Harnessing 1D-CNN for Received Power Prediction in Sub-6 GHz RIS: Part I

In this study, we use a one-dimensional Convolutional Neural Network (1D-CNN) deep learning model to predict the power levels received at the receiver (Rx) in a wireless communication system, taking into account different beamsteering angles (azimuth and elevation). The received power depends on several factors, such as the beamforming method, the distance between the transmitter-RIS and RIS-receiver, and the surrounding environment. We leverage the 1D-CNN’s ability to capture spatial relationships along the beamsteering angles to accurately predict changes in power in a dynamic channel. By training the model on an available dataset that includes received power levels for various beamforming angles, the 1D-CNN detect patterns and trends in power variations, which are essential for optimizing beam selection and enhancing communication performance. The proposed model shows promising results in predicting received power levels with high accuracy compared to other inline deep learning models across all k-fold cross-validation, helping to allocate resources more efficiently and improve overall system performance.