Efficient Dynamic Channel Estimation For Large Scale IRS Assisted Massive MIMO In 6G Communications

Abstract

Intelligent Reflecting Surfaces, assisted massive MIMO systems have good potential in 6G wireless communication, channel estimation is still a major challenge. Conventional estimation methods including LS, KRF and BALS typically provide limited estimation performance and poor scalability in dynamic IRS environments. The paper presents EMRO–DSNN technique for efficient channel state information (CSI) estimation to combat the above-mentioned problem. The enhanced mud ring optimization (EMRO) algorithm is responsible for optimizing the phase shift configurations for the IRS for maximum SNR and spectral efficiency while the deep synergetic neural network (DSNN) utilizes the optimized IRS-assisted channel and received pilot signals to accurately estimate CSI under complex propagation conditions. Simulation Results, EMRO–DSNN outperforms CE-Sub-Wise, CE-PARAFAC, CE-DoubleIRS and CE-KRF-BALS with 95% NMSE reduction, 74% SNR improvement and 83% BER reduction and 14% SE gain with 43% lower computational complexity.