngle recever or tran-balse reader s used, located n front of the tran. Ths confguraton s usually denoted as a Multple Input, three transtters, Sngle Output, one recever, 3x1 system. Ths new balse uses UWB rado assocated wth TR. UWB rado supports hgh data rate short range ground to tran communcaton as well as the tran localzaton process. Localzaton uses tme of flght measurements of the receved UWB sgnals from the balse whch transts ts own absolute localzaton. The TR technque helps focusng balse radaton on a small area over the track, represented by a sphere n Fg. 1. When crossng ths focusng area, the antenna located n front of the tran receves the UWB sgnals and computes the poston to the geo-referenced balse. Ths system would allow the approachng trans to localze themselves wth a very good accuracy. Table 1 establshes a comparson of performance between the communcaton/localzaton systems usng a ralway conventonal balse and the expected performance of our proposed TR-UWB balse. Ths paper evaluates f the expected localzaton accuracy can be effectvely acheved. TABLE 1. Performance Comparson between a Conventonal Balse and The Proposed TR-UWB Balse Conventonal TR-UWB balse balse Operatng frequency MHz tran to ground 4.5 MHz ground to tran 3.1 to 10 GHz Communcaton < 1 m 10 to 100 m Beyond the ralway system, ths combnaton of technques may fnd applcaton n varous areas, such as the detecton and localzaton of persons-through barrers and fndng vctms of accdents, especally n the mountans or n nes. Through-the-Wall sensng also takes advantage of TR technque; ths applcaton s very useful n safety and peace-keepng applcatons [3]. B. Ultra Wdeband rado technque UWB rado s typcally defned as a wreless transsson scheme wth a bandwdth of over 500 MHz, or havng a fractonal bandwdth hgher than 0. [4]. There are bascally two ways of obtanng an UWB sgnal, usng an Orthogonal Frequency-Dvson Multplexng (OFDM) approach or, an Impulse Rado (IR) approach. In our study, the second method s chosen. Ths nvolves the transsson of very short pulses, typcally havng tme duraton of 1 ns or less, therefore, occupyng a very wde frequency spectrum. The Gaussan waveform pulse or ts dervatve are commonly used pulse shapes. Gven these characterstcs, our choce fell on the UWB technology for the followng reasons [5]: - A transsson capacty up to several hundreds of Mbps has been repeatedly acheved; - UWB sgnals have a low probablty of non-ntentonal detecton due to the low power spectral densty (PSD) used; ths property s nterestng for secure operaton of the transport system; - The technque provdes ntrnscally precse localzaton due to the very short pulses used; - The communcaton s robust aganst the multpath due to the very large bandwdth used; - UWB sgnals can share the same frequency bands than other lted bandwdth rado systems, therefore, not nvolvng a dedcated frequency band for the ralway applcaton. The possblty of combnng all these capabltes n a sngle system s a major element that makes UWB a good canddate for communcaton and localzaton n guded transportaton systems. Furthermore, rado localzaton s subject to two major sources of error, the frst beng the lack of lne of sght (LOS) between the transtter and the recever, and the second beng the excessve presence of multpath. The ntroducton of UWB n wreless communcaton has brought mprovements regardng these sources of errors [6]. However, prevous nvestgatons have rased major ssues, such as the complexty of the sgnal processng at the recepton [7], [8]. Therefore, UWB has been assocated wth TR [9], [10], 35

3 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Volume 3, Issue 1, June 014 especally n mult-users communcaton systems, n order to solve part of these problems. The applcaton of the UWB rado technque to transport s a more recent topc that s thoroughly researched [11] consderng several factors ncludng: - The nature of the transport propagaton envronment; - The use of adequate transttng/recevng duty cycle (LDC); - The number of collocated used UWB devces; - The types and levels of nterference. Regulaton Bodes have consdered these ralway applcatons and Table provdes some nputs regardng areas of operaton of UWB systems [1]. TABLE. Crtcal Factors Ltng the Performance of UWB Systems n Ralway Envronment the tme reversed verson of h(; y( the receved sgnal wthout TR and yrt(, the receved sgnal wth TR at the recever; one has: y s( h( n( (1) y RT s( h*( h( n( () Where represents the convoluton operaton and n( s the Gaussan nose. From Eq. (), we deduce the equvalent mpulse response ( whch corresponds to the autocorrelaton functon of the channel: h*( h( (3) C. Tme Reversal Classcally, TR has been appled to acoustcs and underwater systems [13]. TR s closely related to the retro-drectve array n crowave and phase conjugaton n optcs [14]. The frst TR experment usng electromagnetc waves n the.45 GHz band was reported by [15]. Ths contrbuton suggests that the technques developed for ultrasound ght also be used for the study of electromagnetc case. Indeed, t s an nterestng challenge because n many real envronments lke buldngs and confned areas, crowaves, usng wavelengths between 5 and 30 cm, are scattered off by objects such as walls, desks, vehcles and so on, whch produce a multtude of rado communcaton paths from the transtter to the recever. In such stuatons, a TR system should be able not only to compensate for the multpath effect, but also to mprove rado communcaton parameters by takng advantage of the energy dstrbuted n the reflected sgnals [16]. Usually, the followng TR process s used. Frstly, the channel mpulse response (CIR) s measured between the transtter (Tx) and the recever (Rx) and the correspondng Channel State Informaton (CSI) s then loaded nto Tx. Secondly, the selected sgnal and the mpulse response are reversed n tme and transtted by Tx n the propagaton channel, up to Rx. Ths process, represented n Fg., can be mathematcally descrbed by notng s( the transtted pulse, h( the complex mpulse response of the channel and h*(- the complex conjugate of Fg. Prncple of Tme Reversal technque Appled to our ralway balse, ths TR general process becomes the followng one. The local CSI between any balse transtter source and the defned focusng defned area, between the rals, s measured or computed a sngle tme durng the nstallaton phase of the balse. Ths CSI nformaton s then loaded n the transtter equpment to perform the TR operaton. As long as the propagaton envronment remans unmodfed, ths ntal CSI s repettvely used by the balse. Ths nformaton s then ntroduced as pre-flterng data n the dfferent UWB transtters. Therefore, focusng s obtaned n the requred drecton, between the rals, potentally mprovng the absolute localzaton process. A. 1 Parameters to evaluate TR effectveness Temporal focusng (TF) that can be observed at the recever n Fg. 1, and Spatal Focusng (SF) are characterstcs assocated to TR. To study TF, one can evaluate the Focusng Gan (FG), whch s defned as the rato of the spectrum power of strongest ampltude peak n TR receved, to the strongest peak receved by a conventonal UWB system. The focusng gan can be wrtten as: max( yrt ) FG[ db] 0log10( ) (4) max( y( ) Snce the sgnal level s ncreased n the recevng area, hgher FG could potentally translate nto hgher communcaton range and hgher precson of localzaton. As an example, the study of SF consderng a smple transtter to recever confguraton s performed the followng way. The channel mpulse response (CIR) of the ntended recever located n poston p0 s noted h(p0,. The CIR of the unntended recever located n poston p( 0) s noted h(p, 0). The equvalent CIR of the ntended recever 36

4 s then gven by: ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) ( t Volume 3, Issue 1, June 014 p, h *( p, h( p, ) (5) - * h (-: the conjugate of the CIR correspondng to the th staton. Expresson of the mpulse response beng gven by: Whle the equvalent mpulse response of the unntended recever s gven by: ( p, 0, h *( p0, h( p, (6) SF s then evaluated as the rato of the strongest peak power receved by the ntended recever to the strongest peak receved by the unntended recever. The SF parameter can be wrtten as: max( ( p0, ) SF[ db ] 0log 10 ( ) (7) max( ( p, ) 1 III. EVALUATION OF THE TR CHARACTERISTICS IN A MULTI-ANTENNA CONFIGURATION In Fg. 1, the ralway balse uses a 3x1 confguraton. We evaluate the contrbuton of TR n ths confguraton usng the Power Delay Profle (PDP) and then computng FG and SF. Frstly, ther expressons are deterned n the general case of a nx1 confguraton then, the analytcal and smulaton results are presented for cases x1 and 3x1. The study of PDP and FG s performed for channel models explotng successvely the ray channel approach and the IEEE a channel model, the latter s based on the Saleh Valenzuela formalsm [17]. The ray channel model s presented n [11]. It s by consderng a transtter (Tx) to recever (Rx) dstance d 0. The propagaton doman s bounded by a frst horzontal surface, nfnte, homogeneous and perfectly smooth wth a perttvty contrast (Fg. 3). The sgnals from Tx to Rx undergo reflectons on the floor and celng, except n the case of the drect path. An analytcal computaton of all these rays can be performed usng some geometrcal consderatons. A. Case of ray model Fg 3. Ray Channel Model Throughout the development, we denote by: - S (: the sgnal transtted by the th staton; - h (: s the CIR correspondng to the th staton; h m 0 s ( t t ) (8) Where m represents the number of paths of the channel, the correspondng ampltude and tme of arrval of paths. The general expresson for the equvalent mpulse response s gven by Eq. 9: N t 1 h h ( dt (9) Where N t s the number of transt antennas Replacng the expresson of h ( n Eq. 9, we get: Where N t s ( t t ) s ( t t 1 m0 N t 1 m0 0 s ( t t ) s ( t t N t N 1 1 m0 ) dt ) dt (10) s s ( t s ( t 0 s ( t ) dt The PDP s then gven by: PDP N t N 1 TR UWB ( 1 m0 (11) s Slarly, FG s obtaned by consderng the power peaks n the case of UWB wthout TR ( PDP ( ) and n the case of UWB wth RT ( PDPTR UWB( ): FG[ db ] 10Log 10 Nt N 1 1 m0 Nt 0 1 (1) Fg. 4 (a, b) and Fg. 5 (a, b) llustrate, respectvely, the PDP TR-UWB of the x1 and 3x1 confguratons consderng and 10 paths. We observe the strong TF effect. We also note that the ampltude of the PDP ncreases wth the number of paths or the number of antennas. The peak power obtaned n the case of x 1 s [V ] for the path channel model, ncreasng to 0.17 [V ] for the 10 path channel model. In the case of the 3x1, the respectve peak values of the PDP are stll hgher, 0.08 [V ] and 0.45 [V ]. 37

5 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Volume 3, Issue 1, June 014 ======================================================================================== Fg 4. PDP TR-UWB for x 1 confg., a) for path model, b) for 10 path model Fg 5. PDP TR-UWB for 3x1 confg., a) path model, b) 10 path model ======================================================================================== Table 3 and table 4 represent the focusng gan n the case of x1 and 3x1 respectvely. Ths focusng gan IV. CASE OF IEEE A CHANNEL MODEL s evaluated usng successvely, 4, 6 and 10 paths. We note In the case of the IEEE a model, the characterstc that the focusng gan ncreases wth the complexty of the values of Nakaga-m were explored, takng nto account the channel and the number of transtters. Indeed, from to 10 number of antennas N t. paths, the focusng gan ncreases from 6.0 db (5.9 db analytcal) to 1.6 db (1.5 db analytcal), n the case of Expresson of the CIR s equvalent gven by Eq. 13: N x1 confguraton, and from 9. db (9. db analytcal) t to 16.5 db (16.5 db analytcal), n the case of 3x1 E (13) 1 confguraton. TABLE 3. FG n the Case Of x1 Confguraton Usng, Where 4, 6 And 10 Paths (Analytcal and Smulaton) Ray model paths 4 paths 6 paths 10 paths FG analytcal FG smulaton TABLE 4. FG IN THE CASE OF 3X1 CONFIGURATION USING, 4, 6 AND 10 PATHS (ANALYTICAL AND SIMULATION) Ray model paths 4 paths 6 paths 10 paths FG analytcal FG smulaton s ( t ) s ( t t ) d Then, A m 0 Nt E s( s( t d A 1 m0 Nt E ' s 1 m0 N t E ' s (14) 1 m0 38

6 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Where ' s ( t ) s ( t t ) d s A Volume 3, Issue 1, June 014 The calculaton of the average energy of the CIR n generc nterval W [ a, b] (a and b are arbtrary chosen) provdes: E Pg dt (15) I W W Where I W s the random contanng the multpath components. The varance of the energy functon of the CIR s gven by: Var Rg dt (16) I W W Where R g ( s the kurtoss of the delay profle. Explotng Eq. 15 and 16, the expresson ( becomes: N t g 1 E ' (17) s The correspondng ( ( ) s gven by: PDP ULB RT ( ) PDPTR UWB E t (18) After development, Eq. 18 becomes: N E t g PDP TR UWB E g ' s * (19) 1 rms Where, 1 1 (1 ' ) c1 exp( t / ) (1 ) ' s rms s m' and, c1 ' s( s( dd, s A' s the normalzed autocorrelaton of ' ' s ' s (0) 1 The correspondng focusng gan s gven by Eq. 0. Nt Eg rms c1 c 1 (0) FG[ db] 10Log10 Nt Where, 1 1 c (1 ), m ' represents the Nakaga-m m' value. The values obtaned on the evaluaton of focusng gan for x1 and 3x1 confguratons are respectvely presented n table 5 and table 6. These values correspond to the dfferent IEEE a Channel Model confguratons known as CM1, CM and CM3 correspondng to dfferent ncreasng channel complextes. The same observatons are found usng ths IEEE model than usng our prevous ray channel model. Indeed, takng the example of 3x1 confguraton, changng from CM1 to CM3, FG ncreases from 14.6 db to 0.1 db. The focusng gan s also greater n s the 3x1 confguraton as compared to the x1 confguraton. TABLE 5. FG n Analytcal and Smulaton Study, Case Of x1 (IEEE a Channel Model) IEEE a channel CM CM CM3 model 1 FG analytcal FG smulaton TABLE 6. FG IN ANALYTICAL AND SIMULATION STUDY, CASE OF 3X1 (IEEE A CHANNEL MODEL) IEEE a channel model CM1 CM CM 3 FG analytcal FG smulaton V. EXPERIMENTAL VALIDATION The purpose of ths expermental valdaton s to assess the mpact of envronmental complexty on performance related to temporal/spatal focusng and postonng error, and to compare these conclusons to our precedng smulaton results. A.1. Expermental up An Arbtrary Waveform Generator (AWG) assocated wth a fast samplng osclloscope (TDS) s used. These equpment have dfferent avalable ports that can be used to respectvely generate and acqure sgnals. The pulses generated by the AWG are radated usng wdeband horn antennas. Slar antennas are used for recevng; ther outputs are connected to the TDS ports through low nose amplfers (LNA). A portable computer s used to process the sgnals and store the results. We consder two types of envronment: an anechoc chamber envronment and an ndoor envronment. In the anechoc chamber, metallc reflectors are ntroduced to create, on demand, dfferent confguratons of multpath. In each type of envronment, dfferent geometrcal confguratons (SISO or ) and dfferent locatons of the antennas were tested. Therefore, dfferent s of parameters are mplemented to verfy the mpact of these very dfferent propagaton channels on system performance. A.1.1. Anechoc chamber envronment The dmenson of the anechoc chamber we used s 7 x 7 x 3 m, t s operatng from 100 MHz to 10 GHz. Three antennal confguratons are consdered: A SISO confguraton conssts of a sngle transttng antenna and a sngle recevng antenna; A x1 confguraton conssts of two transttng antennas and one recevng antenna; A 3x1 confguraton conssts of three transttng antennas and one recevng antenna. For each type of confguraton, fve cases are consdered: the anechoc chamber envronment type s consdered as t, 39

7 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Volume 3, Issue 1, June 014 wthout addton of metal reflectors ; A sngle alunum plate ( m x 1 m) s ntroduced as a reflector between transtters and recever to generate a frst multpath confguraton; Two reflector are ntroduced to ncrease the number of reflected sgnals; Three are nstalled to further ncrease the number of reflectons n the propagaton envronment; Four are present to maxze the propagaton envronment complexty. Fg. 6 presents a vew correspondng to ths thrd confguraton. Fg 6. Implementaton of thrd confguraton (presence of two reflector n SISO) A.1.. Indoor envronment We also consder the ndoor complex envronment shown n Fg. 7. Ths envronment looks lke a lted secton of tunnel and s subject to multpath propagaton. Ths area presents ts own reflectons and, therefore, do not necesstate addng supplementary reflectors. Confguratons SISO, x1 and 3x1 are also selected to evaluate the TR-UWB localzaton system. the dstance between the transttng antenna and the recevng antenna s 5 m, whch s representatve of the ralway balse applcaton. The x1 confguraton s so as to obtan a dstance of 5 m between Tx1 and Rx, and 3.6 m between Tx and Rx. 3x1 confguraton corresponds to the addton of a thrd transttng antenna usng a dstance of 4 m between Tx3 and Rx. In a frst step, for each selected confguraton, a pulse s transtted usng the AWG; the receved sgnal s acqured by the TDS, and then returned temporally. In the cases of x1 and 3x1, each Tx re-ets ts correspondng reversed n tme sgnal. Fg. 8a shows, n the SISO confguraton, an example of the receved sgnal wthout TR. Ths confguraton uses three reflector. By comparng t wth the correspondng TR sgnals presented n Fg. 8b, we observe a sgnfcant ncrease n the ampltude of the receved sgnal. To assess ths temporal focusng, we calculate the focusng gan (FG) obtaned n each case. The overall results are grouped n table 7. We obtan that, for each of the three confguratons, the focusng gan ncreases wth the number of reflector ntroduced. For example, n the case of SISO confguraton, FG ncreased from. db wth one reflectng plate up to 6.1 db usng four reflector. Moreover, by comparng the values of FG for these three types of confguraton, we fnd that FG ncreases from SISO to confguraton. Consderng the four reflector scenaro, FG s 6.1 db n the SISO confguraton, ncreasng to 9.8 db n the x1 confguraton and reachng 1.8 db n the 3x1 confguraton. These results confrm the beneft of a hgher complexty of the propagaton envronment when usng TR. It also demonstrates the nterest of usng a mult-antenna confguraton. They are n good accordance wth our smulaton results. Fg 7. Indoor envronment (tunnel type) A.. TR focusng effect expermental evaluaton In ths secton, we expermentally evaluate the TR focusng effect before measurng ts mpact n terms of postonng errors. A..1. TR focusng n an anechoc chamber envronment Our objectve s to evaluate the focusng gan (FG), on the one hand, as the complexty of the propagaton channel ncreases, and, on the other hand, as the antenna confguraton evolves from SISO to 3x1. For the SISO confguraton, (a) Fg 8. Receved sgnal wthout TR (case of SISO confguraton, wth 3 reflector ), b) Receved sgnal wth TR (case of SISO confguraton, wth three reflector ) TABLE 7. FOCUSING GAIN (FG) ACCORDING TO THE NUMBER OF REFLECTOR PLATES INSERTED IN THE PROPAGATION ENVIRONMENT (CASE OF SISO, X1 AND 3X1 CONFIGURATIONS) Confguraton SISO x1 3x1 FG (wthout reflector) FG (1 reflector plate) (b) 40

8 FG ( reflector ) FG (3 reflector ) FG (4 reflector ) ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Volume 3, Issue 1, June To evaluate SF, we consder the scenaro usng three reflector. For the three SISO, x1 and 3x1 confguratons, the recever s moved by 10 cm from ts ntal poston. We note the ntal poston p0 and p1 the poston after dsplacement. Fg. 9 represents an llustraton of the SISO expermentaton. The sgnal prevously reversed n tme at poston p0 s transtted and ths sgnal s now receved at poston p1, where the used CSI s no more optmal. Fg 9. Prncple of SF measurement Fg. 10a and Fg. 10b show an example of the receved sgnals after tme reversal for varous postons p0 and p1 n a SISO confguraton usng three reflector. We note that, from poston p0 to poston p1, a loss of focalzaton s obtaned. We then evaluate SF obtaned at poston p0, compared to SF obtaned at poston p1, usng our three confguratons and fve reflector scenaros. The results are summarzed n table 8. By makng a comparson between, on the one hand, the fve consdered reflector cases, on the other hand, the three types of confguraton, we note that SF values ncrease wth the number of reflectors, but also as a functon of number of transttng antennas. Ths confrms, the results obtaned n theory and smulaton. (a) (b) Fg 10. a) Receved sgnal at target poston p 0 (SISO confguraton wth reflector ); b) Receved sgnal at nterferng poston p 1 (SISO confguraton wth reflector ) TABLE 8. SF Accordng to the Number of Reflector Plates Inserted n the Propagaton Envronment Confguraton SISO x1 3x1 SF (wthout reflector) SF (1 reflector plate) SF ( reflector ) SF (3 reflector ) SF (4 reflector ) A... TR focusng n the ndoor envronment The expermental up, ntally nstalled n the anechoc chamber, s now transferred to the ndoor envronment seen n Fg. 7. We consder our three SISO, x1 and 3x1 confguratons and ths partcular ndoor propagaton envronment. We performed the same measurements as n anechoc chamber to evaluate SF and TF. Fg. 11 (a, b) present respectvely, the receved sgnals wthout TR and wth TR for SISO confguraton. By comparng the receved sgnals, we observe agan the phenomenon of TF. Furthermore, by comparng the ampltudes of sgnals receved wth the three confguratons, one can observe that the largest ampltude s obtaned wth the 3x1 confguraton, the x1 confguraton n turn presents a stronger ampltude of the receved sgnal relatve to SISO confguraton. Fg 11. a) Receved sgnal wthout TR, b) Receved sgnal wth TR (SISO confguraton) These results renforced those already obtaned on the assessment of FG and presented n Table 6. In the latter, we notce the ncreasng FG values accordng to the number of transttng antennas used. Next, evaluaton of the spatal focusng on the target poston p 0 compared to p 1 poston s performed usng our three types of confguratons. Fg. 1 (a, b) respectvely show the sgnals receved at postons p 0 and p 1, n the case of SISO confguraton. We also fnd a loss of focusng passng from poston p 0 to poston p 1. Table 10 gves the values of SF for the three types of confguratons, correspondng to the dsplacement relatve to p 1 versus p 0. 41

9 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) TABLE 9. FOCUSING GAIN (FG) IN INDOOR ENVIRONMENT (CASE OF: SISO, X1 AND 3X1) Confguraton SISO x1 3x1 FG (Laboratory type) FG (Tunnel type) Fg 1. a) Receved sgnal wth TR at the target poston p 0, b) Receved sgnal wth TR at nterferng poston p 1 (SISO confguraton) TABLE 10. SPATIAL FOCUSING IN INDOOR ENVIRONMENT (CASE OF: SISO, X1 AND 3X1) Confguraton SISO x1 3x1 SF (Laboratory type) SF (Tunnel type) Volume 3, Issue 1, June 014 pulses s transtted by each staton to the recevng staton. Each transtter has ts own coded sgnal. The receved sgnals are acqured by the TDS. Post processng evaluates the arrval tmes of the varous sgnals usng a Tme Dfference of Arrval algorthm (TDOA). Then, the Chan localzaton algorthm s appled to deterne the recever poston [19]. - For the TR-UWB system, each transttng staton sends frst a sgnal whch s acqured by the osclloscope and respectvely sent back to the transtters. The tme reversed sgnals are transtted from each base staton to the recever. The receved sgnals are then acqured n ths TR condton. Then, the same post processng than n the UWB alone system deternes the recever poston. However, n ths TR confguraton, the receved sgnals are correlated wth ther TR reference sgnals, and not wth the ntal UWB sequence. Experments were also carred out successvely n the anechoc chamber and n the selected ndoor envronment. A.3.. Evaluaton n the anechoc chamber In the anechoc chamber, we consder our fve reflector confguratons. Three seres of acquston were performed n each case. Fg. 13a and Fg. 13b show an example of the receved sgnals from Tx1 usng three reflector respectvely wthout TR and wth TR. As prevously mentoned, a sequence of seven pulses s sent. A..3. Concluson Ths expermental study was performed to assess the characterstcs of TR n terms of temporal and spatal focusng. The results show that RT s an nterestng canddate for UWB appled to the locaton. Indeed, not only does TR take advantage of the complexty of the propagaton envronment, but also t takes advantage of mult-antenna () to mprove focusng. We also found that the best spato-temporal focusng s obtaned for the 3x1 confguraton. Ths confguraton s representatve of our applcaton as a three transttng sources and a sngle recever are currently consdered. In the next secton, we wll study the contrbuton of TR-UWB n terms of postonng accuracy. (a) (b) Fg 13. a) Receved sgnal n the case of conventonal UWB localzaton system (wthout TR); b) Receved sgnal n the case of TR-UWB localzaton system After processng, we deterne the poston of the moble. Fg. 14 shows the correspondng poston errors. Wth the conventonal UWB localzaton system, we get an error of 11.0 cm, decreasng to 3.3 cm n the case of TR-UWB. A.3. Contrbuton of TR-UWB system n terms of postonng accuracy A.3.1. Objectve In ths last secton, we perform a comparatve evaluaton between a conventonal UWB postonng and a TR-UWB postonng system [18]. To compute the D recever poston nformaton, we need at least three transtters. Thus, we consder our prevous 3x1 confguraton. The operatng prncple s as follows: - In the case of the UWB alone system, a sequence of 7 Fg 14. Conventonal UWB and TR-UWB localzaton systems (Scenaro wth 3 reflector n an anechoc chamber) 4

10 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Volume 3, Issue 1, June 014 To study the performance of the two systems n our varous scenaros, we systematcally deterned poston errors for both UWB and TR-UWB systems. To do ths, three s of acquston have been exploted and processed and poston errors were deterned. The results are reported n table 11 and table 1. TABLE 11. POSITION ERRORS, UWB ALONE Poston error [cm] Poston error [cm] Confguraton Wthout reflector 1 reflector plate reflector 3 reflector 4 reflector Frst Second Thrd TABLE 1.POSITION ERRORS, TR-UWB Confguraton Wthout reflector plate 1 reflector plate reflector 3 reflector 4 reflector Frst Second Thrd The average values obtaned for the three acqustons are reported n table 13 and table 14. Comparng the two systems n the anechoc chamber, wthout reflector, leads to lttle dfference. Even f the TR-UWB system provdes a lower error on localzaton accuracy, the dfference s farly lted. Of course n a non multpath envronment lke the anechoc chamber, TR adds lttle mprovement. For all the other consdered cases, TR-UWB sgnfcantly provdes better localzaton accuracy. TABLE 13. AVERAGE POSITION ERROR CONSIDERING THE 3 ACQUISITIONS, UWB ALONE Number of reflectors Poston error [cm] A.3.3. TABLE 14. AVERAGE LOCALIZATION ERROR CONSIDERING THE 3 ACQUISITIONS, TR-UWB Number of reflectors Poston error [cm] Evaluaton n the ndoor envronment The same expermental measurements were repeated usng an dentcal protocol n the case of the ndoor envronment. Three s of acquston were performed. Ths tme, after processng the receved sgnals, we get a poston error of 1.9 cm for the conventonal UWB system. Ths error decreases to 6.8 cm wth TR-UWB. A.4. Concluson For all the confguratons studed and for the all the complex envronments consdered, we obtaned a better postonng accuracy performance usng the combnaton of UWB and TR as compared to the UWB alone technque. VI. GENERAL CONCLUSION In ths paper, a new system for ralway track-to-tran, spot communcaton was analyzed. The proposed balse smultaneously delvers accurate localzaton nformaton to trans. The equpment can be nstalled on the sde of the track, nstead of beng between the rals. The new balse makes use of the assocaton between ultra wde band rado and tme reversal technque. Analytcal and smulaton studes of ther characterstcs were analyzed. Measurements were also performed n two very dfferent propagaton envronments,.e. an anechoc chamber usng an added of metallc reflectors and a tunnel lke envronment. In all the smulated and expermented confguratons, t has been shown, on the one hand, tme reversal has major ass to ultra wde band rado n terms of spato-temporal focusng, and, on the other hand, that ths advantage s transferred on the applcaton to the localzaton. The results obtaned allow us to conclude that n all the consdered cases, the trackng system provded by tme reversal-ultra wde band rado gves better performance compared to ultra wde band rado alone. Usng tme reversal, a poston error below the 10 cm, objectve requested for ths ralway applcaton, was acheved n all cases. REFERENCES [1] F. Barber, M. Abrl, M. A. Saldo, L. P. Ingolott, P. Tormos, A. Lova, Survey of automated Systems for Ralway Management. Techncal report, Unversdad Poltécnca de Valenca, Department of Informaton Systems and Computaton, 007. [] D. Ke-Ln, M. N. Swamy, Wreless Communcaton Systems, ISBN-13: , pp. 100, Aprl 15, 010. [3] N. Maaref, P. Mllot, X. Ferréres, C. Pchot, O. Pcon, Electromagnetc Imagng Methode Based On Tme Reversal Processng Appled to Through the Wall Target Localzaton, 43

11 ISSN: ISO 9001:008 Certfed Internatonal Journal of Engneerng and Innovatve Technology (IJEIT) Volume 3, Issue 1, June 014 Progress In Electromagnetcs research M, Vol. 1, pp , 008. [4] A. F. Molsh, Ultrawdeband Propagaton Channel-theory, Measurement, and Modellng, Vehcular Technology, IEEE Transacton, Vol. 54, No. 5, pp , 005. [5] M.L. Welbom, "System Consderatons for Ultra-Wdeband Wreless Networks," IEEE: Rado and Wreless Conference RAWCON001, Boston, Aug , pp.5-8. [6] H. El-Salab, P. Krts,, A. Paulraj, G. Papancolaou, Expermental Investgaton of Tme Reversal Precodng for Space-Tme Focusng n Wreless Communcatons, Supported by ONR grant Number N [7] D. Porcno, W.Hrt, Ultra-wdeband rado technology: poten-tal and challenges ahead, IEEE Communcatons Magazne, Vol.41, No. 7, pp , Jullet 003. [8] W. Suwansantsuk, M. Z. Wn, L. A. Shepp On the performance of wde-bandwdth sgnal acquston n dense multpath channels, IEEE Trans. Veh. Technol., vol. 54, No. 5, Sept. 005, pp [9] L. G. Van Atta, Electromagnetc Reflector,. U.S. Patent , [10] M. Fnk, Ondes et renversement du temps, Bulletn de l unon des professeurs de physque et de che, 005, pp [11] H. Saghr, M. Heddebaut, F. Elbahhar, A. Rvenq, J.M. Rouvaen, Tme Reversal UWB Wreless Communcaton-Based tran control n tunnel, Journal of communcatons, vol. 4, No. 4, May 009, pp [1] ETSI TR , Electromagnetc compatblty and Rado Spectrum Matters (ERM); Short Range Devces (SRD); UWB locaton trakng devces n the ralroad envronment, Techncal Report V1.1.1, October 01. [13] A. Derode, P. Roux and M. Fnk, Robust acoustc tme reversal wth hgh order multple scatterng, Physcal revew letters, Vol. 75, No. 3, 1995, pp [14] B. E. Henty and D.D. Stancl, Multpath-Enabled Super-Resoluton for RF and Mcrowave Communcaton usng Phase-conjugate Arrays, Phy-Rev. Lett, Vol.93, No.4, 004, pp [15] G. Lerosey, J. de Rosny, A. Tourn, A. Derode, G, Montaldo, M. Fnk, Tme Reversal of Electromagnetc waves, Physcal revew letters, Vol. 9, No , pp [16] X. Lu, B.-Z. Wang, S. Xao, J. Deng, Performance of Impulse Rado UWB Communcaton Based On Tme Reversal Technque, Progress In Electromagnetcs Research, PIER Vol.79, No. 11, 008, pp [17] D. Abass-Moghadam, D. Tabataba Vakl, Channel characterzaton of tme reversal UWB communcaton systems, Wley Internatonal Journal of Communcaton Systems, Vol. 65, No.9-10, 010, pp [18] B. Fall, M. F. Elbahhar, M. Heddebaut, A. Rvenq, Tme-Reversal UWB Postonng Beacon for Ralway Applcaton, Internatonal Conference on Indoor Postonng and Indoor Navgaton (IPIN), IEEE Proceedng, Sydney (Australa), 01, pp.1-8. [19] Y.T. Chan, A Smple and Effcent Estmator For Hyperbolc Locaton, IEEE Transactons on Sgnal Processng, AUTHOR S PROFILE Bouna Fall was born n He receved the M.S and Ph.D. degrees from the Unversty of Valencennes (France) n 009 and 013, respectvely. Currently he s assstant research at ths unversty. Hs prmary nterest s n sgnal processng especally Ultra Wdeband technology and Tme reversal technques appled to communcaton and localzaton systems. Fouza Elbahhar Boukour was born n She receved the M.S and Ph.D. degrees from the Unversty of Valencennes (France) n 000 and 003, respectvely. She s now employed as researcher at IFSTTAR/LEOST, Vlleneuve d Ascq, France. She s nvolved n sgnal processng especally Ultra Wdeband technology. Her major research nterests are sgnal processng localzaton and communcaton technques especally for land transportaton lke communcaton Vehcle to Vehcle and vehcle to nfrastructure and localzaton systems. Marc Heddebaut was born In Soman, France, n He receved the M.S. and Ph.D. degrees n electroncs from the Unversty of Llle France n 1980 and 1983, respectvely. He joned the French Natonal Insttute for Transportaton and Safety Research (INRETS now IFSTTAR) n 1983 and became A senor researcher n Snce 1979, he has been workng n the feld of land moble communcaton and electromagnetc compatblty. Hs prmary nterests nclude telecommuncaton and radar systems dedcated to land transport, EMC, moble localzaton and command control of automated vehcles. Atka Rvenq-Menhaj was born n She receved her Dploma of Engneerng and the M.S j degree n 1993 and then her Ph.D. degree n1996, from the Unversty of Valencennes (France). She s now Professor n electroncs at ths unversty. Her prmary nterest s n sgnal processng appled to ntellgent transportaton systems and telecommuncaton systems 44