For high performance CDMA communications, multiuser detection is often required to suppress the multiple access interference (MAI). Most multiuser detectors rely on accurate channel information to recover the multiuser digital signals. This paper studies the blind channel estimation problem for DS-CDMA systems using aperiodic spreading codes. The maximum likelihood (ML) estimator is formulated for channel estimation. We first convert the multiuser parameter estimation problem into a set of single user optimization problems via alternating optimization, and then determine the channel parameters for each user using an iterative algorithm derived. It is shown by computer simulation that this iterative algorithm can reach global maxima almost always under medium SNR values.

Previous studies showed that multiuser detection in code-division multiple-access (CDMA) communications can be performed without explicit knowledge of users' channel characteristics in a frequency-selective fading environment. However, the computations of these blind approaches are an order of magnitude higher than existing adaptive minimum output energy (MOE) receivers which require at least knowledge of the desired user's channel response. Although the high-complexity problem can be alleviated by constrained adaptive filtering, the tradeoff is a significant drop in receiver performance, especially when the multipath pattern is time varying. In this paper, we present an adaptive receiver for CDMA communications over frequency-selective, and possibly time-varying, wireless channels. A salient feature of the new receiver is that it has complexity and performance comparable to that of the well-known MOE receivers, and yet requires no knowledge of the desired user's channel characteristics.

Differential coding allows signal demodulation without carrier phase estimation, and thus is commonly used to cope with phase ambiguity and residual carriers. In a wireless scenario where the system transfer function is FIR due to multipath reflections, channel estimation and equalization is usually required. Inspired by the work by Tong (1995) on blind sequence estimation, we propose a vector differential coding scheme that allows instantaneous signal detection at the receiver without knowledge of the channel. The new technique can be regarded as a generalization of the standard differential coding method for removing convolutional ambiguities.

The problem of joint carrier offset and channel estimation in code division multiple access (CDMA) communications is studied. An analytic algorithm that provides closed-form channel and carrier offset estimates is proposed. The algorithm first converts the multiuser estimation problem into single-user problems, and then analytically solves the resulting nonlinear multivariate optimization problems utilizing polynomial matrix manipulation. Computer simulations are provided to verify the efficacy of the proposed algorithm.