An implementation methodology of geographical location based Peer-to-peer networks for managing ubiquitous sensors

Conventional Peer-to-peer networks have been focusing on managing large size files, such as musics, movies, or DVD images. On the other hand, huge numbers of tiny data are generated by sensing devices existing all over the globe. To manage these types of data, P2P networks should shift the strategy to construct an overlay and process queries, because lots of queries are relayed through overlay networks and these queries cause serious problems. In this paper, we propose implementation methodology considering the characteristics of queries on ubiquitous sensing environment and clustering sensors and users dynamically. Clustering sensor data stream and supporting event driven behaviors substantially reduce relaying queries. Evaluations show that our P2P network could manage tens millions sensing devices by thousands of P2P nodes.