Proceedings of Meetings on Acoustics

Transcription

1 Proceedings of Meeings on Acousics Volume 9, 2010 hp://asa.aip.org 159h Meeing Acousical Sociey of America/NOISE-CON 2010 Balimore, Maryland April 2010 Session 2aUW: Underwaer Acousics 2aUW9. An overview of underwaer acousic communicaion via paricle velociy channels: Channel modeling and ransceiver design Ali Abdi*, Huaihai Guo, Aijun Song and Mohsen Badiey *Corresponding auhor s address: Elecrical and Compuer Engineering, New Jersey Insiue of Technology, 323 King Blvd, Newark, NJ 07102, ali.abdi@nji.edu Over he pas few decades, he scalar componen of he acousic field, i.e., he pressure channel, has been exensively used for underwaer acousic communicaion. In recen years, vecor componens of he acousic field, such as he hree componens of acousic paricle velociy, are suggesed for underwaer communicaion. Consequenly, one can use vecor sensors for underwaer communicaion. The small size of vecor sensor arrays is an advanage, compared o pressure sensor arrays commonly used in underwaer acousic communicaion. This is because velociy channels can be measured a a single poin in space. So, each vecor sensor serves as a mulichannel device. This is paricularly useful for compac underwaer plaforms, such as auonomous underwaer vehicles (AUVs). Funded by he Naional Science Foundaion, our research effors focus on he research problems in wo closely-relaed caegories: channel modeling and ransceiver design. Channel modeling research aims a characerizaion of hose aspecs of acousic paricle velociy channels such as delay and Doppler spread, ransmission loss, ec., which deermine he communicaion sysem performance. Transceiver design addresses opimal use of vecor sensors and paricle velociy for daa modulaion and demodulaion, equalizaion, synchronizaion, coding, ec. (work suppored by NSF). Published by he Acousical Sociey of America hrough he American Insiue of Physics 2010 Acousical Sociey of America [DOI: / ] Received 10 May 2010; published 26 May 2010 Proceedings of Meeings on Acousics, Vol. 9, (2010) Page 1

2 I. INTRODUCTION Over 75% of he earh surface is covered wih waer, wih los of resources ha human s life depends on. There are numerous naval and civilian applicaions ha pose an increasing demand on high speed underwaer wireless elemery and daa communicaion among sensors, auonomous underwaer vehicles (AUVs), deepwaer moored insrumens, and surface ships. Examples include communicaion among AUVs for collaboraive operaions, harbor securiy and surveillance sysems, oceanographic daa rerieval from underwaer sensors over geographically large areas, offshore oil and gas exploraions, environmenal monioring, Since waer srongly aenuaes elecromagneic waves, acousic waves are he appropriae carriers of informaion in underwaer channels. For decades, underwaer acousic communicaion sysems have been relying on scalar acousic sensors, which measure only he pressure of he acousic field. The idea in his paper is o ake advanage of he vecor componens of he field, such as he hree componens of paricle velociy, sensed by a vecor sensor, for daa communicaion. Applicaion of vecor sensors for informaion exchange over hese degrees of freedom of he acousic field has he poenial o provide new opporuniies for underwaer acousic communicaion and elemery. II. A SHORT SUMMARY OF UNDERWATER COMMUNICATION TECHNIQUES IN ACOUSTIC PRESSURE CHANNELS In general, underwaer acousic channels are seriously bandwidh-consrained and he bandwidh is range-dependen [1]. For disances from 10 km up o 100 km (long range), he available bandwidh is abou few khz, whereas in a 1-10 km medium range seup, he bandwidh is almos few 10 khz. Only for very shor ranges, say, smaller han 100 m, he bandwidh may exceed 100 khz [2]. The harsh mulipah, wih delay spreads up o several hundreds of symbols for high daa raes, and channel emporal variaions due o he waer surface flucuaions, inernal waves, and urbulence, wih Doppler spreads up o several 10 Hz, furher complicaes underwaer communicaion. Afer he firs generaion of underwaer (analog) modems, second generaion (digial) modems in 80 s used noncoheren echniques such as frequency shif keying (FSK) and differenially coheren schemes like differenial phase shif keying (DPSK), which remained o be he mainsream for a decade [3]. Due o he need for higher specral efficiencies over ypical channels of ineres, coheren sysems wih phase shif keying (PSK) and quadraure ampliude modulaion (QAM) were considered as well, mosly in 90 s. Comprehensive reviews of he pas and presen underwaer communicaion and elemery sysems can be found in [2]-[4]. Spaial diversiy (large arrays of hydrophones) and differen ypes of equalizaion, beamforming, coding, channel esimaion and racking are also used for communicaion in pressure channels [3]. Underwaer mulipleinpu muliple-oupu (MIMO) sysems using spaially separaed pressure sensors are also recenly invesigaed [5]- [16]. Among he four ypes of underwaer signals (conrol, elemery, speech, video) [2], low daa raes migh be enough for conrol signals. However, for real-ime ransmission of elemery, speech, and video signals over medium and long ranges, one needs high daa raes and specral efficiencies, specially for video signals. III. A BRIEF OVERVIEW OF VECTOR SENSORS Acousic vecor sensors can measure non-scalar componens of he acousic field such as he paricle velociy, which canno be sensed by a single scalar pressure sensor. Developmen of vecor sensors may dae back o 30 s [17]. Over he pas few decades, a large volume of research has been conduced on design and performance analysis of vecor sensors [18]. They have been mainly used for arge localizaion and SONAR applicaions. Examples include accurae beamforming and azimuh/elevaion esimaion of a source, avoiding he lefrigh ambiguiy of linear owed arrays of scalar sensors, significan acousic noise reducion due o he highly direcive beam paern, ec. [19]-[22]. There are wo major ypes of vecor sensors: pressuregradien and inerial [18]. Pressure-gradien sensors use a finie-difference approximaion o esimae he gradien of he acousic field such as paricle velociy. Inerial sensors ruly measure he paricle velociy by responding o he acousic paricle moion. Each sensor ype has is own advanages and disadvanages. A brief summary of meris and limiaions of hese sensors can be found in [23]. More deails and informaion on oher ypes of sensors such as mulimode vecor sensors can be found in [18]. IV. DATA COMMUNICATION IN UNDERWATER ACOUSTIC PARTICLE VELOCITY CHANNELS A vecor sensor measures he hree orhogonal componens of acousic paricle velociy, as well as he acousic pressure, all a a single poin in space. From his angle, one can hink of a vecor sensor as a compac mulichannel device ha can used for acousic communicaion. This is paricularly imporan for plaforms such as small AUVs where large size array canno be used. For example, he medium frequency 3 khz receive array of a modem recenly designed for he 21-inch diameer Bluefin AUV consiss of four hydrophones and is 1.5 m long. Quoed from [24] Ideally a longer array would be used, bu his is he larges ha can be easily insalled in he vehicle Even a high frequencies, he size of he array is an issue, if i is supposed o fi ino he smaller inch AUVs [24]. Research on daa communicaion in paricle velociy channels can be caegorized ino wo closely relaed areas: channel modeling and ransceiver design. Channel modeling research aims a undersanding and characerizaion of paricle velociy orhogonal componens as communicaion channels. On he oher hand, ransceiver design research inends o deermine how vecor sensors, along wih proper Proceedings of Meeings on Acousics, Vol. 9, (2010) Page 2

3 communicaion echniques and signal processing algorihms, can be used for successful ransmission and recepion of informaion via paricle velociy channels. Possible differences beween velociy channels and he acousic pressure channel ha are idenified a he channel modeling sage can impose differen consrains on sysem design. This raises he need o develop new communicaion and signal processing soluions, as well as perhaps new sensor ypes and echnologies, for proper ransceiver design. V. TRANSCEIVER DESIGN FOR PARTICLE VELOCITY COMMUNICATION CHANNELS The lessons learned from ransceiver design in he acousic pressure channel serve as useful guidelines for sysem design in paricle velociy channels. However, due o he possible differences beween velociy and pressure channels, a proper design should be able o uilize he special feaures of velociy channels. For example, some possible correlaions among he velociy and pressure channels, or perhaps differen delay spreads need o be included in he sysem design, he rae of change of velociy channels migh be differen and should be considered for sysem performance opimizaion, ec. The schemaic of a basic vecor sensor communicaion sysem is shown in Fig. 1. There is one scalar pressure sensor a he ransmi side which ransmis he signal s(). The receiver is a vecor sensor ha measures he x, y and z componens of paricle velociy, in addiion o he acousic pressure. The measured signals are rx (), ry (), rz () and r (), respecively. The vecor sensor is symbolically shown by hree orhogonal dipoles and he single black do a he cener is he scalar pressure sensor. There are four possibly imevarying channels in he sysem, whose impulse responses are vx (,), vy (,), vz (,) and p(,), as shown in Fig. 1. The firs hree represen x, y and z paricle velociy channels, whereas p(,) sands for he pressure channel. The received signals can be wrien as rx() s() vx(,) nx(), ry() s() vy(,) ny(), (1) rz() s() vz(,) nz(), r () s () p(,) n (), where is he convoluion and nx (), ny (), nz () and n () are noise componens. These noise erms may include ambien noise, flow noise, srucure-borne noise, ec. To recover he ransmied daa s() convolved wih he frequency-selecive channels vx (,), vy (,), vz (,) and p(,), a mulichannel equalizer is needed. A. Mulichannel Equalizaion Using a Vecor Sensor Receiver A mulichannel equalizer is proposed in [23] and [25] o demonsrae he feasibiliy of iner-symbol inerference (ISI) removal using vecor sensors. The vecor sensor receiver shows a compeiive performance, compared o an equivalen array of scalar pressure sensors, as well as a single pressure sensor receiver. The impac of unbalanced noise powers in a vecor sensor receiver is sudied in [25]. Since perfec knowledge of he channels a he receiver is no available, he influence of channel esimaion errors on he performance of a vecor sensor receiver is invesigaed in [23]. Feasibiliy of mulichannel equalizaion using a vecor sensor is demonsraed in [26] using measured daa. B. Signal Transmission and Recepion Using Vecor Sensors Since a vecor sensor has a compac size, plaforms such as small AUVs can significanly benefi from signal ransmission using a vecor sensor, in addiion o signal recepion. A sysem is proposed in [27] where boh ransmier and receiver are vecor sensors. In addiion o paricle velociy channels, paricle acceleraion ype channels are also used in his sysem. C. Muliuser Communicaion Using a Vecor Sensor Since i is demonsraed in [23] and [25] ha a vecor sensor can serve as a mulichannel communicaion receiver for ISI cancellaion in a single user sysem, i can be used in a muliuser sysem as well. Using space-ime block codes, a muliuser sysem is developed in [28], where a compac vecor sensor receiver is used for separaing muliple users and also miigaing ISI for each user. Anoher advanage of he sysem in [28] is ha o accommodae muliple users, i does no increase he bandwidh of users via spread specrum and code division muliple access echniques. This is paricularly imporan in highly bandlimied underwaer channels. VI. PARTICLE VELOCITY COMMUNICATION CHANNEL MODELING Fig. 1. A basic communicaion sysem in paricle velociy and pressure channels. The ransmier is a single scalar pressure sensor whereas he receiver is a four-channel vecor sensor. Underwaer communicaion channels, especially in shallow waers, are sochasic mulipah environmens, due o he random roughness of sea boom and is surface, muliple ineracions of he signal wih surface and he boom, and also because of he presence of urbulence, inernal waves, ec. Proceedings of Meeings on Acousics, Vol. 9, (2010) Page 3

4 Modeling sochasic paricle velociy channels as random linear ime-varying filers is a sysem-heoreic mehodology which provides hose channel characerisics ha direcly affec sysem design and performance. Examples include space-imefrequency correlaion funcions in paricle velociy channels, coherence ime and frequency of such channels, level crossing raes and average fade duraions, ec. All hese are vial for he designer when choosing, say, sufficien elemen spacing in an array, proper number of aps for an equalizer, he updaing rae of he equalizer aps, appropriae coding schemes which can handle deep fades of cerain lenghs, he maximum packe lengh over which he channel says almos consan, In [29], [30] and [31], a ray-based sochasic framework for modeling paricle velociy channels is developed, which includes hose channel characerisics ha direcly affec he communicaion sysem performance. The parameric naure of he developed models makes hem adapable o a variey of underwaer channels. In his approach, he channel is characerized using simple probabilisic models for he random componens of he propagaion environmen. In his way, he saisical behavior of he channel can be imiaed, and compac expressions for correlaions and oher relaed channel characerisics such as delay and Doppler spreads can be derived. A. Paricle Velociy Channel Correlaions in Space, Frequency and Time In general, correlaions in a muliple-inpu muliple-oupu (MIMO) channel affec he sysem performance and channel capaciy [32]-[35]. Consider he Fourier ransforms of he paricle velociy and pressure channels shown in Fig. 1, called Vx ( f, ), Vy ( f, ), Vz ( f, ), and P( f, ), respecively. Spaial and frequency correlaions beween hese channel funcions are calculaed in [29] for a single vecor sensor, as well as a verical array of vecor sensors in shallow waers. Exension o an oblique array is addressed in [30]. Temporal channel correlaions for a mobile receiver are derived in [30] and [36]. These paricle velociy channel correlaions are needed for proper communicaion sysem design, o achieve he required performance in he presence of some possible correlaions. B. Delay and Doppler Spreads in Paricle Velociy Channels In wireless mulipah channels, he receiver receives he signal hrough muliple pahs. Each pah has a differen delay (ravel ime), because of is differen pah lengh. Moion of he ransmier or receiver in a mulipah channel also inroduces differen Doppler shifs. Delay spread and Doppler spread are wo key channel parameers for sysem design in frequencyselecive and ime-varying wireless channels [37]. Using he paricle velociy channel correlaion funcions developed in [29], analyical expressions for delay and Doppler spreads of paricle velociy channels shown in Fig. 1 are derived [31]. Knowledge of delay and Doppler spreads in acousic paricle velociy channel is imporan for efficien design of underwaer vecor sensor communicaion sysem. C. Simulaion-Based Sudy of Paricle Velociy Channels Using he Bellhop simulaor [38], impulse responses and frequency responses of paricle velociy channels are simulaed in [23] for differen ranges and sedimens. Some channel characerisics such as delay spreads, disribuion of power among he paricle velociy channels, channel condiion numbers and channel eigenvalues are sudied in [23] via simulaion, whereas channel correlaions are briefly invesigaed in [28]. VII. CONCLUSION A vecor sensor can be considered as a mulichannel device ha can be used for daa communicaion in acousic paricle velociy channels. Research in his area enails he inegraion of he unique physics of underwaer acousic propagaion, i.e., he acousic paricle velociy, ino signaling echniques and daa communicaion mehods. An overview of communicaion in paricle velociy channels is provided in his paper, and some opics in channel modeling and ransceiver design are discussed. REFERENCES [1] M. Sojanovic, On he relaionship beween capaciy and disance in an underwaer acousic communicaion channel, in Proc. ACM In. Workshop Underwaer Neworks, Los Angeles, CA, 2006, pp [2] M. Sojanovic, High speed underwaer acousic communicaions, in Underwaer Acousic Digial Signal Processing and Communicaion Sysems. R. S. H. Isepanian and M. Sojanovic, Eds, Boson, MA: Kluwer, 2002, pp [3] D. B. Kilfoyle and A. B. Baggeroer, The sae of he ar in underwaer acousic elemery, IEEE J. Oceanic Eng., vol. 25, pp. 4-27, [4] H. Sari and B. Woodward, Digial underwaer voice communicaions, in Underwaer Acousic Digial Signal Processing and Communicaion Sysems. R. S. H. Isepanian and M. Sojanovic, Eds, Boson, MA: Kluwer, 2002, pp [5] S. Roy, T. M. Duman, V. McDonald, and J. G. Proakis, High rae communicaion for underwaer acousic channels using muliple ransmiers and space-ime coding: Receiver srucures and experimenal resuls, IEEE J. Oceanic Eng., vol. 32, pp , [6] H. C. Song, P. Roux, W. S. Hodgkiss, W. A. Kuperman, T. Akal and M. Sevenson, Muliple-inpu-muliple-oupu coheren ime reversal communicaions in a shallow-waer acousic channel, IEEE J. Oceanic Eng., vol. 31, pp , [7] H. C. Song, W. S. Hodgkiss, W. A. Kuperman, T. Akal, and M. Sevenson, Muliuser communicaions using passive ime reversal, IEEE J. Oceanic Eng., vol. 32, pp , [8] H. C. Song,W. S. Hodgkiss,W. A. Kuperman,W. J. Higley, K. Raghukumar, and T. Akal, Spaial diversiy in passive ime reversal communicaions, J. Acous. Soc. Am., vol. 120, pp , [9] D. B. Kilfoyle, J. C. Preisig, and A. B. Baggeroer, Spaial modulaion experimens in he underwaer acousic channel, IEEE J. Oceanic Eng., vol. 30, pp , [10] D. B. Kilfoyle, J. C. Preisig, and A. B. Baggeroer, Spaial modulaion over parially coheren muliple-inpu/muliple-oupu channels, IEEE Trans. Signal Processing, vol. 51, pp , [11] A. Song, M. Badiey, and V. K. McDonald, Mulichannel combining and equalizaion for underwaer acousic MIMO channels, in Proc. Oceans, Kobe, Japan, 2008, pp Proceedings of Meeings on Acousics, Vol. 9, (2010) Page 4

5 [12] S. Roy, T. M. Duman and V. Mcdonald, Error rae improvemen in underwaer MIMO communicaions using sparse parial response equalizaion, in Proc. Oceans, Boson, MA, 2006, pp [13] R. F. Ormondroyd, A robus underwaer acousic communicaion sysem using OFDM-MIMO, in Proc. Oceans, Aberdeen, Scoland, 2007, pp [14] M. L. Nordenvaad and T. Oberg, Ieraive recepion for acousic underwaer MIMO communicaions, in Proc. Oceans, Boson, MA, 2006, pp. 1-6 [15] R. F. Ormondroyd and J. S. Dhanoa, Comparison of space-ime block code and layered space-ime MIMO sysems for an underwaer acousic channel, in Proc. Oceans, Bres, France, 2005, pp [16] M. Zaman and B. Tracey, Underwaer acousic MIMO channel capaciy, in Proc. Asilomar Conf. Signals, Sysems, Compuers, Pacific Grove, CA, 2002, pp [17] H. F. Olson, Mass conrolled elecrodynamic microphones: he ribbon microphone, J. Acous. Soc. Am., vol. 3, pp , [18] C. H. Sherman and J. L. Buler, Transducers and Arrays for Underwaer Sound. New York: Springer, [19] A. Nehorai and E. Paldi, Acousic vecor-sensor array processing, IEEE Trans. Signal Processing, vol. 42, pp , [20] B. A. Cray and A. H. Nuall, Direciviy facors for linear arrays of velociy sensors, J. Acous. Soc. Am., vol. 110, pp , [21] B. A. Cray, V. M. Evora, and A. H. Nuall, Highly direcional acousic receivers, J. Acous. Soc. Am., vol. 113, pp , [22] G. L. D Spain, J. C. Luby, G. R. Wilson and R. A. Gramann, Vecor sensors and vecor sensor line arrays: Commens on opimal array gain and deecion, J. Acous. Soc. Am., vol. 120, pp , [23] A. Abdi, H. Guo and P. Suhiwan, A new vecor sensor receiver for underwaer acousic communicaion, in Proc. Oceans, Vancouver, BC, Canada, 2007, pp [24] L. E. Freiag, M. Grund, J. Paran, K. Ball, S. Singh, and P. Koski, Muli-band acousic modem for he communicaions and navigaion aid AUV, in Proc. Oceans, Washingon, DC, 2005, pp [25] A. Abdi and H. Guo, A new compac mulichannel receiver for underwaer wireless communicaion neworks, IEEE Trans. Wireless Commun., vol. 8, pp , [26] A. Song, M. Badiey, P. Hursky and A. Abdi, Time reversal receivers for underwaer acousic communicaion using vecor sensors, in Proc. Oceans, Quebec Ciy, QC, Canada, 2008, pp [27] C. Chen, A. Abdi, A. Song and M. Badiey, Digial ransmission via underwaer acousic paricle velociy channels, in Proc. Conf. Inform. Sci. Sys., Princeon Universiy, Princeon, NJ, [28] H. Guo and A. Abdi, Muliuser underwaer communicaion wih spaceime block codes and acousic vecor sensors, in Proc. Oceans, Quebec Ciy, QC, Canada, 2008, pp [29] A. Abdi and H. Guo, Signal correlaion modeling in acousic vecor sensor arrays, IEEE Trans. Signal Processing, vol. 57, pp , [30] H. Guo, A. Abdi, A. Song and M. Badiey, "Correlaions in underwaer acousic paricle velociy and pressure channels," in Proc. Conf. Inform. Sci. Sys., Johns Hopkins Universiy, Balimore, MD, 2009, pp [31] H. Guo, A. Abdi, A. Song and M. Badiey, Characerizaion of delay and Doppler spreads of underwaer paricle velociy channels using zero crossing raes, in Proc. Conf. Inform. Sci. Sys., Princeon Universiy, Princeon, NJ, [32] S. Barbarossa, Mulianenna Wireless Communicaion Sysems. Boson, MA: Arech House, [33] A. Abdi and M. Kaveh, A space-ime correlaion model for mulielemen anenna sysems in mobile fading channels, IEEE J. Sel. Areas Commun., vol. 20, pp , [34] M. O. Damen, A. Abdi, and M. Kaveh, On he effec of correlaed fading on several space-ime coding and deecion schemes, in Proc. IEEE Vehicular Technol. Conf., Alanic Ciy, NJ, 2001, pp [35] A. Abdi and M. Kaveh, A versaile spaio-emporal correlaion funcion for mobile fading channels wih non-isoropic scaering, in Proc. IEEE Workshop Saisical Signal Array Processing, Pocono Manor, PA, 2000, pp [36] H. Guo, Underwaer communicaion via paricle velociy channels: Principles, channel models and sysem design, Ph.D. hesis, Dep. Elec. Comp. Eng., New Jersey Insiue of Technology, Newark, NJ, [37] A. Molisch, Wireless Communicaions. New York: Wiley-IEEE, [38] A. Dunkan and A. Maggi. Underwaer Acousic Propagaion Modeling Sofware. [Online]. Available: hp://cms.curin.edu.au/ Proceedings of Meeings on Acousics, Vol. 9, (2010) Page 5