Adaptive Interest Managment in Distributed Virtual Environments Thesis Proposal Rsmg Report 3

G. Theodoropoulos

Abstract:This report proposes an investigation of an adaptive architecture for interest management in large-scale distributed virtual environments (DVEs). DVEs, like many distributed systems suffer from large amounts of extraneous network communication in order to ensure replicated data remains consistent. Techniques for minimisation of this overhead by specifying data subscriptions on a ‘per-node’ basis exist and are collectively termed ‘interest management’ systems. Existing techniques tend to require highly complex and application-specific ‘interest expressions’ in order to determine what events must be distributed and to where. The architecture proposed herein uses a run-time analysis of application behaviour to automatically derive optimistic ‘interest expressions’ without the need for application-specific knowledge or the intervention of the application developer.