Research On Blind Equalization In OFDM System Based-on Particle Filter Algorithm

Orthogonal Frequency Division Multiplexing(OFDM) technology has a variety of advantages,such as excellent anti-multipath performance,eliminate inter-symbol interference and high bandwidth efficiency.While the introduction of the Fast Fourier Transform,the OFDM significantly reduced the complexity of wireless communication systems.So it is suitable for transmitting high-speed data in wireless channel.But because of the complexity and unknown character of communication channel,it needs to use equalization for signals.Although equalization method based on using of pilot signals have been introduced,pilot signals take up precious channel resource.Thus it reduces the transmission efficiency of the system.The blind equalization is more fit the OFDM system.So it becomes the most important condition of OFDM system.In this thesis,we first research the basic theory of OFDM,and then introduce the system model and key technology.Based on the theory of Bayesian filter,we research the particle filter.Particle filter is used in the blind equalization of OFDM system through the establishment of state-space model.Particle filter realizes the recursive Bayesian filtering through non-parametric Monte Carlo simulation method.A set of particles,which are randomly sampled from probability function and have corresponding weights,are introduced to approach the posterior distribution.Therefore it suits for nonlinear and non-Gaussian problems which can be expressed by state-space model.The precision of particle filter approaches the precision of best estimating.But there are some existing problems in particle filter such as weight degeneracy,sample impoverishment and heavy computation burden.When it is used in OFDM blind equalization,the algorithm should be improved.This thesis researches four improved algorithms:Sampling Importance Re-sampling(SIR),Auxiliary Sampling Importance Re-sampling(ASIR),Regularized Particle Filter(RPF),and an improved algorithm of better performance which is acquired by particle filter use the system re-sampling and Markov Chain Monte Carlo (MCMC) algorithm.Finally,the four algorithms are simulated in the model of OFDM blind equalization.The result shows that compared to the general particle filter algorithm,the improved algorithms are more reliable and the filter time is shorter.They improve the real-time and accuracy of the OFDM system.