Real-Time Systems Optimization with Black-box Constraints and Hybrid Variables

• DOI: 10.48550/arxiv.2401.11620
• Publication Date: 2024-01-01
• Authors: Sen Wang; Dong Li; S.M. Huang; Xuanliang Deng; Ashrarul H. Sifat; Changhee Jung; Ryan K. Williams; Haibo Zeng
• Keywords: Continuous variable; Black box; Coordinate Descent; Discrete optimization

When optimizing real-time systems, designers often face a challenging problem where the schedulability constraints are non-convex, non-continuous, or lack an analytical form to understand their properties. Although optimization framework NORTH proposed in previous work is general (it works with arbitrary analysis) and scalable, it can only handle problems continuous variables, which limits its application. In this paper, we extend applications of hybrid discrete variables. This achieved coordinate-descent method, variables optimized separately during iterations. The new framework, NORTH+, improves around 20% solution quality than experiments.