Research On Random Set Theory And Its Application In Communication

The dimension of transmitted signals and the number of wireless channel taps are generally assumed as known constants in typical receiver of communication systems, thus the performance of signal processing algorithms based on this assumption is impaired. In this paper, we explore processing algorithm under the new modelling scheme of multi-dimensional signals and multi-path channel taps'variation circumstance with the app lication of Random sets theory (RST), which has significant value theoretically and practically.This paper is mainly studied about the application of RST in the estimation of MIMO-OFDM channel and the multi-user detection(MUD) of MC-CDMA system.First, the basic theory of random set and the application status are introduced, which are the foundation of full research work. Also Bayesian recursive theory and partical filter algorithms are represented in detail.Second, the estimation of MIMO-OFDM channel based on RST is studied. The typical estimator under the assumption of the number of multipath components and delays to be constant could achive good performance, for example, by sending a sequence of pilot symbols known to the receiver or by using blind techniques which do not require the transmission of pilots. However, for the fact that the electromagnetic media is constantly varying, thus the number of channel taps is also unknown in whole process and typical estimator is not suitable. The channel of MIMO-OFDM is modeled to simulate the multi-path channel taps'varying condition. In this paper using RST theory and the estimator is given, and then estimation schemes based on Partical swarm optimization (PSO), basical Particle filter (PF), and Rao-blackwellised partical filter (RBPF) are proposed. For further improvement, the modification concentrates the re-sample space—by abandoming low probability samples and reserving high probability samples, thus acquiring more accurate approximation at each iteration, this scheme is named as RBPFC—a modified RBPF version. Simulation results show that the performance of RBPFC scheme is the best, RBPF scheme followed, the PSO-based channel estimation scheme followed again but is better than that of the PF-based estimation scheme, and the performance of Kalman filter-based estimation scheme is the worst.Third, the multi-user detection(MUD)of MC-CDMA system is also studied, in mobile multi-access communication systems, the number of active users, as well as their parameters, vary with time. The MUD theory has been developed under the assumption that the number of active users is constant and known at the receiver. This assumption weakens the performance in that a set of users might be inactive at any given time. This paper presents a method based on RST to simulate the change circumstance that active users are continuously entering and leaving MC-CDMA system, and puts forward maximum likelihood (RST-ML) detection scheme,maximum posteriori probability (RST-MAP) detection scheme,Bayes filter (RST bayes) detection scheme,and viterbi (RST Viterbi) detection scheme. To reduce the computation complexity which is an exponential growth with the number of users, the detection scheme based on multi-valued particle swarms algorithms and random set theory (RMVPSO) is proposed . Simulation results show that under the dynamic environment, the performance of the RST bayes schemeis the best, and the RST viterbi scheme followed, the performance of the RST-MAP scheme is worse than that of the RST viterbi scheme but better than that of three kinds of RMVPSO schemes. However, RMVPSO schemes have lower computation complexity than the former schemes, and non-linear PSO have better performance than linear PSO methods.