MAC Protocol Adaptation Method in Coordination with Application

In recent years, with the development of wireless communication technologies, various applications use the same networks and complexity of the networks becomes large. Until now, the protocols and parameters used in networks have been determined by humans based on their experience, but the performance degradation due to the inability to adapt to various applications in various environment especially in wireless condition. To solve this problem, a wireless MAC protocol is needed that can adapt to various environments and applications by allowing the network to control itself autonomously. The proposed method collects the information of each node in the gateway (GW) and learns by deep reinforcement learning DQN to adaptively select the parameters of MAC protocols according to the environment. In addition, assuming an environment in which a video streaming application using adaptive bitrate (ABR) is running on each node, the proposed method improves QoE by selecting an appropriate bitrate. We show that the proposed method improves the average throughput and QoE compared to CSMA/CA without parameter adjustment. To evaluate the performance of the proposed method, we compare the performance of the proposed method with CSMA/CA without parameter tuning by using computer simulations. From the simulation results, it is confirmed that the proposed method can obtain higher average throughput and QoE than the protocol without parameter adjustment.