Hybrid reduced-complexity multiuser detector for CDMA communication systems

In a code-division multiple-access (CDMA) system, multiuser detection (MUD) can exploit the information of signals from other interfering users to increase system capacity. However, the optimum MUD for CDMA systems requires the solution of an NP-hard combinatorial optimization problem. It is well known that the computational complexity of the optimum MUD is exponential with the number of active users in the system. In this paper, we apply a hybrid algorithm to develop a suboptimal MUD strategy. The result of the symmetric-successive-overrelaxation-preconditioned conjugate-gradient (SSOR-CG) method is first used to initialize the reduced-complexity recursive (RCR) MUD. Then, the RCR algorithm is applied to detect the received data bit by optimizing the likelihood function. Simulation results for the synchronous and asynchronous case show that the performance of our proposed SSOR-RCR MUD is promising and outperforms the decorrelator, linear minimum mean-squared error, and original RCR MUD.