A New Method Using Machine Learning for Resource Management in Unmanned Aerial Vehicle (UAV) Networks

A new method for resource and energy management in Unmanned Aerial Vehicle (UAV) networks has been introduced, utilizing machine learning to optimize resource allocation and enhance the performance of UAV networks. This method involves simulating the network with auxiliary drones and employing the Kmeans algorithm to determine optimal locations for macro-cells. The Walfisch-Ikegami propagation model is used to calculate losses, and radio resource allocation is based on SINR to ensure communication quality in densely populated areas. Results indicate an increase in the accuracy and efficiency of the proposed model compared to traditional models. Additionally, the network's performance in managing drones and improving services in highdensity areas has been enhanced. With UAV-BS support, a significant reduction in overhead and optimal distribution of drones in dense macro-cells has been observed. Compared to the General Regression Neural Networks (GRNN) model, the proposed model demonstrates higher accuracy in resource prediction using MAE and RMSE metrics, indicating its superiority in optimal resource utilization. Utilizing drones as Base Stations (UAV-BS) in populated areas can lead to significant improvements in coverage and reduced network latency.