Title: An ant-colony approach for the design of optimal chunk scheduling policies in live peer-to-peer networks

Authors: Pablo Romero; Franco Robledo; Pablo Rodríguez-Bocca

Addresses: Departamento de Investigación Operativa, Instituto de Computación, Facultad de Ingeniería, Universidad de la República, Julio Herrera y Reissig 565, Montevideo, Uruguay ' Departamento de Investigación Operativa, Instituto de Computación, Facultad de Ingeniería, Universidad de la República, Julio Herrera y Reissig 565, Montevideo, Uruguay ' Departamento de Investigación Operativa, Instituto de Computación, Facultad de Ingeniería, Universidad de la República, Julio Herrera y Reissig 565, Montevideo, Uruguay

Abstract: Peer-to-peer networks are self-organised communities over the internet infrastructure, in which peers are both clients and servers. The global resources of a peer-to-peer network increase proportionally with the population, promoting scalability. Peers are organised covering neighbouring-strategies and chunk-scheduling policies that determine the success of the cooperation scheme. In this paper, we address the design of chunk-scheduling policies in a cooperative scenario, assuming a complete mesh-topology under regime. All users wish to display a video channel with no cuts and low buffering times. We propose an in-depth analysis of this cooperative system, and develop the best chunk scheduling policy so far, found via a sophisticated ant-colony-based exploration. We introduce the new policy into a real platform. There, users wait five seconds to start watching following our new policy (versus minutes in previous policies), with acceptable number of cuts.

Keywords: combinatorial optimisation problem; COP; travelling salesman problem; TSP; ant colony optimisation; ACO; peer-to-peer networks; P2P networks; chunk scheduling policy; cooperation; mesh topology.

DOI: 10.1504/IJMHEUR.2013.054136

International Journal of Metaheuristics, 2013 Vol.2 No.2, pp.101 - 122

Available online: 23 May 2013 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article