Swarm-Based Algorithms for Decentralized Clustering and Resource Discovery in Grids

In this thesis, some novel algorithms based on swarm intelligent paradigm are proposed. In particular, the swarm agents, was exploited to tackle the following issues: - P2P Clustering. A swarm-based algorithm is used to cluster distributed data in a peer-to-peer environment through a small worlds topology. Moreover, to perform spatial clustering in every peer, two novel algorithms are proposed. They are based on the stochastic search of the ocking algorithm and on the main principles of two popular clustering algorithms, DBSCAN and SNN. - Resource discovery in Grids. An approach based on ant systems is exploited to replicate and map Grid services information on Grid hosts according to the semantic classi cation of such services. To exploit this mapping, a semi-informed resource discovery protocol which makes use of the ants' work has been achieved. Asynchronous query messages (agents) issued by clients are driven towards "representative peers" which maintain information about a large number of resources having the required characteristics.