DS-CDMA synchronization with dual-antenna in ... - IEEE Xplore

1 downloads 0 Views 324KB Size Report
1 Science Park Drive. 118221 Singapore t* School of Electrical and Electronic Engineering,. Nanyang Technological University. Nanyang Avenue, 639798 ...
DS-CDMA SYNCHRONIZATIONWITH DUAL-ANTENNA IN UNKNOWN CORRELATED NOISE

Yugang Mat

G.T.Ye t

K. H.Lit$

Alex C. Kottt

t Singapore Research Laboratory, SONY Electronics (S) Pte. Ltd.

10 Science Park Road, #03-88, The Alpha, Science Park II, 117684 Singapore Singapore Productivity and Standards Board, 1 Science Park Drive 118221 Singapore t* School of Electrical and Electronic Engineering, Nanyang Technological University Nanyang Avenue, 639798 Singapore

Absfrucf- In this paper, we consider the propagation delay estimation of a direct-sequence code-division multipleaccess (DS-CDMA) system in unknown correlated noise. The new propagation delay estimation method proposed here is based on generalized correlation decomposition (GCD). First, the received signal with dual-antenna through a flat-fading channel is modeled. Then, two GCD-based propagation delay estimators: dual-antenna singular value decomposition (SVD)and dud-antenna canonical correlation decomposition (CCD) estimators are presented and compared with common MUSIC estimator. Simulation results show that the proposed estimators, especially the dualantenna CCD estimator, are robust against the near-far problem and outperform the common MUSIC estimator in unknown correlated noise.

(e.g. near-far effect) their performance can be drastically degraded. The near-far effect exists commonly in a wireless environment. Although it can be alleviated by power control schemes, this is not a simple task.

I. INTRODUC~ON

Several near-far resistant propagation delay estimators have been proposedin recent years. They can been classified into two categories: the training-ba$edmethods[ 1, 2, 33 and the blind methods[4,6,7]. One of the merits of the blind methods is that knowledge of the bit sequence is not necessary during the estimation and thus they can be used for either code acquisition or tracking (propagation delay estimation during data transmission)[4]. However, the common subspace-based method, i.e., the common multiple signal classification (MUSIC) estimator, cannot give reliable code timing estimate in correlated noise environments.

Synchronization, which refers here to propagation delay estimation , is a crucial ta$k in direct-sequence code-division multiple-access (DS-CDMA) communication systems. The traditional propagation delay estimation techniques are sliding correlator and its modifications. They are optimal in single-user systems, and have reasonable performance in multiuser environment when the power of the received signal for each user is identical. However, under unequal received power situations

In this paper, we will focus on subspace-ba5edpropagation delay estimation for DS-CDMA signal in unknown correlated noise. The system model here is different from those in the papers mentioned earlier[4, 6, 71 where temporally white background noise is assumed. In practice, the assumption of white background noise is often not valid due to some narrowband interference or the lumping of secondary user signals into with the noise. In this case, the common subspace-based propagation de-

0-7803-6451-1/00/$10.000 2000 IEEE

118

lay estimator can be degraded seriously in performance and can even fail to work. In this paper. we introduce the generalized correlation decomposition (GCD)[S] method for propagation delay estimation in unknown correlated noise. Two GCD-based propagation delay estimators, namely, the dual-antenna singular value decomposition (SVD) estimator and the dual-antenna canonical correlation decomposition (CCD) estimator, are developed to handle the ambient correlated channel noise. In fact, the white noise model is only a special case of the correlated noise model. Therefore, the proposed estimators are applicable more universally to practice environments. In Section 2, the system model is presented. Two propagation delay estimators are developed in Section 3. In Section 4, performance parameters including the probability of correct acquisition and the root mean squared errors (RMSE) with respect to chip duration for the delay estimation are evaluated by computer simulation. Finally, a summary is given in Section 5.

are the path gain, random phase and time delay, respectively, C k ( t ) denotes the spreading waveform of user k with N chips ( a k ( 0 ) a k ( 1 )...ak(N - 1))and c k ( t ) = 0 for t [O,T),T, = T / N denotes the chip duration, and N is equal to the processing gain. The continues-time received signal can be converted to the discrete-time signal r,,(i),i = 1,2... by sampling the output of a chip matched filter[71 at the chip rate. By defining the observation vector at the nth antenna sensor a.. rn(i) = [rn(iN)rn(iN - 1)...~ n ( i N - N l)IT E C N x l,the observation vector can be represented a5

+

In(;)

+ w,,(i),

for 71 = 1,2,

(2)

~L(TK)]

where D = [df(n)di(7-1)...dL(TK) E A,, = diag(A1, Ai, ...AKn AK,,) E B = [bl(i - 1) b l ( i )...b ~ ( -i 1)6 ~ ( i ) E] ~RZKxl, and wn(i) E CNxl is the additive noise vector whose covariance matrix is unknown. The element Ak,, = m a k n e j e b n denotes the complex amplitude of user k at the nth antenna sensor. The vectors dT;(Tk)and dk(Tk) are defined as follows[71

cKxZK,

d;(d d:(Tk)

11. SYSTEM MODEM The system considered here is an a5ynchronous Kuser DS-CDMA system using BPSK modulation in flatfading and near-far environment.It is a w m e d that the receiver employs two well-separated antennas and invokes the standard narrow-band array assumption where the reciprocal of the signal bandwidth is much larger than the signal propagation time across the two antennas such that the path delay differences at two antenna sensors for the same user are negligible[2]. The additive noise at two antenna sensors is spatially uncorrelated[5]. Therefore, the baseband complex envelope representation for the received signal at the nth (n = 1,2) antenna sensor can be written a$

= DA,B

e (1 - E k ) C T ; ( X k ) + W K X k + 11,

(3)

11,

(4)

A

(1- t k ) C : ( X k )

&C:(Xk

+

where Tk = (& xk)T, so that