A Dynamic Multi-Objective Evolutionary Algorithm for Nontrivial Upper Bounds of Real-Time Tasks in Embedded System Design

In the real-time embedded system, how to schedule more real-time tasks has been a difficult point. The nontrivial upper bound of execution time of the real-time task is the key. The analysis of hit-miss behavior of the instruction cache is a problem for the estimation of the nontrivial upper bound, especially more difficult after using shared set-associative instruction caches. And the computation of results for the dynamic problem is a challenge due to the shortcomings of the classic method ILP. In this paper, we present a model for the prediction of nontrivial upper bound, prove that the prediction of the hit-miss number of the shared set-associative instruction cache is a dynamic multi-objective optimization problem, and design a dynamic multi-objective optimization algorithm for the estimation of solution. Simulation experiments demonstrate the effectiveness of our approach for computing the nontrivial upper bound