Reliability evaluation of hypercube multicomputers

An analytic model for the reliability evaluation of hypercube multicomputers is presented. The model is based on the decomposition principle, where a hypercube of a higher dimension is recursively decomposed into smaller hypercubes, until the reliability of the smallest cube is modeled exactly. The reliability of the large n-cube is then obtained from this smallest base model using a recursive equation. The reliability model used is task-based, i.e., it is assumed that the system is operational if a task can be executed on the system. Analytic results are given for n-dimensional hypercubes with up to 75% system degradation. The model is validated by comparison of analytic results with simulation results     

An improved neural network realization for reliability analysis

In this paper we present an improved neural network training algorithm and architecture for reliability analysis of a simplex system and a TMR system which includes the effects of permanent fault and intermittent fault. A fully-connected three-layer neural network represents a discrete-time n-state reliability Markov model of a fault-tolerant system. The desired reliability of the system is fed into the neural network, and when the neural network converges, the design parameters are retrieved from the weights of the neural network. Finally, the simulation results show that the proposed method converges faster than other methods, especially in the case of the state number of the Markov model, which increases. This technique is also suitable for any system.     

A Recursive Algorithm for Computing Exact Reliability Measures

An algorithm is presented to find source-to-K-terminal reliability in a directed graph with independent arc failures. The algorithm is based on a discrete-time Markov chain with two absorbing states. The Markov chain has an upper triangular transition probability matrix, thus the probability of absorption in a state can be found by back-substitution. We show: 1) The source-to-K-terminal reliability is the probability of absorption in a particular absorbing state; 2) The time until absorption can be used as an alternative reliability measure; and 3) The algorithm can be used to find a third reliability measure called the degree of connectedness.     

Markov process based reliability model for laser diodes in space radiation environment

The reliability of laser diodes is a prerequisite for use in satellite optical communication systems, which is significant influenced by both displacement damage effect and annealing effect. A reliability model was proposed to evaluate the reliability and lifetime for laser diodes in space radiation environment. Degradation process is separated into discrete states, and reliability model is subsequently established based on Markov process. The Markov process consists of a Poisson process and an exponential process, representing the displacement damage effect and annealing effect, respectively. Reliability characteristics of a given laser diode are simulated over 100,000 h, and applicability of this reliability model is demonstrated by analyzing the variety trend of probability density of performance states.     

Reliability modeling and assessment of the Star-Graph networks

The reliability of the Star Graph architecture is discussed. The robustness of the Star Graph network under node failures, link failures, and combined node and link failures is shown. The degradation of the Star Graph into Substar Graphs is used as the measure of system effectiveness in the face of failures. Models are provided for each of the failure and re-mapping modes evaluated herein, and the resilience of the Star Graph to failures is emphasized. This paper defines failure of a Star Graph as being when no fault-free (n - 1)-substars remain operational and the intermediate states are defined by the number of (n - 1)-substars that remain operational. A powerful tool (re-mapping) is introduced in which the number of operational (n 1)-substars can be maintained for longer periods, thus improving the overall MTTF (mean time to failure). For comparison the results of a similar reliability analysis of the hypercube is shown. The comparisons are considered conservative due to the failure model used herein for the star graph. One might apply re-mapping to hypercubes; while it would improve the overall MTTF of hypercubes, the hypercubes would still have an appreciably poorer performance than star graphs     

HMT与HMRF联合的SAR图像小波去斑方法

该文将隐马尔可夫树(HMT)和隐马尔可夫随机场(HMRF)两种模型相结合,提出了一种新的估计SAR图像小波系数隐状态的迭代算法.使用该算法可以充分利用小波系数尺度间和尺度内的相关性,更准确地估计隐状态.在此基础上,通过贝叶斯估计分离出小波系数中的信号成分即可消除噪声影响.实验结果表明,该算法能够有效抑制SAR图像相干斑,同时可较好地保持边缘等图像结构特征.     

Analysis of a discrete-time queue with integrated bursty inputs in ATM networks

In this paper, we carry out an exact analysis of a discrete-time queue system with a number of independent Markov modulated inputs in ATM networks, using a generating function approach. We assume that the queueing system has an infinite buffer with M servers. The cell arrival process is characterized by a number of independent Markov modulated geometrical batch arrival processes. We first obtain the generating function of the queue-size distribution at steady-state in vector form, then derive an expression for the average queue-size in terms of the unknown boundary probabilities. To obtain those unknown probabilities, we use the technique proposed in Reference 1. This involves decomposing the system characteristic function to evaluate the roots and solving a set of linear equations. One of the contributions of this paper is presented in Lemma 1, which characterizes the property of the underlying eigenvalues. For one special case of at least M-1 cell arrivals during one slot at one Markov state and of at least M arrivals at all other states, the determination of the unknowns is straightforward. If every Markov modulated arrival process can be further decomposed into a number of i.i.d. two-state, or three-state, or even four-state Markov modulated arrival processes, then each root can be obtained separately using an iterative algorithm. Numerical results are presented to validate the proposed traffic models against actual traffic measurements.     

采用HDPHMM符号化器的语音查询样例检测方法

提出一种基于层级狄利克雷过程隐马尔科夫模型（HDPHMM）符号化器的无监督语音查询样例检测（QbE-STD）方法。该方法首先应用一个双状态层隐马尔科夫模型,其中顶层状态用于表示所发现的声学单元,底层状态用于建模顶层状态的发射概率,通过对顶层状态假设一个层级狄利克雷过程先验,获得非参贝叶斯模型HDPHMM。使用无标注语音数据对该模型进行训练,然后对测试语音和查询样例输出后验概率特征矢量,使用非负矩阵分解算法对后验概率进行优化得到新的特征,然后在此基础上,应用修正分段动态时间规整算法进行检索,构成QbE-STD系统。实验结果表明,相比于基于高斯混合模型符号化器的基线系统,本文所提出的方法性能更优,检索精度得到显著提升。      

Iterative algorithms for state estimation of jump Markov linearsystems

Jump Markov linear systems (JMLSs) are linear systems whose parameters evolve with time according to a finite state Markov chain. Given a set of observations, our aim is to estimate the states of the finite state Markov chain and the continuous (in space) states of the linear system. In this paper, we present original deterministic and stochastic iterative algorithms for optimal state estimation of JMLSs. The first stochastic algorithm yields minimum mean square error (MMSE) estimates of the finite state space Markov chain and of the continuous state of the JMLS. A deterministic and a stochastic algorithm are given to obtain the marginal maximum a posteriori (MMAP) sequence estimate of the finite state Markov chain. Finally, a deterministic and a stochastic algorithm are derived to obtain the MMAP sequence estimate of the continuous state of the JMLS. Computer simulations are carried out to evaluate the performance of the proposed algorithms. The problem of deconvolution of Bernoulli-Gaussian (BG) processes and the problem of tracking a maneuvering target are addressed     

A procedure for probabilistic protocol verification

A new dynamic state exploration algorithm which is applied to probabilistic protocol verification and performance evaluation for synchronous as well as asynchronous models of a protocol is introduced. Properties of the technique are analyzed and proposed to further enhance the verification and evaluation procedure. The procedure based on this technique evaluates the importance of states in the dynamic course of a global reachability graph generation, explores only the most probable subset of states based on their dynamically computed importance, and quantifies and bounds important reliability and performance measures such as mean time to failure and turnaround time. Properties of the algorithm are demonstrated via the call establishment phase of the X.75 protocol     

基于二阶隐马尔可夫模型的文本信息抽取

隐马尔可夫模型是文本信息抽取的重要方法之一.在一阶隐马尔可夫模型中,假设状态转移概率和观察值输出概率仅依赖于模型当前的状态,一定程度降低了信息抽取的精确度.而二阶隐马尔可夫模型合理地考虑了概率和模型历史状态的关联性,对错误信息有更强的识别能力.提出了基于二阶隐马尔可夫模型的文本信息抽取算法;分析了二阶隐马尔可夫模型在文本信息抽取中的有效性;仿真实验表明,新的算法比基于一阶隐马尔可夫模型的算法具有更高的抽取精确度.     

一种改进的用于可靠性分析的神经网络的实现

针对容错系统的可靠性问题,建立基于马尔科夫模型的三层前馈神经网络。提出一种改进的神经网络训练算法,用于包含永久过错．瞬态过错和周期过错影响的一个三模冗余（TMR,Triple Modular Redundancy）系统的可靠性分析。一个全连接的三层神经网络表示一个容错系统的离散时间n状态马尔科夫模型的可靠性。将系统的期望可靠性反馈入网络,当神经网络收敛时,从神经网络的权值中得出设计参数。仿真结果显示,与四层神经网络相比．三层神经网络收敛得更快．收敛可靠十牛更椿沂期望可靠性。     

时变信道中自适应VR-HARQ研究

为了更好地解决时变信道中可靠性与吞吐率这一对矛盾,本文提出一种基于线性预测的自适应冗余可变混合ARQ(VR-HARQ)方案.在该方案中,由于时变信道的慢衰落特性,时变信道被等效为有限状态的Markov过程,每个Markov状态对应信道的不同信噪比,同时,文章详细地描述了信道吞吐率与误码率之间的数学关系,推导了每个Markov状态的最佳编码方案,然后系统采用自适应线性预测算法,根据当前个时段的信道状态,估计下一个时段信道的信噪比以及其所对应的Markov状态,最后根据吞吐率最大原则选择合适的纠错编码方案.仿真结果表明:自适应线性预测VR-HARQ方案的性能明显优于传统VR-HARQ方案.     

Fault-tolerant hypercube multiprocessors

The author presents a new design, called fault-tolerant hypercube (FTH), obtained by augmenting the hypercube topology with some extra links. The FTH has a graceful degradation in performance with the existence of faults. The hardware (link redundancy) is small and negligible for hypercubes with large dimensions. A probabilistic model based on a Markov chain characterizes the FTH-subcube reliability. The mean time to failure is at least 22% better than that for the conventional hypercube. The results have been verified by Monte Carlo simulation. The FTH design is simple and easy to implement. This network can lend itself to the execution of many parallel algorithms designed to run on hypercubes. The FTH contains many more subcubes than the standard hypercube, and thus executes tasks requiring various cube sizes. Where allocation and deallocation of tasks to various subcubes is a common practice, this design achieves an excellent processor usage by efficiently and compactly assigning subcubes     

一种新的边缘检测算法

本文提出了一种基于边缘马尔可夫随机场和波尔兹曼机的边缘检测算法,不仅考虑图象的灰度分布信息,而且利用局部的边缘元素结构。边缘检测过程主要包括两个步骤:(1)利用波尔兹曼机学习算法估计模型参数;(2)波尔兹曼机的迭代搜索算法。实验结果证实了算法的可行性。     

Numerical methods for reliability evaluation of Markov closedfault-tolerant systems

This paper compares three numerical methods for reliability calculation of Markov, closed, fault-tolerant systems which give rise to continuous-time, time-homogeneous, finite-state, acyclic Markov chains. The authors consider a modified version of Jensen's method (a probabilistic method, also known as uniformization or randomization), a new version of ACE (acyclic Markov chain evaluator) algorithm with several enhancements, and a third-order implicit Runge-Kutta method (an ordinary-differential-equation solution method). Modifications to Jensen's method include incorporating stable calculation of Poisson probabilities and steady-state detection of the underlying discrete-time Markov chain. The new version of Jensen's method is not only more efficient but yields more accurate results. Modifications to ACE algorithm are proposed which incorporate scaling and other refinements to make it more stable and accurate. However, the new version no longer yields solution symbolic with respect to time variable. Implicit Runge-Kutta method can exploit the acyclic structure of the Markov chain and therefore becomes more efficient. All three methods are implemented. Several reliability models are numerically solved using these methods and the results are compared on the basis of accuracy and computation cost     

Analysis of a composite performance reliability evaluation for Markovian queueing systems

This paper considers analytical techniques for the evaluation of reliability, availability and combined performance and reliability measures. Markov models are common tools for system reliability evaluation. In this paper, we first compute the state probabilities of Markov model behaviour. We then investigate expected values of interval availability measures. The method is simple and easy to implement, yet very effective in dealing with the problem of computer systems.     

A new formula and an algorithm for reliability analysis of networks

In 1978, Satyanarayana and Prabhakar (SP) proposed a new topological formula for the source-to-terminal (ST) reliability of complex networks. The formula generates only non-cancelling terms of the reliability expression which correspond one-to-one with the p-acyclic subgraphs of the given network. Based on the concept of neutral sequences in acyclic graphs, a powerful SP algorithm was presented for generating all p-acyclic subgraphs and computing the ST reliability of the network. Combining an inclusion-exclusion algorithm with a sum of disjoint products algorithm, we introduce a new topological formula which includes the SP formulas as a special case. A new efficient algorithm which has all features of the SP algorithm is proposed. In general, the reliability expressions obtained by this algorithm are more compact than the ones obtained by the SP algorithm.     

Composite importance measures for multi-state systems with multi-state components

This paper presents & evaluates composite importance measures (CIM) for multi-state systems with multi-state components (MSMC). Importance measures are important tools to evaluate & rank the impact of individual components within a system. For multi-state systems, previously developed measures do not meet all user needs. The major focus of the study is to distinguish between two types of importance measures which can be used for evaluating the criticality of components in MSMC with respect to multi-state system reliability. This paper presents Type 1 importance measures that are involved in measuring how a specific component affects multi-state system reliability. A Monte Carlo (MC) simulation methodology for estimating the reliability of a MSMC is used for computing the proposed CIM metrics. Previous approaches (Type 2) have focused on investigating how a particular component state or set of states affects multi-state system reliability. For some systems, it is not clear how to prioritize system component importance, collectively considering all of its states, using the previously developed importance measures. That detracts from those measures. Experimental results show that the proposed CIM can be used as an effective tool to assess component criticality for MSMC. Examples are used to illustrate & compare the proposed CIM with previous multi-state importance measures.