Improved performance estimation and optimization for chaos-based asynchronous DS-CDMA systems

Recent results proved that there is a constructive procedure to obtain optimum user signatures when the performance of asynchronous DS-CDMA systems can be effectively estimated by means of a double application of Standard Gaussian Approximation (SGA), Such a procedure is based on the statistical theory of chaotic systems and allows the generation of signatures whose auto-correlation profile can be proved to lead to the minimum possible Multiple Access Interference (MAI) power. Regrettably, double SGA may be inappropriate when few users are present in the system. To overcome this, we here apply an alternative approach that has already proved beneficial in this kind of estimation and relies on the evaluation of the fourth-order moment of the interfering signals. These moments will be evaluated relaying on tools of the statistical theory of chaotic systems which have been recently developed by the same authors. The resulting analytical model allows the design of chaos-based signatures depending on the actual system-load in number of users, Such a design procedure reveals that, when few users are accessing the channel, the performance gain with respect to purely random signatures can reach one order of magnitude. © 2001 IEEE.