A novel approach for interference cancellation of DS-CDMA systems using projected successive overrelaxation algorithm

Optimal multiuser detection is a 0-1 quadratic optimization problem with binary constraints which demands the search of the whole signal space to achieve the optimal solution. In the presence of such hardship, some soft interference schemes based on Gaussian-Seidel and Jacobi algorithms have been developed to achieve sub-optimum solution of the original problem. This paper presents a novel constrained successive overrelaxation algorithm for the original optimum multiuser detection problem, which corresponds to a near-optimum weight-controlled multistage soft interference cancellation scheme for the DS-CDMA systems. It is shown this scheme converges much faster than the Gaussian-Seidel method and Jacobi method while keeping similar computational complexity. With a few iterations it can yield performance much better than the conventional linear minimum mean squared error multiuser detection method.