Co-Designing TWT Scheduling and RF Energy Harvesting for Sustainable Heterogeneous Wi-Fi: A Deep Reinforcement Learning Approach

Abstract

IEEE 802.11ax Target Wake Time (TWT) coordinates station sleep and wake schedules for power savings, while RF energy harvesting promises battery-free operation for IoT sensors. Though studied in isolation, the two are coupled; the same access point (AP) that schedules TWT service periods, controls the RF harvest of battery-free stations. We present, to our knowledge, the first framework to co-design them. It features (i) a standard-compliant, datasheet-fidelity ns-3 model of off-the-shelf PowerCast P21XXCSR hardware; and (ii) a deep reinforcement learning scheduler that jointly sets the TWT schedule, the station-to-group assignment, and a novel Power Delivery Window (PDW) that transmits RF energy to sleeping harvesters. The AP is blind to device class and uses only protocol-compliant signals to ensure deployability. Because the PDW reward is delayed through stored charge, an asymmetric critic sees the oracle energy state only during training, while the recurrent, &agr;-fair actor stays realistic. Against a principled online M/D/1 scheduler, while reclaiming slightly less harvested energy, our proposed scheduler consumes 1.9× less and serves 2.8× more harvester demand per joule, raising fairness (+0.11 Jain) and winning the &agr;-fair welfare in 89% of scenarios.