Effect of Mobility Models on Reinforcement Learning Based Routing Algorithm Applied for Scalable AD HOC Network Environment

Abstract

Mobile Ad Hoc Network faces the greatest challenge for better performances in terms of mobility characterization. The mobility of nodes and their underlying mobility models have a profound effect on the performances of routing protocols which are central to the design of ad hoc networks. Most of the traditional routing algorithms proposed for ad hoc networks do not scale well when the traffic variation increases drastically. To model a solution to this problem we consider a reinforcement learning based routing algorithm for ad hoc network known as SAMPLE. Most the scalability issues for ad hoc network performance investigation have not considered the group mobility of nodes. In this paper we model realistic group vehicular mobility model and analyze the robustness of a reinforcement learning based routing algorithm under scalable conditions. KEYWORDS Routing protocols, mobility models, scalability and reinforcement learning.