Study On Channel Estimation Techniques For OFDM System Based On Block Pilot

OFDM technology can not only improve the spectrum efficiency of the system very well,but also reduce the bit error rate of the system and the balanced complexity of the receiver.OFDM technology is a potential technology in the wireless communication technologies. Theperformance of OFDM system is good or not depends largely on the performance of channelestimation results.LS criterion is selected in the paper to get the channel response in the pilot location. ButLS criterion does not take the effect of noise in the channel into account. The traditionalmethod uses one time of DFT time domain filtering or transform domain filtering to filter outthe noise on the basis of LS criterion. It analyzes these two algorithms in detail and improvesthe DFT transform domain filtering algorithm. This improved algorithm redefines the cutofffrequency of low-pass filter by tracking the transform domain curve of channel. Thisimproved algorithm can filter more noise in the channel so that the channel response values inthe pilot location are more closer to the true values. Because the one time of DFT filtering canonly remove a small part of noise in the channel, so a method is designed to filter noise fortwo times. The simulation results show that the performance of channel estimation of the newmethod raises a lot. Compared with the no filtering method, it raises7dB under the samecondition of the MSE. Compared with the one time of DFT time domain filtering method, itraises1dB.After getting the channel response values in the pilot location, it needs to adopt aninterpolation algorithm to get all the response values of the channel. It not only introducesseveral common interpolation algorithms, but also uses a new interpolation algorithm calledNewton interpolation (Newton). The simulation results show that Newton and Lagrangeinterpolation algorithm both can improve the accuracy of channel estimation. But Newtoninterpolation has lower computational complexity by introducing the concept of differencequotient. This advantage is very valuable in the actual channel estimation.