Deep Graph Reinforcement Learning Based Intelligent Traffic Routing Control for Software-Defined Wireless Sensor Networks

Software-defined wireless sensor networks (SDWSN), where the data and control planes are decoupled, more suited to handling big effectively monitoring dynamic environments events. To overcome limitations of using static routing tables under high traffic intensity, such as network congestion, packet loss rate, low throughput, etc., it is critical design intelligent for SDWSNs. In this paper we propose a deep graph reinforcement learning (DGRL) model-based scheme SDWSNs, which combines convolution with deterministic policy gradient. The model fits well task SDWSN, process forwarding can be regarded sampling continuous action space has strong features. policies made by SDWSN controller implemented at nodes optimize process. Simulation experiments performed on Omnet++ platform show that, compared existing algorithms proposed method reduce transmission delay, increase delivery ratio, probability congestion.