Research On Channel Estimation Algorithm Based On Pilot In OFDM System

Communication service with high speed and high quality is required in future wireless mobile communication system. OFDM technique is a kind of high-speed multi-carrier transmission technique, which has high utilization of frequency spectrum and excellent capability of countering multi-path interference. This makes it key technique in the next generation mobile communication system. The wireless mobile channel is the most complex channel in the modern communication system. Since the channel of OFDM system has frequency selective and time-varying characteristics. It is necessary to estimate channel so that the channel state information can be obtained at receiving end and coherent demodulation is realized accurately and in time. The research of channel estimation in OFDM system can be conducted in two categories:one is based on blind channel estimation, the other is channel estimation based on pilots. Because of the high algorithm complexity, the slow convergence and the bad flexibility of the blind channel estimation, it is not suitable in practical use. So this paper focuses on the research of channel estimation based on pilots.In this paper, the basic principle and model of OFDM system and the characteristics of wireless channel are analysed firstly. Then several channel estimation algorithms are researched. And the performance comparisons of these algorithms are made by computer simulation. This paper mainly studies DFT-based channel estimation in OFDM system, and provides two improvements for the shortage of the conventional algorithm:(1) When the channel is sample-spaced multipath channel, the conventional algorithm only removes the noise in the channel impulse response beyond the length of the cyclic prefixs (CP), while the noise within the length of CP is not suppressed at all, affecting the performance of the algorithm. An improved algorithm is provided for this problem. Only by introducing a decision threshold into the length of CP, the influence of the noise could be further removed by threshold processing, and the performance of channel estimation is improved. At last, the simulation results prove the effectiveness of the modified algorithm.(2) When the channel is non-sample-spaced multipath channel, the conventional method will lead to energy leakage of the channel impulse response, and reduce the channel estimation performance. An improved algorithm is present to solve this problem. By calculating the changing rate of the power to select the effective path, and further improve the performance of channel estimation. At last, the simulation results prove the effectiveness of the modified algorithm.The two-dimensional FFT interpolation channel estimation algorithm is researched. However, the channel estimation algorithm is easy to bring noise error, leading to the problem of insufficient precision. To solve this problem, LMS filtering algorithm is applied to channel estimation. According to the changing statistics of the inputsignal, LMS algorithm constantly adjusts the filter weights and further reduces noise in order to achieve optimal output. At last, the simulation results prove the effectiveness of the modified algorithm.