Research On Channel Estimation Algorithm In OFDM System

With the increase of demand for the multimedia, and with the rapid development of internet, future broadband wireless communication system needs higher spectral efficiency and higher data rate. Currently in all the wireless schemes, multi-carrier modulation represented by OFDM (Orthogonal Frequency Division Multiplexing) is one of the most promising ways, and channel estimation is one of the emphasis research subjects in OFDM techniques.In this thesis, the channel estimation technology is researched, including estimation algorithms based on pilot-symbol-aided and the blind estimation algorithms based on subspace. The implementation of channel estimation based on pilot-symbol-aided is easy, but the accuracy is not high, which makes it restricted in practical applications. On the base of analysis of the pilot-based channel estimation algorithm and several interpolation algorithms, a new OFDM channel estimation algorithm based on perfect sequence is proposed in this paper. It uses perfect sequence which has good related performance as the cyclic prefix of transmitting information. At the receiving end, we extract the cyclic prefix, then do related calculations, finally get the estimated value of the frequency response. The simulation shows that, in multipath fading channels, the proposed algorithm can implement a much accurate estimation than the time-domain filtering algorithm and the channel estimation algorithm used m sequence as the cyclic prefix.More and more attention is paid to blind estimation methods by many researchers because of its highly spectrum-using, but its calculation is complex and the convergence rate is slow. So it is necessary to study a blind estimation algorithm of low complexity and fast convergence. One subspace-based algorithm for blind channel estimation is analyzed in this thesis. It is estimated by exploiting the redundancy of the cyclic prefix, which aims at the algorithms complicated and imprecise. The result shows that, this method is more accurately under the same circumstance of Signal to Noise Ratio (SNR).