RIS-assisted Channel Prediction Model for Mine V2X Systems

The complex and dynamic environment of mines leads to time-variability and uncertainty in channel propagation, which affects the stability and reliability of Vehicle-to-Everything (V2X)communication in mine networks. Channel prediction can help mitigate these issues. Introducing reconfigurable intelligence surface (RIS) into the mine V2X system can enhance signal coverage and quality in the complex mine environment. To address the issue of long-term error accumulation in existing channel prediction models, this paper analyzes the prediction error of channel state information (CSI) in mine V2X systems. It was observed that the error distribution exhibits characteristics of an approximate normal distribution, revealing its predictability. Utilizing this characteristic, this paper proposes an error correction strategy to suppress long-term error accumulation during the prediction process. Furthermore, an adaptive method for optimizing RIS phases is designed based on the Snow Ablation Optimizer. Specifically, this paper utilizes a metaheuristic optimization algorithm and deep learning techniques to propose a mine V2X channel prediction model that incorporates the error correction strategy. Experimental results show that optimizing the RIS phase can effectively enhance channel gain, and the mine V2X channel prediction model incorporating the error correction strategy can effectively suppress cumulative errors during the prediction process, thereby improving prediction accuracy.