Based On Bayesian Compressed Perception Of Ofdm Channel Estimation Technology Research

OFDM(Orthogonal Frequency Division Multiplexing), is the key technology of thefourth generation mobile communication system. Transmission speed is increased bymultiplexing orthogonal sub-carriers. A real-time and accurate channel estimationtechnology is highly demanded in order to receive information correctly. The traditionalchannel estimation methods, such as LS(Least Square), MMSE(Minimum Mean SquareError), TD(transform domain)require mass of pilot patterns to estimate channelinformation correctly, resulting in data speed degradation. Therefore, the traditionalchannel estimation methods do not meet the requirements of high efficiency and highspeed of next generation mobile communication system. In order to solve this problem,channel estimation methods based on CS(Compressed Sensing) are proposed. That kindof channel estimation method takes advantage of channel’s sparsity to improveefficiency. However, the known CS channel estimate methods have a bad performancein low SNR condition.Based on that, this paper studied the performance of CS channel estimatetechnology and ways to improve performance of CS channel estimation in low SNRcondition. The main points and achievements are described as follows:(1)Firstly, the key techniques and models in OFDM system and the basiccharacteristics of wireless channel are analyzed. Models for applying compressedsensing technology to OFDM channel estimation are researched. The factors that affectperformance of compressed sensing channel estimation are researched.(3)Secondly, in order to enhance the performance of channel estimation in lowSNR conditions, models for applying Bayesian compressed sensing technology toOFDM channel estimation are researched. The Bayesian framework provides anestimate for the posterior density function of additive noise encountered whenimplementing the compressive measurements. Under Bernoulli Gaussian priorassumptions, OFDM channel estimation methods based on Bayesian compressedsensing are researched. Fast bayesian adapting matching pursuit algorithm is proposedto extend the scope of algorithm. In simulation, the performance in low SNR condition of Bayesian compressed sensing channel estimation is enhanced.(4)Finally, on the basis of bayesian compressive sensing features and priorinformation of the channel, a new pilot pattern design method is proposed. In the pilotpattern design method, by taking advantage of the characteristics of the current channeland BCS channel estimate methods, setting maximum posteriori probability as theiterative atom and designing a fast search algorithm, the optimal pilot pattern is found.The performance of channel estimation in low SNR condition is enhanced, which isproved in simulation.