Q‐learning based flexible task scheduling in a global view for the Internet of Things

Abstract

SummaryWith billions of sensor‐based devices connected to the Internet of Things (IoT), it is a pivotal issue to design an effective task scheduling scheme when the resource of sensor nodes is limited. In the past, Q‐learning based task scheduling scheme which only focuses on the node angle led to poor performance of the whole network. Thus, a Q‐learning based flexible task scheduling with global view (QFTS‐GV) scheme is proposed to improve task scheduling success rate, reduce delay, and extend lifetime for the IoT. First, the Q‐learning framework, including state set, action set, and rewards function is defined in a global view so as to forms the basis of the QFTS‐GV scheme. Then, a task scheduling policy is established with distinguishing rewards for nodes in different areas of the network, so the energy‐strained nodes can be protected to ensure a high lifetime, and the energy‐relaxed nodes can increase their transmission power to promote the benefits of the whole network. Finally, experimental results demonstrate that the QFTS‐GV scheme can achieve a higher task scheduling success rate, lower delay, and less energy consumption. Compared with the Q‐learning based task scheduling scheme, the QFTS‐GV improves the task scheduling success rate by 1.42% to 7.13%, reduces the delay by 24.60% to 42.56%, and saves energy by 21.18% to 36.60%.