Noncoherent Detection of Satellite-Based AIS Signal with Phase Offset Compensation

Size: px

Start display at page:

Download "Noncoherent Detection of Satellite-Based AIS Signal with Phase Offset Compensation"

Martin Glenn
6 years ago
Views:

Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp. 128-134 Sesors & Trasducers 2014 by IFSA Publishig S. L. http://www.sesorsportal.

1 Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp Sesors & Trasducers 2014 by IFSA Publishig S. L. Nocoheret Detectio of Satellite-Based AIS Sigal with Phase Offset Compesatio 1 2 Shexiag MA 1 2 Tig FANG 1 Key Laboratory of Computer Visio ad System Miistry of Educatio Tiaji Uiversity of Techology Tiaji Chia 2 Tiaji Key Laboratory of Itelligece Computig ad Novel Software Techology Tiaji Uiversity of Techology Tiaji Chia 1 Tel.: masx_tjut@126.com Received: 20 December 2013 /Accepted: 28 February 2014 /Published: 31 March 2014 Abstract: Cosiderig the existece of large Doppler frequecy shift fast time-varyig chaels ad multiple user iterferece ad other serious problems i the satellite-based AIS system a ocoheret detectio method called phase adaptive algorithm is proposed. For AIS sigals utilizig the characteristic of the eergy distributio uder the Lauret decompositio of PAM waveform matched filter is evaluated without whiteig filter o the mai compoet. This ca effectively avoid the ifluece of the correlatio legth o the performace of detectio. Ad the complexity of the system is sigificatly reduced. The cost fuctio of decodig is give based o maximum likelihood criterio i the algorithm. Multiple symbols as the observatio widow are decoded at the same time. Meawhile phase adaptive compesatio is employed i the decodig process. Simulatio results show that this method has good detectio performace. The is close to the ideal coheret detectio whe the size of observatio widow is 10 symbols ad is superior to the traditioal differetial decodig. At the same time the algorithm has a higher robust ad is also applicable to the coditios of multiple user co-chael iterferece. Copyright 2014 IFSA Publishig S. L. Keywords: Satellite-based AIS Phase adaptive Nocoheret detectio Phase offset Compesatio. 1. Itroductio Automatic Idetificatio System is a ew aid to avigatio system which is joitly promoted by the Iteratioal Maritime Orgaizatio the Iteratioal Telecommuicatio Uio ad the Iteratioal Associatio of Lighthouse Authorities i It has the fuctio of automatic ship idetificatio commuicatios avigatio ad so o. With the rapid developmet of ecoomic globalizatio i the today s world the origial berthbased ad shore-based AIS system have bee uable to meet the eed of moitorig withi the scope of global ocea. Satellite-based AIS system more ad more gets the attetio of the coutries i recet years due to its wide coverage. Traditioal receiver geerally ca be divided ito two major categories of coheret ad o-coheret. For ideal AIS sigal detectio correlatio detectio performace is best but the Satellite-based AIS system [1] has the characteristics of larger Doppler frequecy shift faster time varyig chael much serious multiuser iterferece. Hece it s harder to estimate the parameters of carrier phase offset Doppler frequecy offset ad chael delay state so that the coheret detectio become more complex. At 128 Article umber P_1915

2 Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp this time how to fid a way to reduce the system complexity with miimal performace loss becomes more ad more importat. Sigle symbol differece detectio techology is studied i the literature [2] eve though the complexity was lower but i terms of performace loss was very large. The multiple symbols differetial detectio ad associated demodulatio techique are studied i the literature [3 4] respectively. Both of them have improved the performace of detectio by icreasig the complexity of system. A o-coheret detectio method based o maximum likelihood Viterbi decodig has bee proposed i the literature [5 6]. Although this method makes full use of the correlatio betwee the cotiuous phase sigal symbol i the process of judgmet ad obtais very good results i terms of performace but the algorithm has much larger complexity because of the use of Viterbi decodig so that it ca t meet the timeliess of the satellite receiver. For the sigals of the satellite-based AIS are always mixed sigals the method of spectral aalysis for blid separatio of mixed sigals was proposed i the literature [7]. But the method is oly applicable to each sigal carrier has larger frequecy differece so that it ca t be applied to the Satellite-Based AIS system. Therefore a o-coheret detectio method for joit decodig ad phase of a ew decodig phase adaptive compesatio is proposed i this paper with the purpose of obtaiig good detectio performace sigificatly reducig the decodig process of complexity ad beig applied to complex coditios with multiple co-chael satellite-based AIS sigals. 2. Sigal Model Let s() t be the complex evelope of the AIS sigal waveform give by j ( t; ) st () = e φ α (1) φ( t α) = π aq( t it) (2) ; i i is the iformatio bearig phase a = { a } are idepedet data symbols takig o the values { 1 1} with equal probability. T is the symbol period ad q() t is the phase pulse t () ( τ) q t = g dτ (3) 1 2π B L+ 1 g() t = Q t T 2T l 2 2 2π B L 1 Q t + T l 2 2 (4) The frequecy pulse g () t is limited to the iterval (0 LT) L is a iteger B is 3 db badwidth ad Q() t is defied as 1 2 /2 Qt () = τ e dτ (5) π 2 t Accordig to the characteristics of the cotiuous phase modulatio a satellite-based AIS sigal after Lauret decompositio ca be expressed as 1 Q 2 L = amplitude pulse modulatio amely PAM waveforms superpositio k k( ) (6) s(; ta) = b c t T k= 0 This paper maily aalysis the frequecy estimatio of the satellite-based AIS sigals o Gaussia chaels ad the complex evelope of the received sigal is modeled as () j( 2πvt+ θ) ( τ ) ( ) r t = e s t + w t (7) v ad θ represet the frequecy offset ad the carrier phase respectively τ is the timig epoch ad wt () is a complex-valued white Gaussia oise process. Let x(t) be the filter output give by () j( 2πvt+ θ) ( τ ) ( ) x t = e s t + t (8) t () is filtered Gaussia oise. 3. Phase Adaptive Nocoheret Detectio Algorithm 3.1. Phase Compesatio i Detectio Based o maximum likelihood (ML) sequece detectio the likelihood fuctio of coheret detectio ad sychroizatio give i the literature [8] is as follows: ( ) Re T0 /2 { ( ) ( ) j a θ θ r t s t a e dt T /2 } Λ = (9) Next substitutig (6) ito (9) yields Lettig 0 j θ Λ ( a θ ) = Ree b k k= 0 r( t) c t T dt T0 /2 T0 /2 k ( ) } k( ) () ( ) + (10) xk = r() t c t T dt (11) = r t c t k t= T 129

3 Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp The equatio (10) ca be writte as: j θ Λ ( a θ ) = Ree xk b k (12) k= 0 I coheret detectio estimates the exact phase iformatio firstly ad the detected the sigal. That is phase estimatio is separated from sigal detectio process. Ad whe the sigal frequecy estimatio is ot iaccurate the detectio performace will decrease sigificatly. This paper presets a detectio algorithm which combied phase compesatio ad decodig. The cost fuctio is expressed i the form jarg( x 1 k b ) Q k k 0 = Λ ( a ) = Re e xk b k (13) k= 0 At this time compared the literature [12] to the origial formula we have phase compesatio i the detectio but it is still differet from the method of the phase referece compesatio i [9 10] i the literature [9 10] they use the method of the judgmet cotrol lockig phase rig. I this paper we avoids tedious PLL desig ad proposes the method of decodig phase adaptive this algorithm combies the process phase compesatio with path selectio phase compesatio ad detectio ru at the same time phase compesatio varies greatly before ad after the path selectio ad this greatly reduces the complexity. The BT of the satellite-base sigal is 0.5. For biary GMSK sigals the eergy is maily cocetrated i the zero order ( k = 0 ) Lauret expasio that is the part of the pricipal compoet whe BT is bigger. So the satellite-base sigal ca be expressed as 0 0 ( ) (14) s(; ta) b c t T b 0 α 1 jπ h m m= = e (15) By formula (1) ad (3) i order to obtai the symbol iformatio sectio of chael trasmissio model of type (1) ad to elimiate the basis fuctio we use matchig filterig to the type (1) ad we obtai () ( ) x = r t c t 0 o t= T L jθ = e glb0 k l + 0 l= L jθ = s0 e + 0 (16) For the coveiece we record x 0 s 0 b 0 0 separately as x s b so the cost fuctio of the o-coheret detectio (12) also be simplified by jarg( xk b k ) Λ ( a ) = Re e xk b k (17) 3.2. Block Detectio Block detectio [11] is a method of multiple symbols detectio ad multiple symbol judgmet based o the maximum likelihood criterio. Namely the received sigals is divided ito blocks which cotai N symbols accordig to the observatio time legth NT the we make related operatios with all sigals may be set block by block ad use the relevat results as a criterio to choose the maximum oe as the output. Because the operatio is processig block by block the decisio is a result of N symbols at oe time. This method takes full use of the correlatio amog the symbols of CPM sigal i the judgmet. It greatly reduces the computatio relative to the Viterbi decodig process ad the decodig complexity is close to 1/N of the Viterbi algorithm which is because Viterbi decodig method oly makes oe symbol decisio at oe time. We defie the states i the decodig as { } θ N+ 1 1 S = a a a (18) N θ = πh i= 0 a i is the cumulative phase. The modulatio idex is 1/2 so mod ( θ 2 ) four possible states they are π oly has π 3π mod ( θ 2π) 0 π 2 2 (19) so there are 4 2 N states for S. Notably we do t eed to kow the exact value of θ i this algorithm supposig θ = 0 so actually there are possible 2 N states i decodig cosiderig the ISI at the same time detectig performace is better whe block N are geerally ot less tha 10. For example whe N = 6 the possible states are show as below. The algorithm uses the same b 0 i every block so that we do ot eed to re-compute it every time as log as storage b 01 b 02 b 0 N + 1 together ad we ca decode the N symbols at oce. 130

4 Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp Table 1. All states table i the decodig process. State umber All states a 5 a 4 a 3 a 2 a 1 a Simulatio Results ad Performace Aalysis The AIS system work i the maritime VHF (VHF) bad with two chaels CH87B (AIS MHz) ad CH88B (AIS MHz) whose symbol rate Rb is 9.6 kbps ad modulatio mode is Gauss miimum shift keyig (GMSK). I the simulatio experimet cosiderig the modulatio feature of AIS sigal both real part ad imagiary part of the complex base-bad sigal use 8 times of symbol rate (76.8 khz) as samplig rate. The iformatio sequece is geerated i radom ad i order to statistic the the legth of the iformatio sequece is more tha 256. Whe decodig we set BT=0.4. Accordig to the problems are faced by the satellite-based AIS receiver the effects of parameters such as frequecy offset phase shift BT delay as well as the decodig performace uder multiple same frequecy iterferece are aalyzed. To be sure the oise i the simulatio experimet is Gauss white oise. The detectio method used i the experimets is block detectio ad the performace has very big relatios with the legth of the block Fig. 1 shows the curve with differet N i this method as ca be see from the graph that with the icrease of N the bit error rate is lower with the same sigal to oise ratio the detectio performace is better. While the umber of iformatio detected a oe-time icreased the umber of the state ad the time complexity is icreased at the same time. I the experimet we take compromise N=10 with which the performace is good ad the decodig time is very short. N=5 N=6 N=7 N=8 N=9 N=10 Fig. 1. Performace of the phase adaptive algorithm with the size of block. Fig. 2 show the detectio rate curve of oe bit differetial detectio [2] two bit differetial detectio [3] ideal coheret detectio [12] ad the detectio method proposed i this paper. We ca see that the bit error rate i this detectio method is close to ideal coheret detectio the detectio performace is better tha oe bit differetial detectio ad two bit differetial detectio ad as the sigal-to-oise ratio icreases the advatage is more obvious. Ideal coheret demodulatio Phase adaptive algorithm 2-bit differece algorithm 1-bit differece algorithm Fig. 2. The performace compariso of four detectio methods. I the respect of ati frequecy offset as we ca see i Fig. 3 that the error rate of coheret detectio growth i the presece of small frequecy offset so it does ot have the character of ati frequecy offset. performace compariso betwee the phase adaptive algorithm ad the method of decisiodirected PLL used i documet [9 10] is show i Fig. 4 we ca see ot oly the complexity of the algorithm is reduced but also the detectio performace is superior. Fig. 5 shows detectio results with a frequecy offset value i 0 Hz 10 Hz 30 Hz 50 Hz from the Fig. 5 we ca be see that 131

5 Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp withi 50 Hz the frequecy offset has little effect o the bit error rate at the same time this simulatio the symbol umber is which is much larger tha a frame data it has better ati frequecy offset performace i the practical applicatio. There is a eed to explai the actual maximum frequecy i the satellite-based AIS receiver is 3.7 KHz we eed to do the blid frequecy offset estimatio ad correctio of the applicatio i this method. Without frequecy offset Frequecy offset 0.01Hz Frequecy offset 0.1Hz Frequecy offset 0.5Hz Fig. 3. of coheret detectio method with small frequecy offset. Without frequecy offset Frequecy offset 10Hz Frequecy offset 30Hz Frequecy offset 50Hz Fig. 5. performace of the phase adaptive algorithm with differet frequecy offset. Without time delay Time delay(1/8tb) Time delay(2/8tb) Time delay(3/8tb) Method of decisio-directed PLL Phase adaptive algorithm Fig. 6. performace of the phase adaptive algorithm with various delays. BT=0.3 BT=0.4 BT=0.5 BT= Fig. 4. performace compariso betwee phase adaptive algorithm ad documet [9 10]. I the respect of ati-delay Fig. 6 shows the detectio performace curves whe the time delay are 0 1/8Tb 2/8Tb ad 3/8Tb we ca see i the iterval of 2/8Tb error rate is still low ad meet the detectio performace requiremets. Fig. 7 show the bit error rate curves whe the sedig sigal i differet BT value from which we ca see that with a larger BT value error rate is small this simplified before is based o the characteristic that satellitebased AIS sigal have large value of BT. Fig. 7. performace of the phase adaptive algorithm with differet values of BT. I order to aalyze the suppressio ability agaist multi-user iterferece cosiderig wave polarizatio ioospheric scitillatio ad other 132

6 Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp factors o the satellite the sigal-to-oise ratio of AIS system o the geeral reach to about 10 db or bigger [13] so it is possible to achieve high sigal to oise ratio of about 20 db 25 db. Fig. 8 ad Fig. 9 show the error curves usig the detectio of this paper with the total sigal to iterferece ratio of S/I=8 db ad S/I=5 db respectively from which we ca see that they ca reach order i a certai sigal-to-oise ratio so the method ca be applied to the AIS receiver system with multiple co-chael iterferece. Without Iterf #1 Iterf #2 Iterf #5 Iterf characteristic of the eergy distributio of AIS sigals uder the Lauret decompositio with larger value of BT we evaluate the matched filter without whiteig filter o the mai compoet the usig block detectio with N symbols ad detect symbols at oe time. Compared to the Viterbi decodig uder the same observatio iterval this method greatly reduced i computatio to approximately to the decodig complexity of Viterbi. The simulatio results show that the effect of AIS sigal i a certai rage of parameters we still achieve good detectio performace at the same time whe there are multiple co-chael iterferece circumstaces we may still get accurate detectio. Whe the total S/I is the detectio performace is still up to about i the high sigal-to-oise ratio which ca fully meets the demad of the satellite-based AIS system. Ackowledgemets This work was supported i part by a grat from the Natioal Natural Sciece Foudatio of Chia (No ) Fig. 8. performace of the phase adaptive algorithm with iterferig sigals total S/I = 8 db. Without Iterf #1 Iterf #2 Iterf #5 Iterf Fig. 9. performace of the phase adaptive algorithm with iterferig sigals total S/I = 5 db. 5. Coclusios For satellite-based the characteristics of AIS receiver sigal this paper presets a ew phase adaptive o-coheret detectio method which jois decodig ad phase compesatio. The algorithm combies the process of phase compesatio ad decodig choose path together phase compesatio ad testig at the same time. Cosiderig the Refereces [1]. Timothy Pratt Charles Bostia Jeremy Allutt et al. Satellite Commuicatios 2 d ed. Joh Wiley & Sos [2]. Su Jihua Li Jia Dog Ji Li-Ju Nocoheret reduced state differetial sequece detectio of cotiuous phase modulatio Joural of Electroics & Iformatio Techology Vol pp [3]. Qiao Zhi Zhou Chuhui Xiao Limi Nocoheret demodulatio of cotiuous phase modulatio sigals based o multi-symbol differetial correlatio Joural of Tsighua Uiversity (Sciece & Techology) Vol. 47 No pp [4]. Zhesheg Zhag Joh Weifield Taru Soi Combied differetial demodulatio schemes for satellite-based AIS with GMSK sigal i Proceedigs of the SPIE 7691 Space Missios ad Techologies Orlado Florida USA 7 May 2010 pp [5]. Giulio Colavolpe Riccardo Raheli Nocoheret sequece detectio of cotiuous phase modulatio IEEE Trasactio o Commuicatios Vol. 47 Issue pp [6]. Paolo Burzigotti Alberto Giesi Giulio Colavolpe Advaced receiver desig for satellite-based automatic idetificatio system sigal detectio Iteratioal Joural of Satellite Commuicatio ad Networkig Vol. 30 Issue pp [7]. G. Gletis A. Jakobsso Time-recursive IAA spectral estimatio IEEE Sigal Processig Letters Vol. 18 Issue pp [8]. J. B. Aderso T. Auli C.-E. Sudberg Digital phase modulatio Pleum New York [9]. G. Colavolpe ad R. Raheli Reduced-complexity detectio ad phase sychroizatio of CPM sigals IEEE Trasactios o Commuicatios Vol. 45 Issue pp

Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp. 128-134 [10]. G. Colavolpe ad R. Raheli Nocoheret sequece detectio of CPM Electroics Letters Vol. 34 No. 3 1998 pp. 259-261. [11]. D. Divsalar M. K.

7 Sesors & Trasducers Vol. 166 Issue 3 March 2014 pp [10]. G. Colavolpe ad R. Raheli Nocoheret sequece detectio of CPM Electroics Letters Vol. 34 No pp [11]. D. Divsalar M. K. Simo Maximum-likelihood block detectio of o-coheret cotiuous phase modulatio IEEE Trasactios o Commuicatios Vol. 41 Issue pp [12]. Zeg Zuo Qi Li Za Performace aalysis ad simulatio of GMSK demodulatio based o Viterbi algorithm Joural of Chogqig Uiversity of Posts ad Telecommuicatios (Natural Sciece Editio) Vol pp [13]. Mu Zhou Alle-Ja va der Vee ad Reé va Leuke Multi-user leo-satelite receiver for robust space detectio of AIS messages i Proceedigs of the IEEE Iteratioal Coferece o Acoustics Speech ad Sigal Processig Kyoto Japa March 2012 pp Copyright Iteratioal Frequecy Sesor Associatio (IFSA) Publishig S. L. All rights reserved. ( 134

Radar emitter recognition method based on AdaBoost and decision tree Tang Xiaojing1, a, Chen Weigao1 and Zhu Weigang1 1

Radar emitter recognition method based on AdaBoost and decision tree Tang Xiaojing1, a, Chen Weigao1 and Zhu Weigang1 1 Advaces i Egieerig Research, volume 8 d Iteratioal Coferece o Automatio, Mechaical Cotrol ad Computatioal Egieerig (AMCCE 7) Radar emitter recogitio method based o AdaBoost ad decisio tree Tag Xiaojig,