Multichannel spectrum access based on reinforcement learning in cognitive internet of things

Abstract With the development of Internet of Things(IoT), the demands for communication spectrum have increased rapidly, resulting in the shortage of limited spectrum resources. Cognitive IoT (CIoT) based on cognitive radio (CR) can improve the spectrum utilization by accessing the idle spectrum licensed to a primary user (PU). In this paper, a multichannel spectrum access scheme based on reinforcement learning (RL) is proposed to improve the spectrum access of CIoT, wherein the CIoT can use multiple channels for transmissions to reduce the communication interruptions. The channels are ranked in the decreasing order of their predicted idle probabilities, which can make the CIoT find enough idle channels quickly via decreasing the number of sensing operations and spectrum handoffs. The simulation results show that our proposed scheme is superior to the single-channel spectrum access scheme in terms of throughput, communication interruption, average collision probability and average spectrum switching frequency.