Interference Avoidance Scheme Based on Reinforcement Learning

This paper proposes an interference avoidance scheme based on reinforcement learning for the communication stations exposed to interference. Firstly, we design the signal to interference noise ratio-mean opinion score (SINR-MOS) as the reward function to obtain optimal channel handoff strategy through reinforcement learning. Then, two specific methods of interference avoidance are proposed by applying Q-learning algorithm, which enables an automatic avoidance by switching to a new idle channel when the station is interfered. Finally, it is proved by computer simulation experiments that the methods have a low miss detection rate and false alarm rate, showing good interference avoidance performance.