A WiSARD Network Approach for 5G MIMO Beam Selection

The integration of context information and machine learning techniques can enhance the capabilities of 5G/6G networks when dealing with the beam selection problem. This paper proposes the use of a Weightless Neural Network (WiSARD) with multimodal data as input to address this problem. The performance of the WiSARD is compared to classic machine learning algorithms (KNN, Decision Tree, SVC, Random Forest) based on the top-k accuracy in a vehicular network. The simulation results indicate that the WiSARD is a competitive method for this scenario and can be a valuable asset for future cellular networks.