Grid-Based Contraction Clustering in a Peer-to-Peer Network

Clustering is one of the main data mining techniques used to analyze and group data, but often applications have to deal with a very large amount of spatially distributed data for which most of the clustering algorithms available so far are impractical. In this paper we present P2PRASTER, a distributed algorithm relying on a gossip–based protocol for clustering that exploits the RASTER algorithm and has been designed to handle big data in a decentralized manner. The experiments carried out show that P2PRASTER returns perfect results under both optimal and non-optimal conditions, and also provides excellent scalability.