A New Distributed Dynamic Spectrum Access Model Based on DQN

Abstract

With the rapid growth of wireless communication services, spectrum resources are increasingly scarce. Cognitive radio and dynamic spectrum access technologies are valued for their ability to improve spectrum efficiency. This paper first gives the classification of dynamic spectrum access technology, then analyzes the research status and shortcomings of dynamic spectrum access technology, and finally proposes a non-cooperative distributed dynamic spectrum access model based on deep Q-Learning. Compared with the existing model, this new model not only adopts overlay access, but also combines with underlay access, and recognizes that users can conduct multichannel access. Compared with the single-channel access model, the bandwidth utilization is improved under the same condition.