The Study Of OFDM Multi Access Scheme Based On Partical Filter

Though the fact that the 3rd and 4th generation communication system has developed maturity and used to business,covering most area,filling the request of voice,data and media communication,and the 5th generation also is developing fast,making it possible for better mobile communication,satellite communication has become a necessary way of communication with the features that it is easy to cover wide area,set base station,communicate far away and take different multi address broadcast,at the same time the cost it take to communicate has nothing to do with the distance between people.Since 1965 in which year that satellite communication started to be used,LEO satellite mobile communication has been more and more important in satellite communication,as the way of its multi address access,the carrier offset estimation of OFDM became a serious problem that need solved.In this paper we focus on the study that based on particle filter to deal with the carrier offset estimation of OFDM in LEO satellite mobile communication system.In this paper,we studied the Doppler frequency shift character of LEO satellite mobile system first,including the value of Doppler frequency shift and its change rate,then we discussed its effect on system,simulated to get the curve that bit error rate change with different normalized Doppler frequency shift,which indicate the necessity of the study of OFDM carrier Synchronization objectively.Secondly,we studied the process of carrier Synchronization,focusing on the effect on integer time estimation and carrier tracking of carrier frequency shift.At last,we come with a plan that base on Doppler frequency pre-compensation and integer time carrier frequency shift estimation to complete rough carrier estimation,and during the state of carrier tracking,we put forward with a plan basing on particle filter to track carrier continuously.Particle filter which also is called sequential Monte-Carlo method is a filter method that based on Monte-Carlo method and recursive Bias estimation,it basis the Large number theorem,take the Monte-Carlo method to get the integral operation of the bias estimation.Particle filter is a method that has been used in many filed,and also widely used in communication filed,because of the fact that computer technology developed very fast.Particle degradation is born with particle filter and is inevitable.There is two main direction to solve this:one is to choose more suitable importance probability density function,the other one is using better resample strategy.In this paper,we focus on this two direction,we came up with unscented gauss particle based on choosing better importance probability density function,and meanwhile we use mean shift algorithm to get more reasonable particle distribution in order to make importance probability density function more suitable to real posterior distribution.