A mixed estimation method for MIMO-OFDM channels

The typical estimation methods for MIMO-OFDM channel generally assume that the number of multi-path components is known and constant,and particle filtering can estimate the Gaussian distribution channel.However,in the fact that the number and the amplitude of channel taps are unknown in practical wireless situation,the performance of particle filtering scheme is impaired.In this paper,a channel estimation scheme based on binary PSO and Kalman filtering(PSO-KF) is proposed.Using RST theory a model of MIMO is established and the channel-taps' varying model is given.According to the model the whole channel is separated into discrete part and continuous part,and the relationship with them is derived.The discrete part could be obtained using binary PSO algorithm,and the channel amplitude value is acquired using Kalman filtering.The likelihood distance between the real observation and the estimated observation based on the channel estimation is chosen as the fitness function.Simulation results show the performance of PSO-KF algorithm is better than the particle filtering scheme.