Deep Reinforcement Learning for Joint Channel Selection and Power Allocation in Cognitive Internet of Things

摘要

With the development of wireless communication technology and the lack of spectrum resources, it is very meaningful to study the dynamic spectrum allocation in the cognitive Internet of Things. In this paper, the system model is firstly established. In an underlay mode, considering the interference between primary and secondary users, jointing channel selection and power allocation, aiming to maximize the spectrum efficiency of all secondary users. Different from the traditional heuristic algorithm, the underlay-cognitive-radio-deep-Q-network frame-work (UCRDQN) based on deep reinforcement learning, is proposed to find the optimal solution efficiently. The simulation results show that the UCRDQN algorithm can achieve higher spectrum efficiency and is more stable and efficient than other algorithms.