The Research For Space-Time Multiuser Detection Technique In Impulsive Environments

Linear multiuser detection (MUD) plays an important role to restrain multiple access interference in code-division multiple access (CDMA) communication system, which can improve the communication system performance and increase the capacity of CDMA system. But with the rapid development of the modern mobile communication, it is difficult for the CDMA system to meet the increasing service demand. By dividing the spatial channel, the space-division multiple access (SDMA) technology can improve the spectrum utilization radio of the communication system, which has became a research hotspot of the wireless communication.The linear multiuser detection (MUD) of the SDMA system is investigated in this paper. In recent years, the minimum mean squares error (MMSE) MUD and the minimum bit error rate (MBER) MUD are widely studied. And it is proved that the MBER algorithm significantly outperforms the MMSE algorithm. However, in most of previous analyses of MBER MUD, the noise presented at the receiver’s front end have been assumed to be Gaussian, which is unreasonable because there are also lots of impulsive noise in real communication environments. Therefore, it is significant to discuss the multiuser detection in the impulsive noise environments.The main contributions of this paper are summarized as follows:1. The MBER MUD for SDMA signals under the Gaussian noise environment is proposed. The steepest descent algorithm and the conjugate gradient algorithm are developed to solve the MBER detector, and the simulation shows the MBER MUD is superior to the MMSE MUD.2. Based on the adaptive least mean squares (LMS) algorithm and the approach of kernel density estimation, an adaptive MBER algorithm, which is called the least bit error rate (LBER) MUD, under the Gaussian noise environment is proposed.3. The MBER MUD algorithm and the LBER MUD algorithm under impulsive noise environments, including both the Bernoulli-Gaussian noise model and the Middleton Class A noise model, are researched. And extensive simulations are given to analyze the performances of the two algorithms in the two impulsive noise environments compared with that in the Gaussian noise environments.