CRAM: clustering-based resource availability measurement in mobile peer-to-peer systems

Abstract

The peer-to-peer (P2P) systems are an alternative to solve the scalability issue raised by the client/server systems. P2P systems are composed of a set of entities that communicate directly without any central server and constitute scalable and robust distributed systems. In these systems, an entity only has a partial knowledge about the overlay state. The challenge is to provide a global knowledge in the system regarding a feature as the resources availability. In this paper, we propose a clustering-based resource availability measurement called CRAM for mobile P2P networks which relies on knowledge from all entities. CRAM can be used in resource search and in replication strategies to improve the network performances. Simulation results show that our estimation of availability is close to the real one such as the deviation between them can equal 0.008. They also show that our algorithm reacts efficiently to the appearance or depletion of a resource replica in the system.