Neural Network based Channel Identification and Compensation

Under the fast fading environment, the estimated channel state information (CSI) is largely different from real channel state particularly in the last part of the packet. To mitigate this influence, an iterative channel estimation (ICE) has been proposed. This method generates the replica signals by using the demodulated signals and the channel state information can be obtained via its remodulation. Channel variation can be compensated by using the error components defined as the difference between the replica symbols and received symbols. Additionally by exploiting forward error correction code, the CSI estimation accuracy can be gradually improved as iteration. However, this method requires symbol-by-symbol operation to track the channel variation which imposes huge computation amount. To reduce the number of channel compensation process, this paper proposes a neural network based channel identification and compensation scheme for OFDM system. It can estimate the whole transition of channel states and efficiently compensate the channel variation using the generalization capability of a neural network. The computer simulation results clarifies that the proposed method can improve the BER performance even under the fast fading environment.