A systematic review and Q-learning-based design of scheduling functions for 6tisch networks.

Abstract

An IPv6 over the TSCH mode of IEEE 802.15.4e (6TiSCH) network provides IPv6 connectivity through IEEE 802.15.4 links governed by Time Slotted Channel Hopping (TSCH). TSCH is a medium access control for low-power and lossy networks, providing low energy consumption, high reliability, and deterministic latency through time-division multiplexing. To achieve this goal, 6TiSCH defines a component responsible for determining the best communication scheduling of devices, called an Scheduling Function (SF). The design and implementation of SFs, being context-dependent, is a current topic of study in the literature. Thus, many different scheduling functions were proposed, each with its particular trade-offs. Additionally, Artificial Intelligence (AI), in particular machine learning, emerges as a prominent tool for its capacity to promote adaptability and flexibility. Although previous works have proposed comparisons of different scheduling strategies, the systematization of AI algorithms for 6TiSCH has not been explored in detail. This work proposes such a review, presenting an analysis of the current state of AI-based scheduling methods. Additionally, this work advances the state of the art by presenting, evaluating, and comparing two new Q-learning SFs with the current state of the art of SFs for 6TiSCH. The experimental results show the promising potential of the proposed approaches.