Research On Methods Of Blind Equalization Based On Particle Filtering In Nonlinear Satellite Channel

In modern communication, satellite communication system has become an important way of information transmission. To fully utilized the power resources from satellite, the high power amplifier in satellite usually work in a state of saturation. However, the nonlinear character of high power amplifier will inevitably cause channel nonlinear distortion and the transmitted signal will distort at different degrees, which badly affects the receiving end signal demodulation and the information recovery. Usually the nonlinear channel equalization can effectively eliminate the nonlinear disturb, improving the symbol error rate performance of receiver. Given that particle filter technology has obvious advantage in dealing with the nonlinear problems, in this paper we study on the application of particle filtering to blind equalization in nonlinear satellite channel, aiming to use particle filter technology to solve the nonlinear equalization problems. The main contributions and innovations of this paper are as follows:1. As for the problem that the high power amplifier in satellite causes nonlinear distortion, in this paper we put forward a blind equalization method based on particle filter technology. That is, this method models the nonlinear channel into a state space model to jointly estimate nonlinear parameters and transmitted sequence. Instead of performing linear process, this algorithm regards unknown parameters and symbol sequence as a higher dimensional state and then approximates the posteriori distribution of this higher dimensional state by particles with weights. Finally we can get the estimations of parameters by minimum mean square error criterion and the estimation of symbol sequence by maximum a posteriori criterion. The simulation results show that compared to traditional Volterra equalization method, the particle filter nonlinear blind equalization can perform better.2. In the algorithm process of particle filter blind equalization, there is a need for large quantities of sample particles to precisely estimate the higher dimensional state, which causes more calculations. As for this problem, this study analyzes the characteristics of the constant modulus modulation signals when they pass channel. By simplifying the state space model, we lower the dimension and then directly estimate signal range and additional modulation phase to reduce calculations. Also, by improving the sampling distribution of parameters, we enhance the parameters estimation accuracy of algorithm and equalization performance.3. Particle degradation is the inherent flaw of standard particle filter algorithm. Repetitive sampling can solve the degradation problem, however, given that particles with higher weights are selected for many times, it causes the loss of particle variety, eventually appearing particle exhaustion. Genetic particle filter method can avoid the problem of particle exhaustion caused by repetitive sampling, but the blindness of mutation operator lower the accuracy of algorithm estimation. To solve this problem, in this paper we introduce a method of self-adaptive evolutionary particle filter to overcome the weaknesses of the blind mutation by self-adaptive guided mutation theory. This method can restrain particle degradation and keep the variety, as well as increasing the accuracy of parameter estimation and improving equalization performance when dealing with nonlinear channel blind equalization.