Query Routing with Ants?

In this paper we propose SemAnt, a novel ant-based algorithm designed for query routing in taxonomy-based peer-to-peer environments. We introduce the reader to the pheronome trail-laying-andfollowing behaviour observed from natural ants and show how it can be applied to query routing in peer-to-peer networks. Our proposed algorithm accounts for network parameters such as bandwidth and latency and optimizes the pheromone trails to results for a given query depending on the query’s popularity. If a query is common, its pheromone trails will converge and lead to the nodes that offer the most results for the given query. Pheromone trails that are used rarely will evaporate over time. In addition, the proposed algorithm accounts for the inherent dynamics of peer-to-peer networks by adjusting pheromone trails when peers join or leave the network.