Research On Channel Estimation Of OFDM System

Channel estimation is one of the key technologies in wireless communication transmission field.It directly affects the performance of the wireless communication system. In this paper,we summarized multipath and time-varing statistic properties of general wireless channel , gave the channel model based on WSSUS(Wide-sense stationary uncorrelated scatterring) hypothesis. The paper analysised the properties of channel, especially multipath fading and doppler shift's influence to system's performance .Based on these analysis, we investigated the channel estimation methods in block-type and comb-type pilots arrangements .To block-type pilots arrangement,channel estimation is discussed with LS (least square criterion) and MMSE (minimum mean squared error) estimation criterion from the point of views of frequency domain and time domain.In frequency domain,combining MMSE with SVD (singular value decomposition) ,we can obtain the same performance while decreasing the estimator's dimention and complexity greatly.To comb-type pilots arrangement ,we put the pilots to the LS estimator to estimate the channel's frequency domain response using linear interpolation and Weiner filtering interpolation. In time domain,pilots are transformed by DFT at first,then LS and MMSE estimator are used to obtain the channel's time domain response. Adopting the channel model we put forward,we simulated all the algorithms and compared their effect to system's SER in OFDM system under different pilots arrangements. The result showed that under the considering channel circumstance ,when the number of pilots is bigger than the channel length scaled by system's sample interval,time-domain estimation is better than frequency-domain estimation. As for the two criterions, when the number of pilots is smaller than the channel length scaled , MMSE estimation which adopted the second moment statistic properties of channel is better than LS estimation.