Machine Learning-Based Resource Allocation for Multi-UAV Communications System

The unmanned aerial vehicle (UAV)-based wireless communication system is prominent in its flexibility and low cost for providing ubiquitous connectivity. In this work, considering a multi-UAV communications system, we propose to utilize a machine learning-based approach to tackle the trajectory design and resource allocation problems. In particular, with the objective to maximize the system utility over all served ground users, a joint user association, power allocation and trajectory design problem is formulated. To solve the problem caused by high dimensionality in state space, the machine learning-based strategic resource allocation algorithm comprising of reinforcement learning and deep learning is presented to design the optimal policy of all the UAVs. Extensive simulation studies are conducted and illustrated to evaluate the advantages of the proposed scheme.