Low-complexity multiuser channel estimation with aperiodic spreading codes

Signal processing techniques for CDMA systems employing aperiodic spreading sequences have gained significant interest recently. Due to the time-varying nature of users' unknown signatures in a multipath communication environment, direct design of blind multiuser detectors is intractable. We focus on estimating the unknown multipath parameters for each active user in the system. The problem is solved in the correlation matching context based on correlations of both the directly received data and the outputs of a bank of matched filters. Three typical scenarios are discussed such as quasisynchronous uplink CDMA system with AWGN, with unknown interference, and downlink CDMA system with AWGN, leading to different solutions. We model the aperiodic spreading codes as random variables. For any priori-known distribution of the spreading codes, their statistics up to the fourth order can be evaluated, resulting in extremely low computational complexity of the methods. The identifiability of the channel parameters only depends on the nonsingularity of a deterministic matrix determined by known system parameters. In the case of unknown code statistics, the methods can be modified to be still applicable by estimating those code statistics from given spreading codes. However, in such a case, more computations are needed. Comparisons with other existing methods show that the proposed computationally efficient approaches can provide satisfactory results while requiring significantly less computations.