Efficient probe selection algorithms for fault diagnosis

Increase in the network usage for more and more performance critical applications has caused a demand for tools that can monitor network health with minimum management traffic. Adaptive probing has the potential to provide effective tools for end-to-end monitoring and fault diagnosis over a network. Adaptive probing based algorithms adapt the probe set to localize faults in the network by sending less probes in healthy areas and more probes in the suspected areas of failure. In this paper we present adaptive probing tools that meet the requirements to provide an effective and efficient solution for fault diagnosis for modern communication systems. We present a system architecture for adaptive probing based fault diagnosis tool and propose algorithms for probe selection to perform failure detection and fault localization. We compare the performance and efficiency of the proposed algorithms through simulation results.     

基于贝叶斯模型的高效主动探测算法(英文)

Fault diagnosis on large-scale and complex networks is a challenging task, as it requires efficient and accurate inference from huge data volumes. Active probing is a cost-efficient tool for fault diagnosis. However almost all existing probing-based techniques face the following problems: 1) performing inaccurately in noisy networks; 2) generating additional traffic to the network; 3) high cost computation. To address these problems, we propose an efficient probe selection algorithm for fault diagnosis base...     

Efficient probe selection for fault localization using the property of submodularity

As the computer network increasingly grows larger and more complex, fault diagnosis has become a challenging task. Active probing is an efficient tool for fault localization. By implementing some test programs and analyzing the results, active‐probing‐based techniques can perform diagnosis efficiently and adaptively. Because probes may generate additional traffic overhead, it is important to appropriately select small number of probes to reach the desired diagnostic capability. However, the computation of probe selection problem in such environment is extremely expensive. Most of the past works purchase the speed at the cost of diagnostic accuracy. In this paper, we first verify that probe selection problem satisfies the property of submodularity. Then we take the use of the property and develop a submodularity‐based selection algorithm with following novel features: (i) it is cost effective, failure resistant and more accurate; (ii) it could deal with the uncertainties about the network structures and the observations; and (iii) it can select the required probes in near‐linear time. Finally, we implement submodularity‐based selection algorithm and other two representative probe selection algorithms (bounded path enumeration approximation algorithm and greedy search algorithm) on different settings of networks. The results have shown how the new algorithm outperforms the former two algorithms. Copyright © 2011 John Wiley & Sons, Ltd.     

A new distributed route selection approach for channelestablishment in real-time networks

We propose a new distributed route selection approach, called parallel probing, for real-time channel establishment in a point-to-point network. The existing distributed routing algorithms fall into two major categories: preferred neighbor based or flooding based. The preferred neighbor approach offers a better call acceptance rate, whereas the flooding approach is better in terms of call setup time and routing distance. The proposed approach attempts to combine the benefits of both preferred neighbor and flooding approaches in a way to improve all the three performance metrics simultaneously. This is achieved by probing k different paths in parallel, for a channel, by employing different heuristics on each path. Also, the proposed approach uses a notion called intermediate destinations (IDs), which are subset of nodes along the least-cost path between source and destination of a call, in order to reduce the excessive resource reservations while probing for a channel by releasing unused resources between IDs and initiating parallel probes at every ID. Further, it has the flexibility of adapting to different load conditions by its nature of using different heuristics in parallel, and hence, a path found for a channel would have different segments (a segment is a path between two successive IDs), and each of these segments would very well be selected by different heuristics. The effectiveness of the proposed approach has been studied through simulation for well-known network topologies for a wide range of quality-of-service and traffic parameters. The simulation results reveal that the average call acceptance rate offered by the proposed route-selection approach is better than that of both the flooding and preferred neighbor approaches, and the average call setup time and routing distance offered by it are very close to that of the flooding approach     

Robust Monitoring of Link Delays and Faults in IP Networks

In this paper, we develop failure-resilient techniques for monitoring link delays and faults in a Service Provider or Enterprise IP network. Our two-phased approach attempts to minimize both the monitoring infrastructure costs as well as the additional traffic due to probe messages. In the first phase, we compute the locations of a minimal set of monitoring stations such that all network links are covered, even in the presence of several link failures. Subsequently, in the second phase, we compute a minimal set of probe messages that are transmitted by the stations to measure link delays and isolate network faults. We show that both the station selection problem as well as the probe assignment problem are NP-hard. We then propose greedy approximation algorithms that achieve a logarithmic approximation factor for the station selection problem and a constant factor for the probe assignment problem. These approximation ratios are provably very close to the best possible bounds for any algorithm.     

SOSP: an efficient multicast fault isolation scheme for multi-source sessions

This paper proposes a one-way source probing mechanism for fault isolation in multi-source multicast sessions. Routers involved in multicast record a routing path based on periodic probes from sources, and receivers isolate a fault region using the probes. We introduce a probe suppression mechanism to enhance the performance. The proposed scheme reduces message complexity and enhances fault isolation latency, which improves scalability. Furthermore, an analytical formula is proposed to estimate suppression time, which provides maximum performance for a given network status.     

Association Rule Mining-Based Dissolved Gas Analysis for Fault Diagnosis of Power Transformers

This paper presents a novel association rule mining (ARM)-based dissolved gas analysis (DGA) approach to fault diagnosis (FD) of power transformers. In the development of the ARM-based DGA approach, an attribute selection method and a continuous datum attribute discretization method are used for choosing user-interested ARM attributes from a DGA data set, i.e. the items that are employed to extract association rules. The given DGA data set is composed of two parts, i.e. training and test DGA data sets. An ARM algorithm namely Apriori-Total From Partial is proposed for generating an association rule set (ARS) from the training DGA data set. Afterwards, an ARS simplification method and a rule fitness evaluation method are utilized to select useful rules from the ARS and assign a fitness value to each of the useful rules, respectively. Based upon the useful association rules, a transformer FD classifier is developed, in which an optimal rule selection method is employed for selecting the most accurate rule from the classifier for diagnosing a test DGA record. For comparison purposes, five widely used FD methods are also tested with the same training and test data sets in experiments. Results show that the proposed ARM-based DGA approach is capable of generating a number of meaningful association rules, which can also cover the empirical rules defined in industry standards. Moreover, a higher FD accuracy can be achieved with the association rule-based FD classifier, compared with that derived by the other methods.     

Fast algorithms for selection of test nodes of an analog circuit using a generalized fault dictionary approach

Selection of test nodes is an important phase of the fault dictionary approach. It is demonstrated in this paper that the techniques used for this purpose in other approaches of analog fault diagnosis like fault analysis and fault verification are not in general suitable for the fault dictionary approach. The ambiguity set is a simple and effective concept for choosing test nodes in the context of dictionaries. These sets are formed such that each faulty condition lies in only one ambiguity set. Deviating from this thinking, overlapping ambiguity sets are proposed in this paper, giving rise to a generalized fault dictionary. These sets use information more fully and hence reduce the number of test nodes. The concept of hashing is applied in this paper for selecting test nodes. This gives a linear time algorithm (linear in the number of fault voltage specificationsf) and it isf times faster than the existing methods. It is not possible to select test nodes faster than this. This technique can also be used to select test nodes by the process of elimination of nodes. This is also linear inf per node elimination. Even a group of nodes can be eliminated or selected within the same computation. This freedom is not possible with the existing methods.     

基于资源签名的Android应用相似性快速检测方法

由于盗版Android应用（Android Application,简称APP）通常保持着与正版APP相似的用户体验,因此本文提出一种基于资源签名的APP相似性快速检测方法.该方法将APP的资源签名视为字符串集合,利用计算任意一对APP资源签名集合的Jaccard系数判断两者的相似性.为了避免遍历全部的APP对,该方法将MinHash和LSH（Locality Sensitive Hashing）算法的思路引入其中,通过从APP集合中挑选候选对并对候选对进行检验的方式获得最终的检测结果.由于挑选候选对的方式将大量相似性较低的APP对排除在外,因此该方法可以明显地提高APP相似性的检测速度.实验结果表明,该方法的检测速度比现有方法FSquaDRA提高了大约30倍,而检测结果与FSquaDRA几乎完全相同.     

AON中基于网络编码的单链路故障定位机制

文章提出了一种AON(全光网络)中基于网络编码理论的单链路故障定位机制.利用网络编码在信息处理和融合上的优势,可使从单个信源发出的信息经过不同的探测通路后得到不同的处理.若经过每条链路的探测通路的集合互不相同,则可以根据目的节点接收到的探测信息定位故障链路.仿真结果表明,该机制能够在保证定位准确性的同时有效地降低故障定...     

基于分布式选择探测算法的服务路由机制

当前网络中间件服务部署方式僵化单一，难以支持未来多样化的网络服务需求.针对此问题，本文从可重构信息通信基础网络中引入元服务的思想，提出一种支持在控制层进行定制编排以及在数据面按需部署的服务路由机制.该机制将服务路径构建问题建模为多约束最优化问题，并针对性设计了分布式选择探测算法，通过对探针的分布式选择处理进行最优服务路径的构建.仿真结果表明，相较于现有算法，本文提出的服务路由机制在付出较小探测开销的情况下，有效提高了路径服务质量以及服务请求成功率.     

An architecture design approach for large satellite networks

Designing a world-wide satellite network that consists of hundreds of user sites and thousands of circuit connections is a complex problem, which involves selecting a set of candidate satellites and satellite beams/frequency bands from among numerous existing and planned satellites, evaluation of circuit connectivity, earth-station compatibility and sizing, and estimating transponder loading. The design process may also require assessment of the impact of a different set of satellites and modified user traffic requirements on the space segment, the earth-station types and quantity, and the total system cost. Although a conventional design approach based on link-by-link and site-by-site analysis provides accurate results, it is time-consuming and impractical for developing high-level network architectures in a time-constrained environment. A design technique is proposed which employs a set of rules for satellite network design, in combination with extensive databases of satellite parameters, earth-station parameters and user traffic requirements, to synthesize a network architecture. The technique is particularly useful for performing high-level trade-offs among alternative architectures in terms of space segment requirements, the number and type of earth-stations and overall system cost. Once the desired architecture has been selected, a detailed design may be developed using conventional methods.     

Parameter selection for health monitoring of electronic products

This paper presents an approach for selecting precursor parameters for health monitoring of electronic products. The approach includes failure modes, mechanisms, and effects analysis (FMMEA) and life cycle profile analysis of a product. The criticality of the failure mechanisms is established using a risk priority number (RPN), where the RPN for each failure mechanism is calculated as a product of the occurrence and the severity of each mechanism. Performance parameters that can be associated with the critical failure mechanisms should be selected for health monitoring of the product. These parameters could be used for diagnostic purposes. A case study is presented to demonstrate the parameter selection approach for a computer server system. FMMEA was performed on the server, and precursor parameters of the server were selected for monitoring based on the failure modes and mechanisms that posed the highest risk. The utilization of identified parameters for fault detection is presented through a diagnostic algorithm. This approach can be used to select parameters for health monitoring of any system.     

Scalable diagnosis in IP networks using path-based measurement and inference: A learning framework

In this paper, we investigate scalability and performance of measurement-based network monitoring, focusing on failure and congestion diagnosis in IP networks for network-based multimedia applications. Path-based measurements using unicast probe-packets are obtained at end-hosts, and diagnosis is performed by exploiting the spatial dependence among those measurements. We formulate network monitoring in a machine learning framework using probabilistic graphical models which perform inference of the network states (on/off) using unicast measurements. We provide fundamental limits on the relationship between the number of probe packets, the size of a network and the ability to diagnose either failed links or congested network components. Specifically, the diagnosis problem is dealt in a two-fold manner. Initially for fault diagnosis, we construct a graphical model using a Bayesian belief network for path-based measurements. We then provide a lower bound on the average number of probes per edge for link failure diagnosis using variational inference under “noisy” probe measurements. Variational inference provides a feasible approximation to address the number of spatially dependent measurements needed for diagnosis in large networks. We then develop an entropy lower (EL) bound by drawing similarities between coding over a binary symmetric channel (BSC) and link failure diagnosis. Both bounds show that the number of measurements needed for diagnosis grows linearly with respect to the number of links. The analytical results are validated by simulation. On the other hand, for congestion diagnosis, we propose a solution based on decoding of linear error control codes on a BSC. In this scenario, we consider path-based probing experiments under both noiseless and “noisy” measurements and compare its performance against the fundamental limits. To identify the congested nodes we construct a factor graph, and congestion is inferred using belief-propagation algorithm. Simulation results demonstrate the ability of our approach to perfectly localize congested nodes using a scalable number of measurements and a computationally efficient algorithm. We believe that this study can ease the problem arising due to lack of QoS support and provide good-quality broadband multimedia services.     

Survivability Approaches Using p-Cycles in WDM Mesh Networks Under Static Traffic

The major challenge in survivable mesh networks is the design of resource allocation algorithms that allocate network resources efficiently while at the same time are able to recover from a failure quickly. This issue is particularly more challenging in optical networks operating under wavelength continuity constraint, where the same wavelength must be assigned on all links in the selected path. This paper proposes two approaches to solve the survivable routing and wavelength assignment RWA problem under static traffic using p-cycles techniques. The first is a non-jointly approach, where the minimum backup capacity against any single span failure is set up first. Then the working lightpaths problem is solved by first generating the most likely candidate routes for each source and destination s-d pair. These candidate routes are then used to formulate the overall problem as an ILP problem. Alternatively, for a more optimum solution, the problem can be solved jointly, where the working routes and the backup p-cycles are jointly formulated as an ILP problem to minimize the total capacity required. Furthermore, only a subset of high merit cycles that are most likely able to protect the proposed working paths is used in the formulation. Reducing the number of candidate cycles in the final formulation plays a significant role in reducing the number of variables required to solve the problem. To reduce the number of candidate cycles in the formulation, a new metric called Route Sensitive Efficiency (RSE) - has been introduced to pre-select a reduced number of high merit cycle candidates. The RSE ranks each cycle based on the number of links of the primary candidate routes that it can protect. The two approaches were tested and their performances were compared.     

Improving Cooperative Transmission Feasibility by Network Reconfiguration in Limited Backhaul Networks

For coordinated multi-point (CoMP), sets of base stations (BSs) have to be selected to jointly serve user equipments (UEs). These sets are typically selected based on wireless characteristics only. However, using CoMP also poses strict capacity and latency requirements on the backhaul network. Hence, these requirements additionally need to be taken into account when selecting BSs for CoMP. We have developed a mixed integer linear program and a BSs selection heuristic for CoMP that takes into account both aspects: the wireless channels and the backhaul network status. This heuristic can also identify which bottlenecks in the backhaul network make a desired BSs selection infeasible. We exploit this to dynamically reconfigure the backhaul network to the wireless requirements. Our simulations show that the heuristic’s solution quality is close to the optimum while execution time and memory consumption are reduced by multiple orders of magnitude compared to solving the problem via mathematical optimization. In addition, we simulate the network reconfiguration in different backhaul network scenarios. The results illustrate how our approach helps to better exploit available backhaul resources.     

Practical Beacon Placement for Link Monitoring Using Network Tomography

Recent interest in using tomography for network monitoring has motivated the issue of whether it is possible to use only a small number of probing nodes (beacons) for monitoring all edges of a network in the presence of dynamic routing. Past work has shown that minimizing the number of beacons is NP-hard, and has provided approximate solutions that may be fairly suboptimal. In this paper, we use a two-pronged approach to compute an efficient beacon set: 1) we formulate the need for, and design algorithms for, computing the set of edges that can be monitored by a beacon under all possible routing states and 2) we minimize the number of beacons used to monitor all network edges. We show that the latter problem is NP-complete and use various approximate placement algorithm that yields beacon sets of sizes within 1+ln(|E|) of the optimal solution, where E is the set of edges to be monitored. Beacon set computations for several Rocketfuel Internet service provider topologies indicate that our algorithms may reduce the number of beacons yielded by past solutions by more than 50% and are, in most cases, close to optimal     

基于半监督学习的SAR目标检测网络

现有的基于卷积神经网络(CNN)的合成孔径雷达(SAR)图像目标检测算法依赖于大量切片级标记的样本,然而对SAR图像进行切片级标记需要耗费大量的人力和物力。相对于切片级标记,仅标记图像中是否含有目标的图像级标记较为容易。该文利用少量切片级标记的样本和大量图像级标记的样本,提出一种基于卷积神经网络的半监督SAR图像目标检测方法。该方法的目标检测网络由候选区域提取网络和检测网络组成。半监督训练过程中,首先使用切片级标记的样本训练目标检测网络,训练收敛后输出的候选切片构成候选区域集;然后将图像级标记的杂波样本输入网络,将输出的负切片加入候选区域集;接着将图像级标记的目标样本也输入网络,对输出结果中的正负切片进行挑选并加入候选区域集;最后使用更新后的候选区域集训练检测网络。更新候选区域集和训练检测网络交替迭代直至收敛。基于实测数据的实验结果证明,所提方法的性能与使用全部样本进行切片级标记的全监督方法的性能相差不大。     

Model-based object tracking in wireless sensor networks

Tracking moving objects is one of the most common requirements in wireless sensor network applications. Most tracking schemes predict a target’s location based on a single object movement model and periodically activate nearby sensors to monitor the target. However, in most real-world situations, a target exhibits multiple movement patterns. Thus, multiple movement models are required to accurately describe the target’s movement. This paper proposes a tracking framework, called model-based object tracking system (MOTS), that allows a sensor network to adaptively apply the most suitable tracking mechanism to monitor the target under various circumstances. To fairly and accurately evaluate all tracking modules, this study further develops a monitoring-cost evaluator to evaluate the monitoring cost of the inactive tracking modules, and then designs three tracking module selection strategies, the Greedy Strategy, Min-Max Strategy, and Weighted Moving Average Strategy, to select the most effective tracking module to monitor the target in each period. A set of experiments is conducted to evaluate MOTS and compare it against existing tracking systems. The obtained results reveal that the cost efficiency of MOTS is considerably better than that of existing tracking systems.     

Improved diagnosis using enhanced fault dominance

Defect diagnosis can benefit from fault dominance relations to reduce the set of defect candidate sites. This paper presents new fault dominance collapsing operators that further reduce the set of candidates considered during the initial phase of diagnosis. In contrast to existing dominance-based methods which operate on pairs of faults, the proposed method operates on sets of faults. Fault-related entities are generated to guide the diagnosis process. The proposed collapsing operators can be used to accelerate effect-cause diagnosis. Experimental results demonstrate that the proposed method achieves a higher collapsing ratio than existing methods.