Content-Based Information Routing and Retrieval in Cluster-Based P2P Overlay Networks

Similarity searching is particularly important in distributed networks such as P2P systems, which use various routing schemes to submit queries to relevant peers. We investigate content-based information routing and retrieval using similarity search in clustered P2P overlay networks and focus on their maintenance cost models and performance issues. We present a query model for a cluster based P2P overlay network, called HON, where data and peers are organized in high dimensional feature spaces. We show through extensive simulations that HON has a low maintenance cost and is resilient to peers' failures. In addition it achievesan efficient similarity search with a high success rate.