Reinforcement learning search mechanism in unstructured peer to peer (p2p) networks
Loading...
Date
2017
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Faculté des Mathématiques et de l’Informatique - Université Mohamed BOUDIAF - M’sila
Abstract
Peer to peer network is one of the ambient and newly fields that take an important place in informatices networking search fielde .existing searching schemes in unstructured p2ps can be categorized as either deterministic or probabilistic .the quality of query results in deterministic schemes is low .probabilistic schemes use simple heuristics that lack the theoretical background to support more accurate and Well results .in this thesis ,we propose to improve searching by reinforcement learning (RL),which has been proven in artificial intelligence to be able to learn the best sequence of actions in order to achieve a certain goal.Our approach, RLKRW (reinforcement learning K random walker ), aims at locating the best path to desired files by exploiting the traffic change .it explores new paths by forwarding queries to K randomly chosen neighbors.it also exploits the paths that have been discovered to reduce the cumulative query cost .its experimental result supports the performance improvement of RLKRW .
Description
Keywords
peer to peer network,reinforcement learning,RLKRW ,unstructured p2ps