Performance Analysis of DFT-Spread based OFDM Transmission System over Underwater Acoustic Channels

Transcription

1 erformance Anaysis of DFT-Sread based OFDM Transmission System over Underwater Acoustic Channes Weijie SHE, Haixin SU 2, *, En CHEG 3, Yonghuai ZHAG 4, 3, 4 Key Laboratory of Underwater Acoustic Communication and Marine Information Technoogy (Xiamen University), Ministry of Education,.R.C Xiamen 365, China 2, * Corresonding Author Deartment of Communication Engineering, Xiamen University, Xiamen 365, China, hxsun@xmu.edu.cn doi :.456/jcit.vo6.issue7. Abstract This aer resents ea-to-average ower ratio (AR) and bit error ratio (BER) erformances of Discrete Fourier Transform Sread Orthogona Frequency Division Mutiexing (DFT-Sread OFDM) system over underwater acoustic channes. Comaring with OFDM transmission systems,the structured maing between the DFT sread symbos and sub-carriers enabes the transmitted signa to have a much ower AR in the scheme. And with the aication of an additiona DFT oeration, sreading the information over severa subcarriers wi mae a diversity gain in a frequency-seective fading underwater acoustic channe. Simuations and the exeriment resuts demonstrate that the DFT-sread OFDM system ossesses satisfactory AR and BER erformances, and it is aicabe for signa transmission in a benign underwater channe. Keywords: AR, DFT-Sread OFDM, Underwater Acoustic Communication. Introduction Underwater acoustic channes are considered as quite ossiby nature s most unforgiving wireess medium [].The ocean environment characteristics drive the comexity of underwater acoustic communications systems which can be summarized as simutaneousy imited bandwidth, excessive mutiath deay sread, severe Doer-drift and time variabiity [2]-[4]. These mae the coherent transceiver of high data rate underwater communication systems much more comicated. evertheess, underwater communication has been used since the beginning of the 2th century and the erformance of signa rocessing systems for underwater acoustic communications has dramaticay increased over the ast two decades [5]-[7]. OFDM has roven to be an efficient underying technoogy for underwater acoustic communication [8]-[]. It increases the symbo interva and thereby decreases the inter-symbo interference. It aso converts a frequency seective channe into a set of arae frequency-fat sub-channes, thus greaty simifying receiver equaization. A major drawbac of OFDM transmission, however, is its AR []-[3] which increases oerationa requirements of the inear ower amifier in the transmitting equiment eading not ony to an increased cost but aso an increased ower consumtion which is not desired eseciay at the uin transmitter. A number of techniques have been deveoed to reduce AR in OFDM communication systems [4] and the DFT-Sread OFDM system [5] is a romising aternative of such investigations. This scheme was first roosed in [6]. In this scheme, data symbos are sread over severa subcarriers by DFT, and this is foowed by ordinary OFDM rocessing. With an extra DFT boc rior to the conventiona OFDM transmitter, it roves to be an effective way of combining the benefits of OFDM with a ow AR transmission signa. As such it has aready been seected as the uin moduation scheme for the ucoming Long Term Evoution of 3G systems under the wor item of Evoved-UTRA by 3G [7]. The rest of the aer is organized as foows. In the next Section, the AR and channe estimation erformances of DFT-Sread OFDM over underwater acoustic are addressed in detai. Simuation and Exeriment resuts in the exerimenta oo of Xiamen University are erformed in Section III, as we

2 as the resuts in shaow water near Xiamen ort. Finay, the concusion is given in Section IV. 2. ERFORMACE AALYSIS 2.. AR Suose that the symbo maing of the inut data { d, i 2M - } is { x = d + jd, M -} and then sread by FFT oeration in a DFT-Sread OFDM 2( i-) 2i- system. i S = FFT ( x ) M - = å x e - j2 / M () The DFT sread symbos { S, M -} are maed onto M sub-carriers. Oversaming by a factor of Q is used to aroximate the true signas since yquist rate saming robaby misses some signa eas. The Q -time oversamed sames can be obtained by erforming QM -oint inverse discrete Fourier transform (IDFT) on the data boc oeration can be described as foows. S with ( Q - ) M zero-adding. The X ìs, M - = í î, M - (2) where denotes the th sub-carrier symbo. The resuting frequency-domain symbos X are then transformed to the time-domain symbo sequence{ xn, n -} by an IFFT oeration. x n = IFFT ( X ) = = = = - å = M - å = X e S e M - M - å å = = M - M - å å = = n j2 n j2 - j2 M x e x e e ( n-q ) j2 n j2 (3) AR is defined as the ratio of the maximum to the average ower during an transmission symbo eriod. For x, we therefore have n 2 max [ xn ] n L - 2 E( xn ) AR = (4) - 8 -

3 where E ( ) denotes the exected vaue. The structured maing between the DFT sread symbos and sub-carriers enabes the transmitted signa to have a ow AR in the DFT-Sread OFDM system. Excet for the DFT sreading and the structured symbo to sub-carrier maing, it is identica to OFDM Channe Estimation The two basic tyes of iot arrangement for DFT-Sread OFDM channe estimations are iustrated in Figure. One is boc-tye iot based channe estimation, and the other is comb-tye iot based channe estimation. Frequency Frequency Figure. Two Basic Tyes of iot Arrangement for DFT-Sread OFDM Channe Estimations The boc-tye iot based channe estimation is erformed by inserting iot tones into a subcarriers of symbos within a secific eriod which is deveoed under the assumtion of sow fading channe. And the comb-tye iot based channe estimation is introduced to satisfy the need for equaization when the channe changes even from one DFT-Sread OFDM boc to the subsequent one. Here the interoation is needed to estimate the conditions of data subcarriers since it is erformed by inserting iot tones into certain subcarriers of each symbo. In the next sections, the rocessing of these two basic methods wi be described Boc-Tye iot Channe Estimation As shown in Figure (a), iot symbos are transmitted eriodicay, and a subcarriers are used as iots in boc-tye iot based channe estimation. Here the tas is to estimate the channe conditions H ( ) through the transmitted iot signas X and the received iot signasy, with or without using certain nowedge of the channe statistics. There are many techniques to estimate the channe state information exoiting iot bocs, such as east-squares (LS), minimum mean square error (MMSE) and inear minimum mean square error (LMMSE). Due to comutationa comexity, LS and LMMSE estimator are chosen here. Suose that the iot symbo is X. The receive symbo isy, and W is the additiona noise. The LS estimation of H is The LMMSE estimation of H can be obtained by Y ( n) W ( n) H LS ( n) = = H ( n) + (5) X ( n) X ( n) 2 T - HH HH w H LMMSE = R ( R + s ( diag( X ) diag( X ) ) - ) H LS (6) - 8 -

4 Here, H is the Channe frequency resonse at the iot subcarriers, and auto-correation between the iot subcarriers, diagona matrix of the iots Comb-Tye iot Channe Estimation As shown in Figure (b), for each transmitted symbo, 2 w R HH reresents the s is the variance of the noise, and diag( X ) is the iot signas are uniformy inserted into X with L with subcarriers aart from each other in comb-tye iot based channe estimation, T where L = /. The iots ocations = [ s ] ( s =,, L, -), the iot T vaues X = [ X ] ( s =,, L, -), and the received signa s receiver. The LS estimations of the channe conditions Y are nown at the H at the iot subcarriers are cacuated by T [ /, /,, / ] - - H = Y X Y X L Y X (7) Then the basic tas is, according to the LS estimates at iot subcarriers H, received signasy, and maybe certain additiona nowedge of the channe statistics, to estimate the channe conditions at the data subcarriers (secified by H with ength ). Due to cacuating amounts, the soutions incude LS estimator with inear interoation and sine interoation. The inear interoation of H is H ( ) = H ( ml + ) = ( - ) H ( m) + H ( m + ) (8) L L The sine interoation of H is H ( ) = H ( ml + ) ' ' = a H ( m) + a H ( m + ) - La H ( m) + La H ( m + ) (9) where ' H ( m) is the first derivative of H ( m) 2 3 a = L - L, 2 3 3( L - ) 2( L - ) a = L L,and the definitions of a and a are : 3. SIMULATIO AD EXERIMET 3.. Simuation The AR erformance of OFDM and DFT-Sread OFDM signas are simuated. The system arameters for the simuation are indicated in Tabe. The resuts are shown in Figure

5 Tabe. SYSTEM SECIFICATIO DFT ength 24 IFFT ength 496 Guard interva ms Symbo duration 256ms Effective bandwidth 6Hz Effective seed 6bit/s The first ones, the AA vaues of OFDM signas, are arger than 8dB in most cases. They have reached db in some arts of the symbos. Whie DFT-Sread OFDM signas have much smaer AR vaues which are between 5 and 8dB. Figure2. AR of Traditiona OFDM and DFT-Sread OFDM systems To better describe the otentia erformance imroving, the Comementary Cumuative Distribution Function (CCDF) is exoited to describe the AR difference of the two inds of signas. The resuts are shown in Figure 3. The abscissa is the AR vaues, and the ordinate is the robabiity that the AR is arger than a certain threshod. As can be seen from figure, when CCDF is in excess of., the AR vaue of DFT-Sread OFDM system is about 7.5db, in which the OFDM system is about.8db. The former has an imrovement about 3.3dB. Figure3. CCDF of Traditiona OFDM and DFT-Sread OFDM systems The BER erformance of the DFT-Sread OFDM system is simuated over a Rayeigh fading channe. Suose that the time synchronization and frequency synchronization are erfect and the signas transmit without any distortion. The guard interva is chosen to be onger than the maximum deay sread in order to avoid inter-symbo interference. Simuations are carried out for different signa-to-noise ratios (SR). The resuts are shown in Figure

6 Figure4. BER erformance of Channe Estimation Methods for DFT-Sread OFDM system in a Rayeigh Channe As can be seen from the figure, the channe estimation methods are sense to SR. As the SR increased, BER decreased raidy. When the BER is beow 2dB, the LMMSE estimator was aroximatey 3dB better than the LS method. The erformance of LS estimator imroves greaty when the SR is above 2dB. Though the boc-tye iot based LMMSE estimator can achieve better erformance, the amount of cacuation of the agorithm is greater Exerimenta oo resuts The exeriment was carried out at the exerimenta oo in Xiamen University. System secifications for the exeriment are the same as in the simuation. The ocation of the transmitter and the receiver is.8m under water, and the distance between them is 7m. Both of them et sti during the exeriment. LFM signas are aended before the data sequence at the transmitter. At the receiver, synchronization is achieved via correating the received sames with the nown LFM sequence. Then the received data are divided into DFT-Sread OFDM rocessing. o. o channe estimation Tabe2. EXERIMET OOL RESULTS (BER) LS Comb-Tye iot Channe Estimation inear Interoation sine Interoation Average As seen from Tabe 2, comaring with no channe estimation, the roosed channe estimation methods can greaty imrove the system erformance. Eseciay, the erformance of boc-tye iot based LS channe estimation is much better than comb-tye iot based channe estimation at the exerimenta oo. The average BER of boc iot based method was.82%, whie the average BERs of comb-tye iot based channe estimation are about 3.6% with a maximum vaue of 6.36% and a minimum vaue of 2.52%. That is because the exerimenta oo channe is stabe which can achieve high SR, and the channe imuse resonse at every sub-carrier is cacuated accuratey by the boc iot based channe estimation. Whie in the comb-tye iot based channe estimation, the channe imuse resonse is estimated by interoation. The actua erformances of the agorithms sti need to be comared in shaow water exeriments Shaow water exeriments

7 The exeriment was aso carried out in the shaow water near Xiamen ort. The distance between transmitter and receiver was 2m. And the ocations of the transmitter and receiver were about 4 meters under water. Tabe 3 is the BER erformance of the exeriment in the rea marine underwater acoustic channes. o. Tabe3. EXERIMET RESULTS I SHALLOW WATER (BER) Boc-Tye iot Channe Estimation Comb-Tye iot Channe Estimation LS LMMSE inear Interoation sine Interoation Average As can be seen from the tabe that the boc-tye iot based LS channe estimation didn t achieve desirabe resuts. The minimum vaue of BER is.67%. The average BER is 3.49%, and the maximum vaue of BER even reached 8.57%. The boc-tye iot based LMMSE channe estimation achieves a much better resut, which gets an average BER of 6.9%. The minimum even reached 5.97%, and the maximum was just 8.2%. In comarison, comb-tye iot based channe estimation can achieve a stabe and good erformance. BER were beow % in the most cases. Exoiting different inds of interoation methods, the average BER were 8.8% and 6.96% resectivey. The minimum BER even reached 4.37%. 4. Concusion Through simuations and exeriments, this aer resents the AR and BER erformances of the DFT-Sread OFDM system over underwater acoustic channes. The structured maing between the DFT sread symbos and sub-carriers enabes the transmitted signa to have a ow AR. A diversity gain is achieved by the aication of an additiona DFT oeration in this scheme. The resuts in a benign underwater channe show that, comaring with the OFDM systems, the roosed DFT-sread OFDM system wi achieve satisfactory AR suression and BER erformances. The study demonstrates that the DFT-sread OFDM transmission system is aicabe for underwater acoustic communications. 5. Acnowedgments This wor was suorted by The atura Science Foundation of Fujian rovince, China under Grants 29J555. It is aso suorted by 2 roject on information technoogy of Xiamen University. The authors woud ie to than the unnown reviewers for their vauabe suggestions and critique. 6. Reference [] D. Brady and J. C. reisig, Underwater Acoustic Communications, In H. V. oor and G. W. Worne, editors, Wireess Communications: Signa rocessing ersectives, USA, 998. [2] J.A. Catiovic, erformance imitations in underwater acoustic teemetry, IEEE Journa of Oceanic Engineering, Vo.5, o.3,.25-26, 99. [3] M. Stojanovic, J.A. Catiovic, J.G. roais, hase-coherent digita communications for underwater acoustic channes, IEEE Journa of Oceanic Engineering, Vo. 9, o.,.-,

8 994. [4] A.C. Singer, J.K. eson, S.S Kozat, Signa rocessing for Underwater Acoustic Communications, IEEE Communications Magazine, Vo.47, o.,.9-96, 29. [5] M. Stojanovic, Recent advances in high-seed underwater acoustic communications, IEEE Journa of Oceanic Engineering, Vo.2, o.2,.25-36, 996. [6] E.M. Sozer, M. Stojanovic, and J.G. roais, Underwater acoustic networs, IEEE Journa of Oceanic Engineering, Vo.25, o.,.72-83, 2. [7] W. Shen, H. Sun, E. Cheng, Y. Zhang, SR Estimation Agorithm Based on iot Symbos for DFT-Sread OFDM Systems over Underwater Acoustic Channes, Journa of Convergence Information Technoogy, AICIT, Vo.6, o.2,.9-96, 2. [8] Y. Li, L. J. Cimini, Jr., and. R. Soenberger, Robust channe estimation for OFDM systems with raid disersive fading channes, IEEE Transactions on Communication, Vo.46, o.7,.92-95, 998. [9] S. M. Riazu Isam, Kyung Su Kwa, On Channe Estimation in MB-OFDM UWB Systems with Time Varying Disersive Fading Channe, JDCTA, AICIT, Vo.4, o.2,.8-24, 2. [] M. Romoozi, H. Ebrahimour-omeh, A ositioning Method in Wireess Sensor etwors Using Genetic Agorithms, JDCTA, AICIT, Vo.4, o.9,.74-79, 2. [] H. Ochiai and H. Imai, On the distribution of the ea-to-average ower ratio in OFDM signas, IEEE Trans. Communications, Vo.49, o.2, , 2. [2]. Dinur, D. Wuich, ea-to-average ower ratio in high-order OFDM, IEEE Trans. Communications, Vo.49, o.6,.63-72, 2. [3] T. Jiang, M. Guizani, H. Chen, W. Xiang, and Y. Wu, Derivation of AR distribution for OFDM wireess systems based on extreme vaue theory, IEEE Transactions on Wireess Communications, Vo.7, o.4, , 28. [4] T. Jiang, Y. Wu, An Overview: ea-to-average ower Ratio Reduction Techniques for OFDM Signas, IEEE Transactions on Broadcasting, Vo.54, O.2, , 28. [5] M.D. isar, H. ottensteiner and T. Hindeang, On erformance Limits of DFT Sread OFDM Systems, Mobie and Wireess Communications Summit, 6th IST,.-4, 27. [6] Rui Dinis, David Faconer, Chan Tong Lam and Maryam Sabbaghian, A mutie Access Scheme for the Uin of Broadband Wireess Systems, Goba Teecommunications Conference, 24, Vo.6, , 24. [7] Standardization Committee 3G. hysica ayer asects for E-UTRA, 3G TR Onine, htt://3g.org,