Multi-UAV Cooperative NOMA User Assignment and Power Allocation Optimization Based on Proximal Policy Optimization

For the UAV relay collaborative NOMA system, power allocation is one of the core challenges to improve spectral efficiency. the goal of maximizing the minimum transmission rate of users is achieved by establishing a Markov chain, and the fairness among the users is also ensured. We propose multi-UAV cooperative power allocation method based on proximal policy optimization (PPO) for practical scenarios involving measurement errors in user location information. Simulation results show that the optimization effect of our designed PPO algorithm improves about $3.57 \%$ and is more stable compared to the Deep Q-Network (DQN) algorithm, considering the effect of user’s location error.