Gibbs Sampling based Near MAP Receiver Design for Multi-User Spread Spectrum Systems

In this thesis, we first establish a unified system model for downlink synchronous direct-sequence code division multiple access (DS-CDMA), uplink asynchronous DS-CDMA, and multi-carrier CDMA (MC-CDMA) systems. Several near-maximum-a-posteriori (MAP) and near-maximum-likelihood (ML) algorithms, including Gibbs sampling, iterated conditional modes (ICM), and expectation-maximization (EM) are next introduced. These algorithms are adopted in the receiver design of uplink asynchronous direct-sequence CDMA
(DS-CDMA) system. In addition to user data estimation, channel estimation for each user and noise variance estimation for blind receiver design are also investigated. Simulation results show that the performance of the Gibbs sampling based receiver outperforms ICM and other suboptimal receivers.

Based on the tree search structure in the minimum distance optimal receiver, we also propose an irregular tree search (ITS) method to greatly reduce the overall complexity. ITS can be viewed as a stochastic procedure of adaptively controlling the survivors at each levels and thus, the complexity can be greatly reduced compared to the M-algorithm at high SNR. Simulation results show that at high SNR, ITS outperforms the M-algorithm both in error performance and complexity.