A new method for multi-objective optimization problem

Space embedded systems are becoming increasingly-complex and highly-integrated. Researchers found it very hard to coordinate software and hardware design by traditional methods. To bridge the gap between both ends of the design, we resorted to co-design, a method aimed at unifying the view of software and hardware to achieve a multi-objective optimal design under many constraints. We had clearly understood the route map towards co-design, but some key problems hampered our past work. The multi-objective optimization problem involved tasks schedule and resources assignment was one great challenge to us. Therefore, we focus on resolving this kind of problem in this paper. Firstly, major evolutionary methods based on Pareto-optimal are summarized. Afterwards, a branch-and-bound algorithm named BBMOO is proposed. The algorithm can effectively compress the search space by moving and updating a Pareto-rectangle to lower the upper-bound dynamically. Finally, some experiments are designed to verify the algorithm, and the results are shown in visual graphs. In comparison with SPEA2, the good convergence and computational time of BBMOO are presented.