A multiobjective optimization model for exploring multiprocessor mappings of process networks

TLDR: A optimization software tool, implementing an evolutionary algorithm from the literature, has been developed to achieve a set of best alternative mapping decisions under multiple criteria, some of which were further evaluated by the Sesame simulation framework.

Abstract: In the Sesame framework, we develop a modeling and simulation environment for the efficient design space exploration of heterogeneous embedded systems. Since Sesame recognizes separate application and architecture models within a single system simulation, it needs an explicit mapping step to relate these models for co-simulation. So far in Sesame, the mapping decision has been assumed to be made by an experienced designer, intuitively. However, this assumption is increasingly becoming inappropriate for the following reasons: already the realistic systems are far too complex for making intuitive decisions at an early design stage where the design space is very large. Likely, these systems will even get more complex in the near future. Besides, there exist multiple criteria to consider, like processing times, power consumption and cost of the architecture, which make the decision problem even harder.In this paper, the mapping decision problem is formulated as a multiobjective combinatorial optimization problem. For a solution approach, an optimization software tool, implementing an evolutionary algorithm from the literature, has been developed to achieve a set of best alternative mapping decisions under multiple criteria. In a case study, we have used our optimization tool to obtain a set of mapping decisions, some of which were further evaluated by the Sesame simulation framework.