Double Deep Q-Learning Based Channel Estimation for Industrial Wireless Networks

Double deep Q-learning (DDQL) algorithm for channel estimation for industrial wireless networks is proposed. Mean square error (MSE) and bit error rate (BER) as two key metrics are considered to evaluate the performance of the proposed scheme for channel estimation in modified Rician and Rayleigh channels. The estimation error (MSE) shows that the proposed algorithm is comparable to minimum mean square error (MMSE) and has better performance than approximated linear MMSE (ALMMSE). The simulation results demonstrates and further compounds that for number of pilot carriers and BER, the proposed algorithm is robust and efficient than conventional algorithms that are used for channel estimation.