A Grid Enabled Problem Solving Environment for Monte Carlo Matlab Simulations

Abstract

This paper presents a problem solving environment (PSE) for the grid-enabled execution of Monte Carlo simulations. Our main concern during the design and development of the PSE was to hide the complexity of using the grid infrastructure from the actual users and simplify the way users run their application on grid resources. In order to assist non grid-expert users to execute their simulations on the grid, we have implemented transparent mechanisms that automatically create the appropriate scripts and components required for grid execution by taking as input plain Matlab files. The PSE sets up a temporary distributed Matlab environment on grid resources by using the Matlab compiler and allows the concurrent execution of standard Matlab code. The PSE has been used for several wideband code division multiple access (WCDMA) system simulation scenarios in the production grid provided by the EGEE infrastructure. Simulation results regarding performance of the PSE are presented.