Optimal Recharge Scheduler for Drone-to-Sensor Wireless Power Transfer

Abstract

Wireless recharging by autonomous power delivery vehicles is an attractive maintenance solution for Internet of Things devices. Improving the operating efficiency of power delivery vehicles is challenging due to complex dynamic environments and the need to solve difficult optimization problems to determine the best combination of routes, number of vehicles, and numerous safety thresholds prior to deployment. The optimal recharge scheduling problem considers minimizing discharged energy of drones while maximizing devices' recharged energy. In this paper, a configurable optimal recharge scheduler is proposed that incorporates several evolutionary and clustering approaches. A modified version of the Black Hole algorithm is presented, which is shown to execute on average 35% faster than the state of the art genetic approach, while delivering comparable performance in simulation across 18 scenarios with varying area and density of sensor nodes deployed under different initialization scenarios.

Practical Mission Planning for Optimized UAV-Sensor Wireless Recharging

“…Mission planning in a Wireless Rechargeable Sensor network [19] generally involves energy supply management of sensor nodes [2], [20], [21], [22], [23], [24] and autonomous vehicles [25], [26], data gathering from sensor nodes [3], [27] and surveillance [28], [29]. Our focus is to solve the optimization problems raised in the scenario of sensor node energy replenishment as a maintenance task.…”

“…Cheng et al apply Earlier Deadline First (EDF)-based and Nearest Job First (NJF)-based clustering strategies to assign the number of mobile chargers, and a GA to compute the near-optimal solution. In a more general case, Qian et al assign the task to each vehicle with a K-Means Clustering initialization, which allows for a flexible number of vehiclenode visitations in each route plan [21]. Simulated results in [21] show that the proposed BHA could deliver comparable performance, with faster execution than a state-of-the-art GA.…”

“…In a more general case, Qian et al assign the task to each vehicle with a K-Means Clustering initialization, which allows for a flexible number of vehiclenode visitations in each route plan [21]. Simulated results in [21] show that the proposed BHA could deliver comparable performance, with faster execution than a state-of-the-art GA. These clustering strategies do not, however, account for the recharging capability of the autonomous vehicles, which may result in infeasible missions being assigned to energy-constrained vehicles.…”

“…This paper develops the optimal mission planner for autonomously recharging wireless sensors using a DJI M100 quadrotor drone as the PDV. We select temperature sensor LMT84 [37] with capacitor HB1030-2R510 [39] and pressure sensor NPA300 [38] with capacitor PHB-5R0 [40] as the sensor node prototypes, following [20] and [21]. Specifications of sensor nodes and the UAV are listed in Table . 1.…”

“…CBHA (see Alg. 1) comes from a modified version given in [21], which includes N pop members in the population to evolve N gen generations. Population solutions are initialized from the same initial solution with a length N sol .…”