Bio-inspired Search and Distributed Memory Formation on Power-Law Networks

In this paper, we report a novel and efficient algorithm for searching P2P networks having a power law topology. Inspired by the natural immune system, it is a completely decentralized algorithm where each peer searches by sending out random walkers to a limited number of neighbors. As it finds other peers having similar content, it restructures its own neighborhood with the objective of bringing them closer. This restructuring leads to clustering of nodes with similar content, thus forming P2P communities. Alongside, the search algorithm also adapts its walk strategy in order to take advantage of the community thus formed. This search strategy is more than twice as efficient as pure random walk on the same network.