Research On Joint Detection For TD-SCDMA System

TD-SCDMA is one of the three main standards for the third generation communication system, and joint detection is a key technique in TD-SCDMA. Joint detection can reduce or eliminate multiple access interference (MAI), inter-symbol interference (ISI) and near-far effects, simplify power control, reduce power control precision, and remedy the negative influence caused by the nonideal relativity of the orthogonal spreading code. Consequently, joint detection technique can improve the system performance, increase the system capacity, and extend the coverage area of the cell. Blind joint detection for TD-SCDMA System based on parallel factor technique is investigated in this paper. The main contents are as follows.TD-SCDMA system and its key technique are introduced. Smart antenna and joint detection are particularly explained. Then wireless channel model, array model and space-time channel model are studied, and a kind of space-time signal model over a multipath channel for CDMA system is given.Traditional space-time multiuser detection algorithms are discussed in section 3, including space-time minimum mean square error (ST-MMSE) algorithm and space-time interference cancellation algorithm. The bit rate error (BER) performance of these algorithms is analyzed by simulation experiment.The basic mathematic knowledge of the parallel factor is introduced in section 4. Blind joint detection algorithms based on parallel factor technique are investigated in section 5. First, the paper studies joint detection for TD-SCDMA system over a single path synchronous channel. An analysis of the received signal shows that the received signal has trilinear model characteristic. So, joint detection based on trilinear alternating least square (TALS) algorithm is proposed. The simulation results reveal that the BER performance of TALS algorithm is close to the ST-MMSE algorithm. The larger the number of snapshots is, the better the BER performance will be. When the number of users increases, the BER performance will degrade. The algorithm also has robust characteristic in array error condition. Then the paper studies two kinds of parallel factor algorithms with fast convergence speed: Complex parallel Factor analysis (COMFAC) and trilinear alternating least square with linear search (TALS-LS). The BER performance of the two algorithms is very close to TALS algorithm, but the two algorithms have better convergence.Joint detection algorithms based on parallel factor technique over a multipath channel are investigated in section 6. The methods that using guard interval with zero padding and cyclic prefix before every symbol to eliminate ISI are proposed in this paper, and joint detection based on parallel profiles with linear dependencies (PARALIND) algorithm is also proposed. Then, a novel joint detection method which can improve the efficiency of frequency band over a multipath channel is presented. The method uses PARALIND2 algorithm. The simulation results reveal that the BER performance of PARALIND algorithm and PARALIND2 algorithm is close to the ST-MMSE algorithm. The larger the number of snapshots is, the better the BER performance will be. When the number of users increases, the BER performance will degrade. The two algorithms also have robust characteristic in array error condition. The algorithms are blind, which do not need the channel parameter.