离散时间区间时序逻辑可满足性的判定

目前还没有模型检查的方法自动检测模型是否满足时间区间时序逻辑描述的性质.我们约束时间域到离散时间,证明了离散时间区间时序逻辑的可满足性是可判定的,因而是可模型检查的.提出了时间正则图模型,通过从离散时间区间时序逻辑到时间正则图的构造,提出了基于该逻辑的判定算法,该算法可以推广到其它的时序逻辑模型检查,并优于现有的基于自动机的时序逻辑判定方法.     

以DNA为载体的线性时序逻辑模型检测

线性时序逻辑模型检测被广泛应用于处理器设计与验证、网络协议验证、安全协议验证等领域.然而到目前为止,该技术只能在电子计算的平台上实现.为了以脱氧核糖核酸(Deoxyribo Nucleic Acid,DNA)为载体对线性时序逻辑(Linear Temporal Logic,LTL)实施模型检测,给出了使用粘贴自动机实现Until算子模型检测的方法.首先,使用粘贴自动机对Until公式的有穷状态自动机(Finite State Automata,FSA)模型进行编码;然后,将系统模型转换为粘贴自动机的输入字符串;最后,用粘贴自动机验证系统是否满足公式.仿真实验结果证实,新方法可实现对LTL逻辑时序算子的检测.     

PSL的有界模型检验

 基于SAT的有界模型检验被视为是基于OBDD的符号化模型检验技术的重要补充,是并行反应式系统的一种有效验证方法.然而,直至现在,有界模型检验已验证的属性逻辑还十分有限.PSL是一种用于描述并行系统的属性规约语言(IEEE-1850),包括线性时序逻辑FL和分支时序逻辑OBE两部分.通过模型检验可验证系统的PSL属性,本文提出了PSL的有界模型检验方法及其算法框架.首先,定义PSL逻辑的有界语义,而后,将有界语义进一步简化为SAT,分别将PSL性质规约公式和系统M的状态迁移关系转换为SAT命题公式,最后验证上述两个SAT命题公式合取式的可满足性,这样就将时序逻辑PSL的存在模型检验转化为一个命题公式的可满足性问题,并用一个队列控制电路实例具体解释算法执行过程.     

基于随机森林的网表级时序预测模型

在超大规模集成电路设计中,时序分析的准确性对指导时序优化,保证芯片时序收敛和运行性能至关重要.目前,时序分析绝大多数都是采用商用签核(Sign-off)工具时序报告,作为主要依据.在逻辑综合阶段,由于缺少物理布局布线之后的模块位置和布线结果等信息,因此很难得到准确的电容电阻等寄生参数,用于预测其对应的Sign-off时序.为提高逻辑综合阶段时序预测的准确性,在给定工艺库的情况下,以电路网表作为输入,采用线负载模型对网表的电容电阻等进行估算,并在此基础上利用Elmore Delay模型计算时延作为时序特征.在时序模型训练阶段,提取训练集电路网表的时序特征,以训练模型对应的Sign-off时序结果为标准,采用机器学习中的随机森林算法进行模型训练,包括构建三个模型:互连线时延(Wire delay)、互连线信号转换时延(Wire slew),以及输出负载(Output load).在测试阶段,本文以同工艺库下,新的电路网表作为测试集,输入给训练后的时序模型进行预测.我们的方法与商用工具PrimeTime相比,在Wire delay和Wire slew的Sign-off结果预测上,平均一致性(Correlation)分别提高了49%、37%.此外,我们的方法所预测的Output load与Sign-off结果的一致性在0.99以上.     

基于服务分组和调用轨迹的Web服务组合验证方案

针对Web服务组合的有效性验证问题,提出了一种基于服务分组和调用轨迹的Web服务组合形式化验证方案。首先,基于服务调用顺序,利用提出的Web服务集分组(WSSG)算法将候选Web服务划分为几个子集,并结合调用轨迹编排这些子集组成WSSG图,作为系统的抽象模型;然后,推理出系统所需的预期交互规范,并利用线性时序逻辑(LTL)来描述交互规范;最后,通过检测模型是否符合交互规范来验证组合模型的可行性。实验结果表明,该方案能够有效验证Web服务组合的正确性,且避免了死锁现象。     

RFID信息服务中发布/订阅系统若干关键技术问题研究

 本文将描述逻辑中基于个体的推理方法引入到RFID信息服务的发布/订阅系统中,利用时间本体来描述事件之间的时序关系,通过判断事件断言集与订阅的一致性来解决语义匹配以及与时序相关的复合事件与复合订阅的匹配问题,并给出了订阅语言和匹配算法.实验结果表明,匹配算法的效率能够满足实际应用的需要.     

基于特征值的多模式匹配算法

高速网是当今网络发展的必然趋势,采用现行匹配算法的入侵检测系统(IDS)很难在高速网中有效地运行。本文主要从特征值的多模式匹配算法、模式库的组织和逻辑实现这三个方面来大幅度地提高系统检测速率,完全适应于高速网络的入侵检测。     

一种快速的突变点在线检测算法设计与实现

传统TSTKS算法是一种离线突变点检测算法,该算法在待检测数据存在多个突变点时准确度较低。针对这一问题,文中结合TSTKS算法与滑动窗口理论,提出了一种快速时序数据突变点在线检测方法。该方法利用滑动窗口的思想将待检测数据切分为若干子段,并根据窗口顺序对每个子段采用TSTKS算法进行突变点检测,进而实现时序数据多突变点快速检测。实验结果表明,相比于常见的几种突变点检测算法,采用文中提出算法对存在多突变点的时序数据进行检测时耗时较少,相对误差率较低且命中率较高。     

具有模糊时态的广义可能性线性时序逻辑的模型检测

本文首先定义了具有模糊时态的广义可能性线性时序逻辑GPoFLTL（Generalized Possibilistic Fuzzy Linear Tempora Logic）的语构以及基于路径和基于语言的两种语义解释,证明了GPoFLTL在模糊时态方面对GPoLTL（Generalized Possibilistic Linear Tempora Logic）进行了扩张,并通过实例说明了GPoFLTL比GPoLTL具有更强的表达能力;其次在广义可能性测度下通过模糊矩阵运算讨论了"不久","几乎总是"等几类模糊时态性质的模型检测问题;最后研究了模糊时态性质的必要性阈值模型检测问题,给出了基于自动机的GPoFLTL的阈值模型检测算法及算法的复杂度.     

基于改进SSD算法的铁路隧道漏缆卡具检测

针对目前隧道内漏缆卡具检测数据量大、人工检测效率低的问题,提出了一种基于改进single shot MultiBox detector(SSD)算法的隧道漏缆卡具检测算法.该算法使用不同尺度的特征图检测卡具目标,并在网络宽度和网络深度上对SSD网络结构进行改进.结合Inception结构,增加网路宽度;采用残差结构,在提高网络深度的同时优化网络深度结构;使用深度可分离卷积和1×1卷积结构,减少模型参数量,改善模型结构,从而提高模型检测效率.将改进后的模型应用于隧道漏缆卡具图像检测,实验结果表明,该算法检测的平均准确率达到了86.6％,检测速度达到了26.6 frame/s,相较于原始SSD算法和MobileNet SSD算法,具有明显优势.     

概率时态认知逻辑模型检测中三值抽象技术的研究

 为缓解概率时态认知逻辑模型检测中的状态空间爆炸问题,提出了概率时态认知逻辑的三值抽象技术.具体研究内容包括:定义抽象模型及模型上概率时态认知逻辑的三值语义,依据状态空间等价划分建立初始抽象模型,并证明抽象技术对概率时态认知逻辑的满足性保持关系;提出概率时态认知逻辑模型检测算法;依据初始模型检测的结果,给出利用最小证据和最小反例引导的抽象系统的求精过程.最后通过Dining Cryptographer协议说明了抽象技术的应用,及其在约简系统状态空间方面的效果.     

概率实时时态认知逻辑模型检测中抽象技术的研究

概率实时时态认知逻辑PTACTLK模型检测面临着与传统模型检测同样的挑战,即状态空间爆炸问题.抽象是缓解状态空间爆炸问题的最为有效的方法之一.为了缓解概率实时时态认知逻辑模型检测中的状态空间爆炸问题,我们给出了一种抽象技术:对于PTACTLK中的实时部分PTACTL,采用抽象离散时钟赋值,把概率实时解释系统的无限状态空间转化成有限形式;对于PTACTLK中的认知算子K,给出了抽象状态关于智体认知等价的定义.定义了概率实时解释系统的抽象模型,给出了抽象模型上概率实时时态认知逻辑的语义,并证明了由抽象技术演绎得到的抽象模型是原始模型的上近似.最后通过一个通信协议来说明抽象技术的有效性.     

改进的安全协议自适应分析算法

提出一种改进的安全协议自适应分析算法,即修正学习算法La*,解决部分教师缺乏经验的问题,并将字符集扩展为大字符集。对提出的修正学习算法,进行了正确性证明和复杂度分析。该修正学习算法将有助于提高安全协议自适应模型检测的效率、降低分析和设计成本、缓解状态空间爆炸并增强协议本身对环境和各种攻击手段的防御能力。     

一种基于区间时序逻辑模型检测的入侵检测算法

Model checking based on linear temporal logic reduces the false negative rate of misuse detection. However, linear temporal logic formulae cannot be used to describe concurrent attacks and piecewise attacks. So there is still a high rate of false negatives in detecting these complex attack patterns. To solve this problem, we use interval temporal logic formulae to describe concurrent attacks and piecewise attacks. On this basis, we formalize a novel algorithm for intrusion detection based on model checking interval temporal logic. Compared with the method based on model checking linear temporal logic, the new algorithm can find unknown succinct attacks. The simulation results show that the new method can effectively reduce the false negative rate of concurrent attacks and piecewise attacks.     

On the performance bounds of motion-compensated deinterlacing

The terms ‘proscan up-conversion’ or ‘deinterlacing’ stand for algorithms for the conversion of interlaced image sequences into progressive format by interpolating the lines which are missing due to interlaced scanning. Although deinterlacing is already known to improve TV image quality, such techniques will become even more important in future. Firstly, standard interlaced image sequences will be increasingly processed and displayed on computer platforms and thus usually require progressive scanning and an increased temporal resolution. Secondly, the emerging market of light valve projection equipment demands high-quality proscan up-conversion since the image quality of these display techniques relies heavily on the light efficiency of non-interlaced scanning. By describing deinterlacing as a problem of linear prediction theory, this paper derives upper MSE performance bounds for these algorithms. Thus, a theoretical framework is provided which allows to analyse the effects of various parameters like motion vector inaccuracy, temporal integration and memory constraints on the efficiency of motion-compensated deinterlacing. It is, for example, shown that these algorithms can be very sensitive to motion vector inaccuracies especially if the image sequence is recorded with very short exposure time. Finally, it is briefly demonstrated that the presented analysis can easily be extended to the case of field- and frame-rate up-conversion of interlaced (‘100 Hz’) and non-interlaced image sequences.     

MONSOON: SAT-Based ATPG for Path Delay Faults Using Multiple-Valued Logics

As technology scales down into the nanometer era, delay testing of modern chips has become more and more important. Tests for the path delay fault model are widely used to detect small delay defects and to verify the correct temporal behavior of a circuit. In this article, MONSOON, an efficient SAT-based approach for generating non-robust and robust test patterns for path delay faults is presented. MONSOON handles tri-state elements and environmental constraints occurring in industrial practice using multiple-valued logics. Structural techniques increase the efficiency of the algorithm. A comparison with a state-of-the-art approach shows a significant speed-up. Experimental results for large industrial circuits demonstrates the feasibility and robustness of MONSOON.     

Minimum cost localization problem in wireless sensor networks

Localization is a fundamental problem in wireless sensor networks. Current localization algorithms mainly focus on checking the localizability of a network and/or how to localize as many nodes as possible given a static set of anchor nodes and distance measurements. In this paper, we study a new optimization problem, minimum cost localization problem, which aims to localize all sensors in a network using the minimum number (or total cost) of anchor nodes given the distance measurements. We show this problem is very challenging and then present a set of greedy algorithms using both trilateration and local sweep operations to address the problem. Extensive simulations have been conducted and demonstrate the efficiency of our algorithms.     

Modal space three-state feedback and feedforward control for 2-DOF electro-hydraulic servo shaking table with dynamic coupling caused by eccentric load

Shaking table is a device to simulate the vibration environment, which is used to test the reliability and seismic resistance of many products. However, the control performance of the shaking table is restricted by the serious output coupling caused by eccentric load. In order to solve the coupling problem, a new control strategy, modal space three-state feedback and feedforward control, is proposed. For a 2-DOF electro-hydraulic servo shaking table, its kinematics model and dynamics model with eccentric load are built. And then, the dynamic coupling is analyzed, which makes it impossible for each channel to be controlled independently. To avoid the effect by coupling, a modal space control framework is established based on the vibration theory and the decoupling characteristics of the modal space are researched. In modal space, although a controller of each channel can be designed dependently, the ultimate requirements are decoupling for channels in the degree of freedom space, so it is necessary to discuss the mapping relation between traditional physical space and the presented modal space. The effect of modal channel consistency on the decoupling for physical space is revealed. Furthermore, the correlation of the frequency responses between the two spaces is also analyzed. To meet the obtained conclusions of modal space, a modal space three-state feedback and feedforward controller is designed for improving the dynamic performance and the consistency of modal channels. The experimental results show the correctness of the theoretical analysis, and validate that modal space three-state feedback and feedforward control method is effective, which can significantly reduce the output coupling for the 2-DOF electro-hydraulic servo shaking table with eccentric load.