基于二型模糊逻辑的多线程数据竞争检测方法研究

发布时间：2018-04-29 04:11

本文选题：二型模糊逻辑 + 隐马尔科夫模型　；参考：《计算机科学》2017年12期

【摘要】：多线程机制以其诸多优势在程序开发中被广泛使用,然而随着多线程软件规模的增长,程序中潜存着许多并发缺陷,最常见的并发缺陷是数据竞争和死锁。目前,针对这些并发缺陷的检测手段都无法处理线程时序的不确定性,无法处理运行时环境对线程时序的影响,同时也不能计算这些并发缺陷发生的概率并根据概率生成其处理优先级。针对以上问题,提出了一种基于二型模糊逻辑的多线程数据竞争检测方法。该方法将传统的多线程时序分析和缺陷检测方法作为预处理,考虑程序运行时环境因素对线程时序的影响,利用二型模糊逻辑和隐马尔科夫模型对待检测程序建模,计算待检测程序在某一系统负载下的时序概率,并根据时序概率生成时序缺陷处理优先级列表供软件开发人员参考。
[Abstract]:Multithreading mechanism is widely used in program development because of its many advantages. However, with the growth of multithread software scale, there are many concurrency defects hidden in the program. The most common concurrent defects are data competition and deadlock. At present, none of the detection methods for these concurrent defects can handle the uncertainty of thread timing and the influence of runtime environment on thread timing. At the same time, the probability of these concurrent defects can not be calculated and the processing priority can be generated according to the probability. In order to solve the above problems, a multithread data competition detection method based on two types of fuzzy logic is proposed. This method takes the traditional method of multithread timing analysis and defect detection as preprocessing, considers the influence of environment factors on thread timing, and uses type 2 fuzzy logic and hidden Markov model to model the detection program. The time sequence probability of the program to be detected under a certain system load is calculated, and the priority list of timing defect processing is generated according to the timing probability for the reference of software developers.
【作者单位】：苏州大学计算机科学与技术学院;
【基金】：国家自然科学基金项目(61202029,61272449,61572339) 江苏省科技支撑计划重点项目(BE2014005-4)资助
【分类号】：TP311.1

【相似文献】