| This dissertation presents a Multi-objective System level synthesis Utilization model framework that achieves near optimal application, algorithm, and architecture mappings of design implementation scenarios. The framework approach uses integrated genetic algorithm (GA) and mixed integer linear programming (MILP) methods. The use of both the non-deterministic GA and deterministic MILP approaches allows for a quick realization of the best mapping and the accuracy of the mapping respectively.; The proposed framework performs top-down design space explorations over different levels of abstractions as a way of reducing design cost, reducing time-to-market, and also evaluating design performance tradeoffs. A complete model framework was developed on Matlab platform and near optimal design mappings were demonstrated at various design abstraction levels including the mapping exploration onto a field programmable analog array. |
