Channel BlaQLisT: Channel Blacklist using Q-Learning for TSCH

In time-slotted channel hopping (TSCH) communications included in IEEE 802.15.4, there are no direct scheduling guidelines, but communication schedules are determined by the distribution of slots according to Slotframes and the allocation of channels. Among them, channel allocation is directly exposed to effects such as frequency mix caused by the increasing number of wireless communication equipment. In this paper, we propose a technique to create a Hopping sequence list based on the Blacklist according to the channel quality of each slot by directly matching the Q-table of Q-learning, a technique of reinforcement learning, to maintain the quality and maximize the efficiency of communication. To verify the above technique, Contiki-NG's Cooja simulation was verified, and one of the existing studies was selected as a comparison group. Simulation results show higher PDR and energy charge compared to previous studies, showing relatively good efficiency, reliability, and the potential of this technique.