Analysis of factors influencing the efficiency of applying reinforcement learning to routing control in wireless mesh network

Abstract

The application of reinforcement learning to routing control in mesh wireless network is attracting the attention of many research groups. Recent works have shown that reinforcement learning-based routing is more efficient than traditional routing protocols, especially in mesh wireless network models with heavy traffic loads, wide area, and multiple routers. In this paper, we focus on investigating the factors that affect the effectiveness of applying reinforcement learning to routing control. Thence, we provide appropriate recommendations for improving network performance when employing reinforcement learning-based routing protocols.