The Channel Estimation and Signal Detection in OFDM systems using Deep learning.

25 £

This is a technical report supported with novel results which proves that OFDM system based on Deep learning is more reliable and efficient than the conventional system.



 This report presents our recorded results for channel estimation and signal detection in orthogonal frequency division multiplexing (OFDM) systems using a Deep learning toolbox called as long short term memory (LSTM) network. In this research we exploit deep learning to obtain a better symbol classification at the wireless receiver with enhanced signal detection in OFDM systems. Our proposed system can directly estimate the channel state information (CSI) and recover the symbols that are transmitted at same time. Whereas the existing system separately estimates the CSI and using that information it separately detects or recovers the transmitted symbols. The proposed approach based on deep learning address the channel distortion also it detect the transmitted symbol where the performance is comparable to the Least Square and minimum mean square error (MMSE) method. Therefore it is observed that deep learning is a power tool that reduces the complexity of the existing OFDM system and makes it more robust against it. In wireless communication deep learning has great potential for channel estimation and signal detection in complex channel distortion and interference.