摘要
随着嵌入式实时系统复杂度提升,传统最坏情况执行时间分析面临挑战,推动研究者采用概率统计技术解决时序分析问题.本文首先了概述概率时序分析方法的发展历程,探讨了验证模型选择对分析结果的影响.然后介绍极值理论在时序分析中的应用,并特别关注了从独立同分布到非独立同分布系统的理论扩展.由于现代计算机系统难以满足独立同分布假设,本文重点描述了MBPTA使能方法中的随机化技术.通过引入随机性打破执行时间间的依赖关系,并可由硬件与软件方法实现.此外,本文讨论了在各种随机化条件下为MBPTA提供可靠统计验证的方法.最后提出了适用于多种随机化系统的通用统计学验证流程.
With the increasing complexity of embedded real-time systems,traditional worst-case execution time analysis faces significant challenges,prompting researchers to adopt probabilistic and statistical techniques to address timing analysis problems.This paper first provides an overview of the development of probabilistic timing analysis methods and discusses the impact of verification model selection on analysis results.It then introduces the application of extreme value theory in timing analysis,with particular focus on theoretical extensions from independent and identically distributed(i.i.d.)to non-i.i.d.systems.Since modern computer systems struggle to meet i.i.d.assumptions,this paper emphasizes randomization techniques in Measurement-Based Probabilistic Timing Analysis(MBPTA)enablement methods.These techniques introduce randomness to break dependencies between execution times and can be implemented through both hardware and software methods.Furthermore,this paper discusses reliable statistical verification methods for MBPTA under various randomization conditions.Finally,it proposes a universal statistical verification process applicable to multiple randomized systems.
作者
王艺霏
张轶
邓庆绪
WANG Yifei;ZHANG Yi;DENG Qingxu(School of Computer Science and Engineering,Northeastern University,Shenyang 110004,China;College of Medicine and Biological Information Engineering,Northeastern University,Shenyang 110004,China)
出处
《小型微型计算机系统》
北大核心
2025年第11期2807-2816,共10页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(62072085)资助
辽宁省科技攻关项目(2024JH2/102400070,2023JH2/101700370)资助。
关键词
基于测量
概率时序分析
统计学验证
实时
WCET
随机
measurement-based
probabilistic timing analysis
statistical testing
real-time
WCET
random
