Q-Learning-Based Traffic-Aware Parent Selection for Wireless Powered Sensor Networks

This paper presents a Q-learning-based traffic-aware parent selection for wireless powered sensor networks (WPSNs), abbreviated QTaPS. The QTaPS employs Q-learning to independently select power and data parents, thereby improving the efficiency of energy harvesting and data transmission in WPSNs. To this end, QTaPS uses two Q-learning agents with different reward functions to select power and data parents, respectively. In order to consider the characteristics of each type of traffic, different metrics are used in each reward function. The simulation results showed that QTaPS obtained better performance than the existing scheme in terms of aggregate throughput and average end-to-end delay.