AI-Driven Jamming Detection and DRL-Based Anti-Jamming for Ambient Backscatter Communications in 6G Networks

Abstract

Abstract: This paper proposes an energy-efficient Deep Reinforcement Learning (DRL)-based antijamming framework for backscatter-enabled Internet of Things (IoT) systems in next-generation
wireless networks. The proposed approach adopts a Deep Q-Network (DQN) with a well-defined
state-action space and a unified reward function to jointly optimize channel selection, power control,
and reflection coefficient under dynamic jamming conditions. Simulation results demonstrate that
the proposed method converges significantly faster than the tabular Q-learning baseline, achieving
stable performance within approximately 300 episodes. In terms of throughput, the proposed framework achieves up to 3.5 bps/Hz at low jammer power and maintains about 2.0 bps/Hz under strong
jamming (20 dBm), providing a gain of approximately 25% over Q-learning and more than 80%
compared to heuristic baselines. The proposed method also improves energy efficiency by around
20% and achieves up to 15–20% higher detection accuracy across different SNR levels. These results validate the effectiveness, robustness, and adaptability of the proposed DRL-based framework
for intelligent anti-jamming in resource-constrained IoT environments.
Keywords: AI-based jamming detection, Ambient Backscatter Communication, Anti-jamming
communication, Deep Reinforcement Learning, Energy-efficient wireless networks.