Dual-Stage Channel Estimation using ANN and Hybrid RIS Aided MIMO Systems

In Multiple-Input-Multiple-Output (MIMO) communications, the Reconfigurable Intelligent Surface (RIS) can be described in a crucial enabler. It is used to enhance the signal coverage on it. It is also used to reduce the frequency obstructions in it. The Channel State Information (CSI), is one of the main obstacles to the constructive distribution of the Reconfigurable Intelligent Surface for the collecting purpose on it. Since, cascaded or descend or individual channels can only be manipulated at a base station (BS). It can be done through the uploading training or a MS, which is possibly referred to as Mobile Station can be done through the downlink training, the passive Reconfigurable Intelligent Surface in the absence of baseband processing capability makes channel estimation (CE) challenging. Now the study focuses on this hybrid Reconfigurable Intelligent Surface architecture. In this architecture, there are a very low number of elements. It can have an active and capable of receiving and proceeding the pilot signals on RIS, with subjective to facilitate the CSI collection. ANN channels are a subset of machine learning and are at the heart of deep learning algorithms. To retrain the channel, atomic norm minimization and ANN channel is ensured, the CE is carried out in two stages. With the help of same training overhead, the proposed technique can perform better than the passive ANN and RIS CE according to simulation findings. Proposed research work additionally manipulate the theoretical performance via CRLB analyses. It can be reduce based upon the mean square error (MSE).