一种混合模型的时序数据异常检测方法

对设备性能指标、用户数据指标的异常检测能有效地发现系统潜在故障,本文提出了一种混合异常检测方法。该方法利用k-means将历史数据按照时间进行划分,使用grubbs算法剔除历史数据中的噪音,并计算各时间段的阈值形成动态阈值,同时利用曲线拟合和ARIMA算法对预处理后的历史数据进行训练,得到对应的模型,作为判断异常的依据。该方法结合了统计学的高效、机器学习的准确,无需对数据进行标注,该方法能自动发现单指标和多指标异常。通过在几个系统的实际运维的检验,本文提出的方法能有效地发现缺数异常和系统异常,提高告警准确率,单指标的查全率达到100%,平均查准率为95.7%,算法的效率满足生产环境中的性能要求。     

Full‐Textile Wireless Flexible Humidity Sensor for Human Physiological Monitoring

Textile‐based electronic techniques that can in real‐time and noncontact detect the respiration rate and respiratory arrest are highly desired for human health monitoring. Yarn‐shaped humidity sensor is fabricated based on a sensitive fiber with relatively high specific surface area and abnormal cross‐section. The response and recovery time of the yarn‐shaped humidity sensor is only 3.5 and 4 s, respectively, with little hysteresis, because of the hydrophobic property of these functional fibers and the grooves on the surface of the fibers, which is much faster than those of the commercial polyimide materials. Moreover, a battery‐free LC wireless testing system combined with the yarn‐shaped sensor is fabricated, which is further successfully imbedded into the intelligent mask to detect human breath. Based on the detection of LC wireless testing system, the frequency of 50.25 MHz under the exhaled condition shifts to 50.86 MHz under the inhaled situation of humidity sensor. In essence, the functional yarns with proper structure, would be an excellent candidature to the yarn‐shaped humidity sensor, in which there are good performance and wide application possibilities, eventually offering a facile method for the wireless detection of human physiological signals in the field of electronic fabrics.     

A Strong Authentication Scheme with User Privacy for Wireless Sensor Networks

Wireless sensor networks (WSNs) are used for many real‐time applications. User authentication is an important security service for WSNs to ensure only legitimate users can access the sensor data within the network. In 2012, Yoo and others proposed a security‐performance‐balanced user authentication scheme for WSNs, which is an enhancement of existing schemes. In this paper, we show that Yoo and others' scheme has security flaws, and it is not efficient for real WSNs. In addition, this paper proposes a new strong authentication scheme with user privacy for WSNs. The proposed scheme not only achieves end‐party mutual authentication (that is, between the user and the sensor node) but also establishes a dynamic session key. The proposed scheme preserves the security features of Yoo and others' scheme and other existing schemes and provides more practical security services. Additionally, the efficiency of the proposed scheme is more appropriate for real‐world WSNs applications.     

Stretchable and Temperature‐Sensitive Polymer Optical Fibers for Wearable Health Monitoring

Stretchable physical sensors that can detect and quantify human physiological signals such as temperature, are essential to the realization of healthcare devices for biomedical monitoring and human–machine interfaces. Despite recent achievements in stretchable electronic sensors using various conductive materials and structures, the design of stretchable sensors in optics remains a considerable challenge. Here, an optical strategy for the design of stretchable temperature sensors, which can maintain stable performance even under a strain deformation up to 80%, is reported. The optical temperature sensor is fabricated by the incorporation of thermal‐sensitive upconversion nanoparticles (UCNPs) in stretchable polymer‐based optical fibers (SPOFs). The SPOFs are made from stretchable elastomers and constructed in a step‐index core/cladding structure for effective light confinements. The UCNPs, incorporated in the SPOFs, provide thermal‐sensitive upconversion emissions at dual wavelengths for ratiometric temperature sensing by near‐infrared excitation, while the SPOFs endow the sensor with skin‐like mechanical compliance and excellent light‐guiding characteristics for laser delivery and emission collection. The broad applications of the proposed sensor in real‐time monitoring of the temperature and thermal activities of the human body, providing optical alternatives for wearable health monitoring, are demonstrated.     

An incremental intrusion detection system using a new semi‐supervised stream classification method

In recent years, the utilization of machine learning and data mining techniques for intrusion detection has received great attention by both security research communities and intrusion detection system (IDS) developers. In intrusion detection, the most important constraints are the imbalanced class distribution, the scarcity of the labeled data, and the massive amounts of network flows. Moreover, because of the dynamic nature of the network flows, applying static learned models degrades the detection performance significantly over time. In this article, we propose a new semi‐supervised stream classification method for intrusion detection, which is capable of incremental updating using limited labeled data. The proposed method, called the incremental semi‐supervised flow network‐based IDS (ISF‐NIDS), relies on an incremental mixed‐data clustering, a new supervised cluster adjustment method, and an instance‐based learning. The ISF‐NIDS operates in real time and learns new intrusions quickly using limited storage and processing power. The experimental results on the KDD99, Moore, and Sperotto benchmark datasets indicate the superiority of the proposed method compared with the existing state‐of‐the‐art incremental IDSs.     

Contention‐based geographic forwarding in asynchronous duty‐cycled wireless sensor networks

In asynchronous duty‐cycled wireless sensor networks, it is desirable that the data forwarding scheme is adaptive to the dynamics caused by the uncertainty of sensor nodes’ working schedules. Contention‐based forwarding is designed to adapt to the dynamic environments. In this work, we are interested in the contention‐based geographic forwarding (CGF) for two asynchronous duty‐cycling (ADC) models, which we refer to as uninterruptible ADC (U‐ADC) and interruptible ADC (I‐ADC). We propose a new residual time‐aware routing metric for CGF in the I‐ADC model and present a residual time‐aware forwarding scheme using this metric. We evaluate the performance of CGF in both asynchronous duty‐cycling models. Simulation results show that CGF in the U‐ADC model provides a shorter delivery delay while suffering from a high sender effective duty cycle problem. CGF in the I‐ADC model incurs a very long data delivery delay, but it can achieve a good load balancing among nodes. It is also demonstrated that the proposed residual time‐aware forwarding scheme lowers the effects of the performance degradation caused by the pure asynchronous duty‐cycling operation. Copyright © 2011 John Wiley & Sons, Ltd.     

Belief based data cleaning for wireless sensor networks

Wireless sensor network (WSN) data is often subjected to corruption and losses due to wireless medium of communication and presence of hardware inaccuracies in the nodes. For a WSN application to deduce an appropriate result it is necessary that the data received is clean, accurate, and lossless. WSN data cleaning systems exploit contextual associations existing in the received data to suppress data inconsistencies and anomalies. In this work we attempt to clean the data gathered from WSN by capturing the influence of changing dynamics of the environment on the contextual associations existing in the sensor nodes. Specifically, our work validates the extent of similarities among the sensed observations from contextually (spatio‐temporally) associated nodes and considers the time of arrival of data at the sink to educate the cleaning process about the WSN's behavior. We term the data cleaning technique proposed in this work as time of arrival for data cleaning (TOAD). TOAD establishes belief on spatially related nodes to identify potential nodes that can contribute to data cleaning. By using information theory concepts and experiments on data sets from a real‐time scenario we demonstrate and establish that validation of contextual associations among the sensor nodes significantly contributes to data cleaning. Copyright © 2010 John Wiley & Sons, Ltd.     

Vision‐based garbage dumping action detection for real‐world surveillance platform

In this paper, we propose a new framework for detecting the unauthorized dumping of garbage in real‐world surveillance camera. Although several action/behavior recognition methods have been investigated, these studies are hardly applicable to real‐world scenarios because they are mainly focused on well‐refined datasets. Because the dumping actions in the real‐world take a variety of forms, building a new method to disclose the actions instead of exploiting previous approaches is a better strategy. We detected the dumping action by the change in relation between a person and the object being held by them. To find the person‐held object of indefinite form, we used a background subtraction algorithm and human joint estimation. The person‐held object was then tracked and the relation model between the joints and objects was built. Finally, the dumping action was detected through the voting‐based decision module. In the experiments, we show the effectiveness of the proposed method by testing on real‐world videos containing various dumping actions. In addition, the proposed framework is implemented in a real‐time monitoring system through a fast online algorithm.     

An efficient approach for outlier detection in big sensor data of health care

In recent years, wireless sensor networks are pervasive and are generating tons of data every second. Performing outlier detection to detect faulty sensors from such a large amount of data becomes a challenging task. Most of the existing techniques for outlier detection in wireless sensor networks concentrate only on contents of the data source without considering correlation among different data attributes. Moreover, these methods are not scalable to big data. To address these 2 limitations, this paper proposes an outlier detection approach based on correlation and dynamic SMO (sequential minimal optimization) regression that is scalable to big data. Initially, correlation is used to find out strongly correlated attributes and then the point anomalous nodes are detected using dynamic SMO regression. For fast processing of big data, Hadoop MapReduce framework is used. The experimental analysis demonstrates that the proposed approach efficiently detects the point and contextual anomalies and reduces the number of false alarms. For experiments, real data of sensors used in body sensor networks are taken from Physionet database.     

TREE: Routing strategy with guarantee of QoS for industrial wireless sensor networks

As considerable progress has been made in wireless sensor networks (WSNs), we can expect that sensor nodes will be applied in industrial applications. Most available techniques for WSNs can be transplanted to industrial wireless sensor networks (IWSNs). However, there are new requirements of quality of service (QoS), that is, real‐time routing, energy efficiency, and transmission reliability, which are three main performance indices of routing design for IWSNs. As one‐hop neighborhood information is often inadequate to data routing in IWSNs, it is difficult to use the conventional routing methods. In the paper, we propose the routing strategy by taking the real‐time routing performance, transmission reliability, and energy efficiency (TREE, triple R and double E) into considerations. For that, each sensor node should improve the capability of search range in the phase of data route discovery. Because of the increase of available information in the enlarged search range, sensor node can select more suitable relay node per hop. The real‐time data routes with lower energy cost and better transmission reliability will be used in our proposed routing guideline. By comparing with other routing methods through extensive experimental results, our distributed routing proposal can guarantee the diversified QoS requirements in industrial applications. Copyright © 2012 John Wiley & Sons, Ltd.     

基于层级实时记忆算法的时间序列异常检测算法

时间序列异常检测是数据分析中一个重要的研究领域.传统的时间序列的异常检测方法主要通过比较检测数据和历史数据的差异程度,以判断被检测数据是否为奇异点（Surprise）、离群（Outlier）点等.然而序列和窗口的划分,状态的划分或者异常的定义和判定等问题,使得这类方法存在一定的局限性.本文针对传统时间序列检测算法不足,提出一种基于层级实时记忆算法的时间序列异常检测算法.该方法对时间序列内在模式关系进行学习,建立预测模型,通过比较预测值和真实值的偏离程度来判断数据是否异常.首先使用稀疏离散表征在保证保留数据相关性的同时又将数据离散化;然后输入到模型网络,预测下一时刻的数据值;最终根据预测值和真实值的差异为数据异常程度进行定量评分.在人造数据和真实数据上的实验表明,该方法能够准确、快速地发掘时间序列中的异常.     

MEMS压阻式三轴加速度传感器的双路循环电桥检测法

A cycle bridge detection method, which uses a piezoresistive triaxial accelerometer, has been described innovatively. This method just uses eight resistors to form a cycle detection bridge, which can detect the signal of the three directions for real time. It breaks the law of the ordinary independent Wheatstone bridge detection method, which uses at least 12 resistors and each four resistors connected as a Wheatstone bridge to detect the output signal from a specific direction. In order to verify the feasibility of this method, the modeling and simulating of the sensor structure have been conducted by ANSYS, then the dual cycle bridge detection method and independent Wheatstone bridge detection method are compared, the result shows that the former method can improve the sensitivity of the sensor effectively. The sensitivity of the x, y-axis used in the former method is two times that of the sensor used in the latter method, and the sensitivity of the z-axis is four times. At the same time, it can also reduce the cross-axis coupling degree of the sensor used in the dual cycle bridge detection method. In addition, a signal amplifier circuit and adder circuit have been provided, Finally, the test result of the ＂eight-beams/mass＂ triaxial accelerometer, which is based on the dual cycle bridge detection method and the related circuits, have been provided. The results of the test and the theoretical analysis are consistent, on the whole.     

Reference node placement and selection algorithm based on trilateration for indoor sensor networks

The key problem of location service in indoor sensor networks is to quickly and precisely acquire the position information of mobile nodes. Due to resource limitation of the sensor nodes, some of the traditional positioning algorithms, such as two‐phase positioning (TPP) algorithm, are too complicated to be implemented and they cannot provide the real‐time localization of the mobile node. We analyze the localization error, which is produced when one tries to estimate the mobile node using trilateration method in the localization process. We draw the conclusion that the localization error is the least when three reference nodes form an equilateral triangle. Therefore, we improve the TPP algorithm and propose reference node selection algorithm based on trilateration (RNST), which can provide real‐time localization service for the mobile nodes. Our proposed algorithm is verified by the simulation experiment. Based on the analysis of the acquired data and comparison with that of the TPP algorithm, we conclude that our algorithm can meet real‐time localization requirement of the mobile nodes in an indoor environment, and make the localization error less than that of the traditional algorithm; therefore our proposed algorithm can effectively solve the real‐time localization problem of the mobile nodes in indoor sensor networks. Copyright © 2008 John Wiley & Sons, Ltd.     

Intrusion detection in RFID system using computational intelligence approach for underground mines

The radio frequency identification technology (RFID) is commonly used for object tracking and monitoring. In this paper, we discuss a model for intrusion detection system based on RFID to identify the abnormal behavior of underground mines' toxic gases. This model consists of various types of sensor nodes that are integrated with RFID tag, which are deployed in the underground mines by using Zigbee protocol. It consists of coordinators, routers, and sensor nodes, according to different capabilities and the probabilities of intrusive activities that occur in underground mines. It can detect the real‐time abnormal behavior of the toxic gases viz. methane, carbon monoxide, carbon dioxide, hydrogen sulfide, and nitrogen dioxide gases, using artificial neural network middleware techniques. It increases the detection accuracy and reduces the false alarm rate, using multilayer perceptron, radial basis function network, and probabilistic and general regression neural network (PNN/GRNN) techniques. The simulations are performed on the toxic gas dataset, which has been generated in a real‐time scenario by using different gas sensors. The real‐time dataset contains intrusive and nonintrusive values of methane, carbon monoxide, carbon dioxide, hydrogen sulfide, and nitrogen dioxide gases. Experimentally, the PNN/GRNN provides higher detection accuracy as 90.153% for carbon monoxide, 86.713% for carbon dioxide, 93.752% for hydrogen sulfide, and 75.472% for nitrogen dioxide. The PNN/GRNN also provides low false alarm rate as 9.85% for carbon monoxide, 13.29% for carbon dioxide, 6.24% for hydrogen sulfide, and 24.53% for nitrogen dioxide compared with the multilayer perceptron and radial basis function networks.     

Delay aware reliable transport in wireless sensor networks

Wireless sensor networks (WSN) are event‐based systems that rely on the collective effort of several sensor nodes. Reliable event detection at the sink is based on collective information provided by the sensor nodes and not on any individual sensor data. Hence, conventional end‐to‐end reliability definitions and solutions are inapplicable in the WSN regime and would only lead to a waste of scarce sensor resources. Moreover, the reliability objective of WSN must be achieved within a certain real‐time delay bound posed by the application. Therefore, the WSN paradigm necessitates a collective delay‐constrained event‐to‐sink reliability notion rather than the traditional end‐to‐end reliability approaches. To the best of our knowledge, there is no transport protocol solution which addresses both reliability and real‐time delay bound requirements of WSN simultaneously. In this paper, the delay aware reliable transport (DART) protocol is presented for WSN. The objective of the DART protocol is to timely and reliably transport event features from the sensor field to the sink with minimum energy consumption. In this regard, the DART protocol simultaneously addresses congestion control and timely event transport reliability objectives in WSN. In addition to its efficient congestion detection and control algorithms, it incorporates the time critical event first (TCEF) scheduling mechanism to meet the application‐specific delay bounds at the sink node. Importantly, the algorithms of the DART protocol mainly run on resource rich sink node, with minimal functionality required at resource constrained sensor nodes. Furthermore, the DART protocol can accommodate multiple concurrent event occurrences in a wireless sensor field. Performance evaluation via simulation experiments show that the DART protocol achieves high performance in terms of real‐time communication requirements, reliable event detection and energy consumption in WSN. Copyright © 2007 John Wiley & Sons, Ltd.     

HDF: Hybrid Debugging Framework for Distributed Network Environments

Debugging in distributed environments, such as wireless sensor networks (WSNs), which consist of sensor nodes with limited resources, is an iterative and occasionally laborious process for programmers. In sensor networks, it is not easy to find unintended bugs that arise during development and deployment, and that are due to a lack of visibility into the nodes and a dearth of effective debugging tools. Most sensor network debugging tools are not provided with effective facilities such as real‐time tracing, remote debugging, or a GUI environment. In this paper, we present a hybrid debugging framework (HDF) that works on WSNs. This framework supports query‐based monitoring and real‐time tracing on sensor nodes. The monitoring supports commands to manage/control the deployed nodes, and provides new debug commands. To do so, we devised a debugging device called a Docking Debug‐Box (D2‐Box), and two program agents. In addition, we provide a scalable node monitor to enable all deployed nodes for viewing. To transmit and collect their data or information reliably, all nodes are connected using a scalable node monitor applied through the Internet. Therefore, the suggested framework in theory does not increase the network traffic for debugging on WSNs, and the traffic complexity is nearly O(1).     

Real‐Time Liquid Crystal pH Sensor for Monitoring Enzymatic Activities of Penicillinase

A liquid crystal (LC)‐based pH sensor for real‐time monitoring of changes in localized pH values near a solid surface is reported, along with its application for the detection of enzymatic activities. It is found that 4‐cyano‐4′‐pentylbiphenyl (5CB), when doped with 4′‐pentyl‐biphenyl‐4‐carboxylic acid (PBA), shows a bright‐to‐dark optical response to a very small change in pH (from 6.9 to 7.0). The pH‐driven optical response can be explained by using orientational transitions of 5CB induced by the protonation and deprotonation of PBA at the aqueous/LC interface. Because of its high pH sensitivity, the LC‐based sensor is further exploited for monitoring local pH changes resulting from enzymatic reactions. As a proof of concept, the hydrolysis of penicillin G by surface‐immobilized penicillinase is monitored using the system, even when the concentration of penicillin G is as low as 1 nM . This type of LC‐based sensor may find potential utilities in high‐throughput screening of enzyme substrates and enzyme inhibitors.     

Development of Film Sensors Based on Conjugated Polymers for Copper (II) Ion Detection

A fluorescent film sensor was prepared by chemical modification of a polyfluorene derivative on a glass‐plate surface. X‐ray photoelectron spectroscopy and ellipsometry measurements demonstrate the covalent attachment of the polyfluorene derivative to the glass‐plate surface. The sensor was used to detect Cu²⁺ ions in aqueous solution by a mechanism exploiting fluorescence quenching of conjugated polymers. Among the tested metal ions, the film sensor presents good selectivity towards Cu²⁺ ions. Further experiments show that the sensing process is reversible. Moreover, sensory microarrays based on conjugated polymers targeting Cu²⁺ ions are constructed, which display similar sensing performance to that of the film sensor. The structural motif in which conjugated polymers are covalently confined to a solid substrate surface offers several attractive advantages for sensing applications. First, in comparison with film sensors in which small fluorescent molecules are employed as sensing elements, the sensitivity of our new film sensor is enhanced due to the signal‐amplifying effect of the conjugated polymers. Second, the film sensors or microarrays can be used in aqueous environments, which is crucial for their potential use in a wide range of real‐world systems. Since the sensing process is reversible, the sensing materials can be reused. Third, unlike physically coated polymer chains, the covalent attachment of the grafted chains onto a material surface precludes desorption and imparts long‐term stability of the polymer chains.     

一种基于节点实时流量的异常检测算法

针对无线传感器节点缺乏移动性和可预知的流量模式等特征,从而难以在传感器网络中进行异常检测的问题,该文提出了一种基于节点的实时异常检测算法.根据节点流量的到达过程,提出了一种新的节点流量到达模型,根据多层次的滑动窗口事件存储原理,将动态统计值进行短时间的保持,对不同时间段的到达过程指标进行比较,包括节点的可计算资源,低复杂度,融合特性等,以此来判别流量到达过程是否发生异常,从而对传感器网络进行有效的异常检测.     

Real‐time performance evaluation of asynchronous time division traffic‐aware and delay‐tolerant scheme in ad hoc sensor networks

Wireless infrastructureless networks demand high resource availability with respect to the progressively decreasing energy consumption. A variety of new applications with different service requirements demand fairness to the service provision and classification, and reliability in an end‐to‐end manner. High‐priority packets are delivered within a hard time delay bound whereas improper power management in wireless networks can substantially degrade the throughput and increase the overall energy consumed. In this work a new scheme is being proposed and evaluated in real time using a state‐based layered oriented architecture for energy conservation (EC). The proposed scheme uses the node's self‐tuning scheme, where each node is assigned with a dissimilar sleep and wake time, based on traffic that is destined for each node. This approach is based on stream's characteristics with respect to different caching behavioral and storage‐capacity characteristics, and considers a model concerning the layered connectivity characteristics for enabling the EC mechanism. EC characteristics are modeled and through the designed tiered architecture the estimated metrics of the scheme can be bounded and tuned into certain regulated values. The real‐time evaluation results were extracted by using dynamically moving and statically located sensor nodes. A performance comparison is done with respect to different data traffic priority classifications following a real‐time asymmetrical transmission channel. Results have shown the scheme's efficiency in conserving energy while the topology configuration changes with time. Copyright © 2009 John Wiley & Sons, Ltd.