| 本文已被：浏览 4487次 下载 6082次 |
码上扫一扫！ |
|
|
| 基于树状线性规划搜索的单调速率优化设计 |
|
陈力1,2, 王永吉1,3,4, 吴敬征1,3, 吕荫润1,2
|
|
1.中国科学院软件研究所基础软件国家工程研究中心, 北京 100190;2.中国科学院大学, 北京 100049;3.计算机科学国家重点实验室(中国科学院软件研究所), 北京 100190;4.中国科学院软件研究所互联网软件技术实验室, 北京 100190
|
|
| 摘要: |
| 改善单调速率(rate monotonic,简称RM)可调度性判定算法的效率,是过去40年计算机实时系统设计的重要问题.最近,研究人员把可调度性判定问题扩展到了更一般的优化设计问题,即,如何调节在区间可选择情况下的任务运行时间,使得:(1)系统RM可调度;(2)系统的某个性能(如CPU利用率)达到最优.在已有的求解实时系统RM优化设计问题的方法中,都是先把原问题建模成广义约束优化问题,然后再对广义约束优化问题进行求解.但现有方法的求解速度较慢,任务数较多时不再适用.提出一种求解优化问题的方法——基于树状的线性规划搜索(linearprogramming search,简称LPS)方法.该方法先将实时系统RM优化设计问题建模成广义约束优化问题,再将其分拆成若干线性规划子问题,然后构造线性规划搜索树,利用剪枝搜索算法求解部分线性规划子问题,最后得到优化解.实验结果表明:LPS方法相比于已有的方法能够节省20%~70%的求解时间,任务数越多,节省时间越多.该研究成果可以与计算机可满足性模定理(satisfiability modulo theories,简称SMT)领域的多个研究热点问题联系起来,并可望改善SMT问题的求解效率. |
| 关键词: 实时系统 单调速率 最优化 搜索算法 线性规划 可满足性模定理 |
| DOI：10.13328/j.cnki.jos.004853 |
| 分类号: |
| 基金项目:国家自然科学基金(61170072);国家青年科学基金(61303057);中国科学院、国家外国专家局创新团队国际合作伙伴计划 |
|
| Rate-Monotonic Optimal Design Based on Tree-Like Linear Programming Search |
|
CHEN Li1,2, WANG Yong-Ji1,3,4, WU Jing-Zheng1,3, LÜ Yin-Run1,2
|
|
1.National Engineering Research Center for Fundamental Software, Institute of Software, The Chinese Academy of Sciences, Beijing 100190, China;2.University of Chinese Academy of Sciences, Beijing 100049, China;3.State Key Laboratory of Computer Science(Institute of Software, The Chinese Academy of Sciences), Beijing 100190, China;4.Laboratory for Interact Software Technologies, Institute of Software, The Chinese Academy of Sciences, Beijing 100190, China
|
| Abstract: |
| Over the last four decades, a critical problem in real-time system is to improve the efficiency of the decision algorithm for the rate-monotonic(RM) scheduling. Nowadays researchers extend the decision problem to a generalized optimal design problem, that is, how to adjust the task execution time in the corresponding interval such that(1) the system is schedulable and(2) certain system performance(e.g. CPU utilization) is optimized. All the existing methods for solving this problem are to formulate the problem as the generalized constrained optimization problem(GCOP). However, these methods run very slowly and cannot be applied to the systems with large numbers of tasks. In this paper, a new method for solving the optimization problem is proposed. The method is called tree-like linear programming search(LPS). First, the problem is transformed into a GCOP. Next, the GCOP is partitioned into several linear programming sub-problems. Then, a linear programming search tree is constructed and the node of linear programming is solved by depth-first-search as the optimal solution. The experimental results illustrate that the new method can save 20%~70% of runtime comparing with other existing methods. This work also relates to the research areas of satisfiability modulo theories(SMT), and is expected to improve the efficiency for solving SMT problems. |
| Key words: real-time system rate-monotonic optimization search algorithm linear programming satisfiability modulo theory |
