Heuristic Algorithms for Optimization of Data Distribution in P2P Networks

Peer-to-Peer (P2P) systems have gained much popularity over the past few years. P2P networks contribute a significant portion of the Internet traffic and cause increasing costs incurred by Internet service providers (ISP). Present P2P systems are generally selecting peers in random manner, regardless of the network cost, which generates much redundant traffic over the core network and congest the Internet. Therefore the need to optimize P2P networks arises. The following article presents the research on heuristic algorithms to optimize distribution problems in P2P networks. We propose a construction algorithm called backward cost selection and a tabu search algorithm. Simulation comparison against optimal results yielded by CPLEX solver shows that both algorithms produce results close to optimal ones, while the execution time is significantly lower than CPLEX.