| （英） |
In recent years, with the development of the Internet of Things (IoT), the number of devices performing wireless communication has rapidly increased. In an environment where a variety of devices are mixed, conventional general-purpose protocols cannot cope with changes in the environment and the operation of multiple applications, which causes performance degradation. Therefore, in this study, we investigate a method for autonomous protocol control and application parameter selection by learning communication logs as input values using DQN, a kind of deep reinforcement learning. In the proposed method, in a protocol based on Carrier Sense Multiple Access/Collision Avoidance (CSMA/CA), learning is performed by collecting information such as the number of devices and the number of retransmissions, and functions such as Carrier Sense and parameters such as RTS/CTS are autonomously selected according to the environment. In addition, it adjusts the parameters of each application in an environment where multiple applications are running. It is confirmed that the proposed method can obtain high QoE compared with conventional CSMA/CA. |