Research On Improved Channel Estimation For OFDM Based On Particle Filter Algorithm

Due toits good anti-multipath ability and high spectrum efficiency, orthogonalfrequency division multiplexing (OFDM) system has become the key technology ofthe fourth generation(4G) mobile communication and attracted widespread attentions.Wireless channel transmission is an important component in OFDM system，of whichthe multipath propagation and time-varyingcausedistortion of the received signals, thusdecreasing the system performance. To address this distortion in the channeltransmission, the channel estimation is necessary. Among many channel estimationmethods proposed in these years, the particle filter based channel estimation algorithmis one of the best methods in performace and it is widely used in target tracking andstatus estimation.This paper describes a standard particle filteralgorithm and applies itto OFDMchannel estimation, with emphasis on its improvements. The main contents are listedas follows:1.Introduce the characteristics and effects of each wireless fading from a varietyof wireless channel fading, which elicits good background for OFDM channelestimation. Describe OFDM system construction and principles with it develepmemntand list the advantages and disadvantages of OFDM system.2.Introduce the particle filter algorithm based ont the detailed classification ofthe current OFDM channel estimation methods. Starting from bayesian filtering,describe the basic ideas and process flow of the particle filter algorithm in detail. Astwo core factors in the particle filter, theimportance function selection andresamplingtechniques are proposed. And then, the particle filter algorithm isintroduced into OFDM channel estimation based on the self-regression model anddynamically tracks the channel state.3. Aiming at particle filter algorithm’s low accuracyproperty,a new improvedparticle filter is introduced, which combines with the unscented kalman filter. Theproposed method fitst generates distribution and extract particles by UKF algorithm,and then estimates the time-varying channel.Based on UKF, a new method usingsubspace decompositionto obtain the initial channel state is proposed, which is theabbreviation as SSD–UPF. Based on AR model and combined with UPF algorithmcomplete the track of time-varying channels. Simulation results show that this methodhas comparable performance with UPF, and is significantly better than the standard particle filter and the pilot approach.