Study On Projection-Based Channel Blind Estimation Algorithm Of MIMO-OFDM System

With the development of The Times, people is higher and higher requirement for the quality of communication. the traditional communication system is no longer applicable, in the fourth generation of mobile communication in MIMO-OFDM systems is proposed. The system will be OFDM (orthogonal frequency division multiplexing) is combined with MIMO (multiple input and output), complement each other, not only has high transmission rate, and the advantages of resistance to frequency selective fading and multipath fading.For both sides of the wireless communication system, all need to know the channel state information (CSI)--the relevant information of the real channel matrix. So, a new technology produced, so-called channel estimation techniques.The classification in MIMO-OFDM channel estimation algorithm can be divided into three categories, based on the pilot or training sequence estimation, a half blind estimation,blind estimation. Each algorithm have advantages and disadvantages, the first kind of algorithm complexity is low, easy to implement, but the bandwidth overhead is large, cannot meet the requirements of high-speed communication. The second algorithm,using a small amount of pilot to estimate, the complexity of the algorithm between blind estimation and pilot or training sequence estimation. The third type of algorithm complexity has greatly increased, but without extra bandwidth overhead.This paper mainly researches on the channel blind estimation for the system. First.First, this paper introduces the principle of MIMO - OFDM systems, wireless channel fading model, the classification of the channel estimation algorithm. Second, using the tapped delay mo-del on MIMO-OFDM system.Third, order of channel is obtained by subspace projection, then, using the theory of oblique projection and fanaticism convolution channel separation to estimate channel. in the end, the matlab simulation results show that it is feasible. Compared with the classical subspace method, although the estimation precision difference is not big, but the algorithm complexity is greatly reduced.