AI-assisted mmWave beamforming for QoS enhancement in networked multimedia systems

Millimeter-wave (mmWave) communication offers ultra-high data rates for next-generation multimedia systems but remains vulnerable to beam misalignment, interference, and rapid channel variations. To address these challenges, we propose an AI-assisted mmWave beamforming optimization framework that leverages a deep artificial neural network (ANN) to enhance quality of service (QoS) in a vehicular network. The proposed ANN performs real-time QoS prediction using features such as signal-to-noise-plus-interference ratio (SNIR), throughput, latency, jitter, packet loss, PDR, and energy consumption. It also incorporates dropout-based regularization and early stopping for improved robustness. SHapley Additive exPlanations (SHAP) further identifies SINR, PDR, and throughput as the most influential QoS factors. The numerical results show that the proposed framework achieves an R2 score of 0.9941 and significantly reduces MSE and RMSE compared with traditional machine learning models.