AMPS - Analysis of Massively Parallel Stochastic Systems

From March 2009 to August 2012
Project Leader(s): Dr. Nigel Thomas
Staff: Prof. Aad van Moorsel, Mr. James Turland, Mr. Eason Zhao, Mr. Xiao Chen
Contact: Dr. Nigel Thomas
Sponsors: EPSRC
Partners: Imperial College

The key technical aim of AMPS is to be able to perform quantitative response-time analysis of massively parallel real world systems, through a combination of compositional analysis and fluid approximation. The main focus of AMPS case studies will be on two classes of computer-communication system where, in both cases, massive parallelism is an inherent feature: software architectures (publish/subscribe architectures for distributed and mobile applications; file sharing applications over peer-to-peer networks; large-scale Internet caching infrastructures) and network and computer security-related models (Internet worm models; distributed voting models; performance of trust establishment in Web of Trust; trust economics decisions in large enterprises).