Search and index in locality-based clustering overlay

A locality-based clustering peer-to-peer overlay networks (LCO) architecture is introduced in this paper. LCO differs from the pure unstructured P2P networks such as Gnutella in two key aspects. First, LCO partitions peers into clusters such that peers belonging to the same cluster are relatively close to one another in terms of network latency. Multiple floods are initiated for one query, with the character that each flood is restricted within one cluster, hence reducing the unnecessary traffic produced by the topology mismatching between the P2P logical overlay network and the physical underlying network. Second, an efficient inter-cluster index scheme is used in LCO such that the search scope can be retained even though only a few clusters are directly probed. Our simulation results indicate that LCO is efficient in both resource usage and data retrieval.