A New Semidefinite Programming Method for Asynchronous Multiuser Detection Problem

Based on the special kind of spreading sequence and semidefinite programming relaxation, A new multiuser detector for asynchronous code division multiple access system is presented. At each transmitter, the special spreading sequence whose second half is the replica of the first is employed. Maximum-likelihood(ML) multiuser detection only need detect 2K once. Then ML detection can be efficiently and accurately approximated using the semidefinite relaxation. Its time complexity per bit decrease to the order of O(2K^3.5) , while the Viterbi Algorithm (optimum detector) requires a exponentially computational complexity. We compare the performances of the new SDP method and a joint method of block coordinate ascent and semidefinite programming relaxation(BCD-SDR) method for the asynchronous multiuser detection problem in various situations, and simulations demonstrate that the new semidefinite programming method often yields better bit error rate (BER) performances than the BCD-SDR method, but the average CPU time of this method is significantly reduced.