A Blind Signal Channel Estimation Algorithm Based On Extended-natural Gradient In MIMO-OFDM System

The combination between of the MIMO and OFDM can overcome the radio channel frequency selective decline, increase system capacity, improve the utilization ratio, and it becomes one key 4G spectrum technical. The combination of the MIMO and OFDM can overcome the radio channel frequency selective decline, increase system capacity, improve the utilization ratio, and it becomes one key 4G spectrum technical. However the coherent demodulations of MIMO-OFDM need estimating and tracking the channel. The traditional method is adding the sending data or training sequence, which greatly reduces the system frequency utilization. So the blind signal channel estimation algorithm in MIMO-OFDM system becomes the hot spot of this.Presently, there are still lots of issues to be deeply investigated in BSS theory. The performance of BSS algorithms also needs to be enhanced. Therefor, in this paper we study some traditional BSS algorithms, and propose improvements based on others'previous work. The main work and innovations are as follows:(1) A conception of extended-natural gradient is induced based on natural gradient, and it has been applied in new ICA algorithm of BSS by deeply researching on conventional natural gradient ICA algorithm. The results show that, with proper translation function chosen, extended algorithm converges much faster than conventional natural gradient algorithm.(2) It uses the natural gradient and extended-natural gradient in MIMO-OFDM system and uses five transmitting antenna and five receiving antenna to simulate. Simulation results show that the execution speed of the new algorithm are improved greatly, and its convergence speed is faster compared with the traditional natural gradient algorithm when it selects appropriate parameter transformation matrix functions .