A gradient guided search algorithm for multiuser detection

This letter introduces a new multiuser detection algorithm based on a gradi ent guided search that can achieve near-optimum performance while its imple mentation complexity is linear in the number of users. The new algorithm at tempts to perform jointly optimum multiuser detection by updating one user' s bit decision each iteration in the best possible way. When a decorrelatin g multiuser detector is employed to initialize the proposed algorithm, in a ll cases tested the gradient guided search converges to a solution with per formance very close to optimum. Further, tests indicate that the proposed a lgorithm provides excellent performance for cases where other suboptimum al gorithms perform poorly. Although we focus only on synchronous multiuser de tection, the algorithm is equally applicable to asynchronous multiuser dete ction problems.