Output details
11 - Computer Science and Informatics
Imperial College London
Fluid analysis of energy consumption using rewards in massively parallel Markov models
<13>Acceptance 17%/110 This was the first paper to apply rapid fluid analysis to the problem of deriving accumulated measures during operation of large stochastic behavioural systems. Analysis of this sort on explicit state-space systems was previously limited to O(10^3) states. This is crucial for optimising energy usage and other dynamic cost measures in complex industrial applications and leads naturally to design-time minimisation of such measures while maintaining operational service level agreements. Accumulated reward minimisation techniques are now implemented in the open-source GPA tool, http://code.google.com/p/gpanalyser/. An important output for EPSRC project on Analysis of Massively Parallel Stochastic Systems (AMPS), EP/G011737/1.