Priority-Aware Coordinated Anti-Jamming Channel Access in Multi-Band Wireless Networks: A Game-Theoretic Hierarchical Learning Approach

Abstract

Multi-band networks (MBNs) face severe challenges from dynamic jamming and heterogeneous user priorities, which can degrade communication reliability and quality of experience (QoE). In this paper, we propose a hierarchical learning algorithm, Priority-based Multi-band Anti-jamming Access Learning (PMAAL), to jointly address anti-jamming and priority-aware channel coordination in a distributed manner. The framework separates the problem into two layers: a lower layer employing Multi-Agent Independent Q-learning (IQL) to learn jammer patterns and construct a Recommended Channel Set (RCS), and an upper layer formulating channel access within the RCS as an exact potential game, solved efficiently by the Priority-based Win-Adapt Lose-Shift (P-WALS) algorithm. Furthermore, a satisfaction feedback mechanism is introduced to propagate user QoE information back to the lower-layer learning module, enabling dynamic refinement of the RCS by eliminating channels that repeatedly fail to satisfy communication requirements. A priority-driven backoff mechanism ensures higher priority users gain preferential access, directly enforcing user priorities and QoE requirements. Simulation results demonstrate that PMAAL converges rapidly and outperforms benchmarks in network-wide weighted satisfaction, confirming its effectiveness in resilient, priority-aware communication under adversarial jamming.