Authors:
Aymen Gammoudi
1
;
Daniel Chillet
2
;
Mohamed Khalgui
3
and
Adel Benzina
4
Affiliations:
1
IRISA Lab, LISI Lab and Tunisia Polytechnic School, France
;
2
IRISA Lab, France
;
3
LISI Lab and School of Electrical and Information Engineering, Tunisia
;
4
LISI Lab and Tunisia Polytechnic School, Tunisia
Keyword(s):
Heterogeneous Multi-core Platform, Reconfiguration, Task Mapping and Scheduling.
Related
Ontology
Subjects/Areas/Topics:
Application Integration Technologies
;
Applications
;
Service-Oriented Software Engineering and Management
;
Software and Systems Development Methodologies
;
Software Change and Configuration Management
;
Software Engineering
Abstract:
Multi-core Real-time Systems (MRS) powered by a battery have been adopted for a wide range of high performance applications, such as mobile communication and automotive systems. A system is composed of N dependent and periodic Operating System (OS) tasks to be assigned to p heterogeneous cores linked by a network-on-chip (NoC). This paper deals with the problem of task allocation in MRS in such a way that the cost of communication between cores is minimized by trying to place the dependent tasks as close as possible to each other. The main objective is to develop a new strategy for allocating N tasks to p cores of a given distributed system using task clustering by considering both the cost of inter task communication and that of communication between cores. The proposed strategy guarantees that, when a task is mapped into the system and accepted, then it is correctly executed prior to the task deadline. A novel periodic task model based on elastic coefficients is proposed to compute
useful temporal parameters allowing to assign all tasks to p cores,
by minimizing the traffic between cores. Experimental results reveal the effectiveness of the proposed strategy by comparing the derived solutions with the optimal ones, obtained by solving an Integer Linear Program (ILP).
(More)