Improving the performance of P2P routing using adaptive machine learning

Abstract

Effective Peer to Peer (P2P) routing is a very complex problem to solve, due to the fact that the nodes are not location aware, and the network is constantly changing in terms of number of nodes, link quality and other network parameters. For such adhoc networks protocols like adhoc on demand distance vector routing (AODV), AODV with multi-channel (AOMDV), viceroy and others have been proposed. These protocols show a moderate to low quality of service (QoS) for the network when routing is concerned, in some scenarios these protocols have low delay, while in some other scenarios these protocols have better energy efficiency than others. In this paper, we propose a delay and energy efficient machine learning based routing protocol for P2P networks, which allows the network designers to improve upon the network QoS without compromising on the routing efficiency of the network. The results indicate that our proposed protocol is better in terms of both delay and energy efficiency when compared to the existing ones, and provides alternate routing paths, in case of node or link down scenarios. We also propose some further work which can be taken up by using the proposed protocol to make the network more secure and efficient in terms of overall privacy.