Blind Detection of Multi-user Signals in Fading Channels via Discrete Particle Swarm Optimization

Based on the feature that transmission signals belong to a limited alphabet in mobile communications, this paper formulates the blind detection problem of multi-user signals in fading channels into a quadratic programming with binary constraints. This formulation breaks down the restrictive identification condition that transfer function of MIMO channels must be irreducible and column reduced. In the transformation into quadratic programming, a complementary projection operator of a received data matrix is used as an annihilator of a transmitted signal vector to de detected. The quadratic programming problem is solved with discrete particle swarm optimization. Simulations suggest that our method achieves better performances with a lower bit-error rate and robustness without regard to the conventional channel identification condition