A Dynamic Approach for Handling UAV in Crowded Environment Using Deep Q-Networks

The demand for wireless connections is rising, and the advent of mobile internet has sparked a hunt for fresh, creative ways to expand the capacity and reach of cellular networks. One strategy to provide ground mobile users with long-term communication coverage is to employ Unmanned Aerial vehicles (UAV) as mobile base stations. The proposed method incorporates a Deep Reinforcement Learning (DRL) method to address the challenges that persist with traditional approaches. We introduced Deep Q-Learning with an attention mechanism-based multi-agent model to develop an energy-efficient control system for UAVs in cellular networks. Due to the proposed strategy’s dynamic nature, it efficiently handles the crowded environment by resolving scalability and mobility issues. The researchers evaluated the proposed model using the benchmark dataset, producing promising results. Further, the researchers tested the proposed model using the sum rate metric to analyze the capability of the proposed algorithm to provide a reasonable throughput rate.