A Q-Learning-based distributed routing protocol for frequency-switchable magnetic induction-based wireless underground sensor network

Magnetic Induction (MI) based Wireless underground sensor networks (WUSNs) consist of magnetic-antenna sensors that are buried in and communicate through soil, considered as part of full-coverage 6G wireless system. The wireless channel in MI-WUSNs appears a considerable frequency-selective property due to those conductivity objects in underground environment. This results in that a single frequency is almost impossible to ensure that all sensors in MI-WUSNs are both connected. To this end, the frequency-switch strategy is introduced into MI-WUSNs to cope with its frequency-selective property and ensure network connectivity, forming frequency-switchable MI-WUSNs. In this paper, the distributed routing protocol for frequency-switchable MI-WUSNs is studied to contribute to the network design of underground wireless systems. To be specific, we first map the frequency-switchable MI-WUSN to a multi-layer network and give a description about its routing decision in the Q-Learning framework. Then, we discuss several special designs in system implementation for the proposed routing protocol. In addition, simulation experiments are also considered to verify the convergence and effectiveness of our routing protocol.(c) 2022 Elsevier B.V. All rights reserved.