Genetic algorithm assisted multiuser detection in asynchronous CDMA communications

The up-link transmissions of mobile stations are typically uncoordinated, which lead to asynchronous DS-CDMA systems. Provided that the propagation delay differences of users are less than one symbol duration, every bit of each user is interfered by two consecutive bits of every other user supported by the system which are overlapping with the bit of interest. This is only true, however, with the proviso of an identical channel bit rate for all the users. Hence the multiuser detector (MUD) must have knowledge of these two overlapping bits, in order to efficiently detect the desired bit (DB). Suboptimal MUDs have been proposed based on a truncated observation window, in which the overlapping `edge´ bits are tentatively estimated by some other means. Using a similar approach, a MUD is developed in this contribution which invokes genetic algorithms (GAs), in order to estimate the DBs within the truncated observation window as well as to simultaneously improve the edge bits error probability (EBEP). Computer simulations showed that by using GAs for improving the reliability of the edge bits, our proposed MUD can achieve a near-optimum EBEP performance, while imposing a lower complexity compared to that of the optimum MUD