Design of a Priority-Differentiated MAC Protocolfor UANETs Based on MAD3QN

Unmanned Aerial Vehicle Ad Hoc Networks (UANETs) have been widely deployed in both military and civilian applications. This paper proposes an adaptive priorityaware MAC protocol for UANETs, named Multi-Agent Priority Medium Access Control(MP-MAC), based on the Multi-Agent Double Dueling Deep Q-Network(MAD3QN) framework. Each UAV node is equipped with an individual D3QN model, which dynamically adjusts its transmission probability and contention window size according to task priority, historical transmission behavior, and current network conditions. This design ensures the throughput of high-priority nodes while flexibly accommodating low-priority nodes, thereby improving overall network throughput. The protocol is implemented and evaluated in the ns-3 simulator via the ns3-ai interface to emulate multi-agent communication in competitive wireless environments. Simulation results demonstrate that MP-MAC achieves higher network throughput and lower end-to-end delay compared to traditional Q-Learning-based MAC protocols, while also enhancing traffic differentiation capabilities.