Adaptive Channel Estimator for Multiuser Cdma Receivers

An adaptive channel estimator for frequency selective fading channels is proposed for the uplink in Code Division Multiple Access (CDMA) mobile communication systems. The proposed scheme is well suited for both linear and nonlinear multiuser detectors. In this paper we will show the importance of channel estimator adaptivity especially to synchronization of weak users. A simple Least Mean Squares (LMS) predictor was used to adapt the weights of channel estimation lters of diierent users. The performance of such an algorithm was compared to two non-adaptive schemes by computer simulations. The simulations do show that the adaptive predic-tor outperforms clearly non-adaptive estimators regardless of the interference level. Code Division Multiple Access (CDMA) communication systems have recently gained considerable attention both among industrial and academic communities. The so-called conventional approach to CDMA reception is to neglect multiple access interference (MAI) and near-far problems. This would require strict power control and good crosscorrelation properties of the spreading codes. Another much more eecient way to detect diierent users in a CDMA system is based on multiuser reception 1]. An ideal multiuser receiver suuers no degradation due to other users in the system. Unfortunately the optimum receivers are far too complex for practical implementations and hence several suboptimal receivers 2] have been proposed. The performance of both optimum and subop-timum receivers is well understood in AWGN channels. A great deal of the recent CDMA research addresses the problems of multipath fading 3]. In a multipath fading channel the complex amplitudes, propagation delays and data must be estimated for every user. The optimum receiver in a frequency selective channel leads to a multidimensional optimization problem of aforementioned parameters. Especially, when tracking the time-varying channel parameters, the multidimen-sional optimization problem can be well approximated by KL separate estimation problems. The other users' parameters are assumed to be known when estimating the k th user's l th path delay and complex amplitude. This leads to interference cancellation 2] in channel parameter estimation. Basically, an estimate of the multiple access interference (MAI) must be formed for every user before subtracting the MAI estimate from the matched lter outputs. The complex channel estimates are formed by subtracting the MAI estimates from matched lter outputs, by multiplying the delayed MAI-free matched lter output by the complex conjugate of the previously detected data symbol and nally by ltering the result to suppress noise. Any lter causes some delay to the ltered signal. …