Intelligent mechanism for communication in P2P networks

Abstract

Peer-to-Peer networks have become a major research topic over the last few years. Object location is a major part in the operation of these distributed systems. In this work, we present an overview of several search methods for unstructured peer-to-peer networks, in unstructured P2Ps can be categorized as either blind or informed schemes in a blind search, nodes do not keep information about data location. In an informed search, nodes store some metadata that facilitates the search, so we propose to improve searching in the blind schemes by using the idea of Reinforcement Learning and applied to each element in the network, each node in our proposal RLKRW (Reinforcement Learning Iterative Deepening Search) learn from the past researches other nodes we achieved good results.