A New Approach for Optimal Implementation of Multi-core Reconfigurable Real-time Systems

Abstract

This paper deals with a multi-core reconfigurable real-time system specified with a set of implementations, each of which is raised under a predefined condition and executes multiple functions which are in turns executed by threads. The implementation as threads generates a complex system code. This is due to the huge number of threads and the redundancy between the different implementations which may lead to an increase in the energy consumption. Thus we aim in this paper to optimize the system code by avoiding the redundancy between implementations and reducing the number of threads while meeting all related real-time constraints. The proposed approach adopts mixed integer linear programming (MILP) techniques in the exploration phase in order to provide a feasible task model. An optimal reconfigurable POSIX-based code of the system is manually generated as an output of this technique. An application to a case study and performance evaluation confirm and validate the expected results.