Modeling Performance Evaluation of Reinforcement Learning Based Routing Algorithm for Scalable Non-cooperative Ad-hoc Environment

Scalable performance analysis of routing protocols for ad-hoc network reveals the hidden problems of routing protocols in terms of performances. Wireless nodes in ad-hoc networks may exhibit non-cooperation because of limited resources or security concerns. In this paper we model a noncooperative scenario and evaluate the performance of a reinforcement learning based routing algorithm and compare it with ad-hoc on-demand distance vector a de facto routing standard in ad-hoc networks. Mobility models play an important role in ad-hoc network protocol simulation. In our paper we consider a realistic optimized group mobility model to aid the performance of the reinforcement learning based routing algorithm under scalable non-cooperative conditions.