Adaptive optimization of binary/quaternary CDMA signatures in multipath fading environments

We develop a new algorithm to optimize adaptively binary and quaternary signatures for code-division multiple-access (CDMA) communications over multipath channels and/or asynchronous channel access. Using the observed signal autocorrelation matrix, the algorithm attempts to maximize over the binary or quaternary antipodal sequence field the signal-to-interference-plus-noise ratio (SINR) at the output of the maximum SINR linear filter. While this maximization problem is NP-hard, the algorithm is seen to produce in short polynomial time highly desirable solutions that approach in performance the theory-only complex/real-field optimal signature vectors. Signature adaptation may be carried out in either a single or a multi-user mode. Simulation studies included herein offer direct performance comparisons with other known binary signature set designs and the theoretical complex/real-valued optimal vectors