Channel Estimation Using Deep Learning Techniques with Hippopotamus Optimization Algorithm for Millimeter Wave MIMO-OFDM System

Millimeter wave Massive Multiple Input Multiple Output (MIMO) and Orthogonal Frequency Domain Multiplexing (OFDM) utilize a lens antenna array to minimize the count of Radio Frequency (RF) chains.However, MIMO-OFDM faces a significant issue due to having fewer RF chains, as compared to the number of antennas.The estimation of beam space channel in the sparse signal recovery makes it challenging due to the channel exploitation.The traditional approaches-based channel estimation faces significant challenges like degradation of the model training process due to inappropriate adoption of Deep Learning (DL) architectures.In addition, they have complex architecture and require higher training data.So, this research develops a sparse channel estimation approach with the help of DL approach, along with the optimization technique.The Adaptive Deep Neural Network (ADNN) helps in uplink channel estimation and the Bidirectional Long Short-Term Memory (Bi-LSTM) helps to perform hybrid pre-coding.The Hippopotamus Optimization Algorithm (HOA) is introduced to optimize hidden neurons for enhancing accuracy.The experimental outcomes exhibit that the proposed approach achieves a better accuracy of 0.9831 at the learning rate of 35, as opposed to the existing method, Trail-based Red Deer Algorithm (T-RDA).