Resource Efficient Actuation in UAV-aided Sensor-Actuator Networks

Integrating uncrewed aerial vehicles (UAVs) into wireless sensor-actuator networks (WSANs) offers flexibility and improved system performance. However, imprecise localization and limited battery capacity constrain the UAV operation. This paper explores the use of battery swap station (BSS)-assisted UAV to facilitate timely actuation in WSANs while addressing the above limitations. The UAV collects data via backscatter communication from the sensor nodes and delivers to the energy-constrained actuator nodes along with the required energy. Incorporating UAV location uncertainty and Nakagami-m wireless channel fading, closed-form expressions are derived for the ergodic capacity in backscatter communication and the expected energy harvesting rate. To minimize the maximum delay in actuation, an optimization problem is formulated. To reduce complexity, the problem is transformed into an equivalent node visit sequence optimization and solved using sequential deep reinforcement learning (SDRL). We verify the accuracy of our analysis through Monte Carlo simulations. Our results show that the proposed SDRL-based strategy consistently offers reduced actuation delay with a significantly small computation overhead.