SYSTEM-ON-CHIP DESIGN SPACE EXPLORATION AT SYSTEM LEVEL

The sheer complexity of today s embedded systems forces the designer to start with modeling and simulating system components in the very early design stages. This article presents a multi-objective evolutionary algorithm approach to solve the allocation, binding and scheduling problems of VLSI at system level. From a behavioral description using specification graph, the multi- objective genetic algorithm generates the cheapest and fastest heterogeneous hardware or software architecture. The evaluation of the proposed approach is performed on a video decoder. The experimental results show the efficiency, accuracy and scalability of our approach.