Wireless-Powered Multi-Channel Backscatter Communications Under Jamming: A Cooperative Reinforcement Learning Approach

Wireless-Powered Backscatter Communication (WPBC) is emerging as a promising technology for battery-less solutions to many Internet-of-Things (IoT) applications. In this paper, we study the channel selection and operation mode control problem in a multi-channel WPBC system that turns jamming signal into energy harvesting opportunities. We propose a cooperative reinforcement learning (RL) approach that enables multiple agents, namely the access point (AP) and backscatter device (BD), to exploit unknown jamming pattern present. The learning of jamming channels and patterns not only enables the AP to communicate with the BD with interference-free but also empowers the BD to harvest energy from the jamming signals. Additionally, we address the limitation of BDs, which cannot decode information and harvest energy simultaneously, by incorporating this constraint into our design of a cooperative RL approach. This innovative strategy empowers the BD to make intelligent decisions regarding its operational mode-whether to prioritize communication or energy harvesting. Our aim is to optimize the trade-off between throughput and energy harvesting efficiency, ensuring that data reception requirements are met while adhering to constraint on the energy level stored in the battery. Unlike traditional approaches that only consider paired states and actions, our proposed cooperative RL algorithm incorporates channel state, jamming state, and action. These elements represent channel operation, jamming experience, and channel selection, respectively. The awareness of jamming experience derived from the jamming state enables the AP agent to select an action that can evade jamming. The proposed scheme experienced low computation and storage overhead as an inherent feature of the algorithm. Remarkably, the results show that the proposed method achieves optimal performance under static and round-robin jamming, and even in scenarios of random jamming.