Global Optimisation Problems in P2P Networks

This page aims to collect different global optimisation problems arising in peer-to-peer networks.

Global optimisation is, generally speaking, the task of finding the absolutely best value of a (usually) nonlinear function under given constraints.

Neighbor-selection strategy

The paper

S.G.M.Koo, K.Kannan and C.S.G.Lee: On neighbor-selection strategy in hybrid peer-to-peer networks. Future Generation Computer Systems 22 (2006) 732-741

discusses the problem of neighbor-selection (NS) which is the process where one or more entities in the P2P network police the system and determines the "neighbors", the other peers that they will connect to for obtaining and/or distributing the content, for each peer. The NS process has great influence on the network efficiency. The related problem, which appers in the paper cited above, is to maximise the disjointness of content from peer i to peer j (which means the collection of content pieces that peer i has but peer j does not, or in another interpretation, means the pieces that peer i can upload to peer j). The proposed solution is to maximise the number of content pieces each peer can contribute to its neighbors by determining the connections between the peers. Using mathematical formalism this leads to a constrained integer programming formalism. The paper applies genetic algorithm to solve the problem.

Two quesions:

• (Not so important:) Can we use better algorithms to solve the problem?
• (More important:) The problem statement above is in networks with tracker server. Can we formulate the similar problem in a server-free network?