A new approach of anomaly detection in wireless sensor networks using support vector data description

Anomaly detection is an important challenge in wireless sensor networks for some applications, which require efficient, accurate, and timely data analysis to facilitate critical decision making and situation awareness. Support vector data description is well applied to anomaly detection using a very attractive kernel method. However, it has a high computational complexity since the standard version of support vector data description needs to solve quadratic programming problem. In this article, an improved method on the basis of support vector data description is proposed, which reduces the computational complexity and is used for anomaly detection in energy-constraint wireless sensor networks. The main idea is to improve the computational complexity from the training stage and the decision-making stage. First, the strategy of training sample reduction is used to cut back the number of samples and then the sequential minimal optimization algorithm based on the second-order approximation is implemented on the sample set to achieve the goal of reducing the training time. Second, through the analysis of the decision function, the pre-image in the original space corresponding to the center of hyper-sphere in kernel feature space can be obtained. The decision complexity is reduced from O( l) to O(1) using the pre-image. Eventually, the experimental results on several benchmark datasets and real wireless sensor networks datasets demonstrate that the proposed method can not only guarantee detection accuracy but also reduce time complexity.

Download Full-text

Anomaly detection in wireless sensor networks via support vector data description with mahalanobis kernels and discriminative adjustment

2017 4th NAFOSTED Conference on Information and Computer Science ◽

10.1109/nafosted.2017.8108030 ◽

2017 ◽

Author(s):

Van Vuong Trinh ◽

Kim Phuc Tran ◽

Anh Tuan Mai

Keyword(s):

Wireless Sensor Networks ◽

Sensor Networks ◽

Anomaly Detection ◽

Wireless Sensor ◽

Support Vector ◽

Support Vector Data Description ◽

Vector Data ◽

Data Description

Download Full-text

Outlier Detection Using Improved Support Vector Data Description in Wireless Sensor Networks

Sensors ◽

10.3390/s19214712 ◽

2019 ◽

Vol 19 (21) ◽

pp. 4712

Author(s):

Pei Shi ◽

Guanghui Li ◽

Yongming Yuan ◽

Liang Kuang

Keyword(s):

Wireless Sensor Networks ◽

Sensor Networks ◽

Relative Density ◽

Outlier Detection ◽

Sensor Data ◽

Wireless Sensor ◽

Support Vector ◽

Support Vector Data Description ◽

Vector Data ◽

Data Description

Wireless sensor networks (WSNs) are susceptible to faults in sensor data. Outlier detection is crucial for ensuring the quality of data analysis in WSNs. This paper proposes a novel improved support vector data description method (ID-SVDD) to effectively detect outliers of sensor data. ID-SVDD utilizes the density distribution of data to compensate SVDD. The Parzen-window algorithm is applied to calculate the relative density for each data point in a data set. Meanwhile, we use Mahalanobis distance (MD) to improve the Gaussian function in Parzen-window density estimation. Through combining new relative density weight with SVDD, this approach can efficiently map the data points from sparse space to high-density space. In order to assess the outlier detection performance, the ID-SVDD algorithm was implemented on several datasets. The experimental results demonstrated that ID-SVDD achieved high performance, and could be applied in real water quality monitoring.

Download Full-text

Outlier Detection in Wireless Sensor Networks Using Model Selection-Based Support Vector Data Descriptions

Sensors ◽

10.3390/s18124328 ◽

2018 ◽

Vol 18 (12) ◽

pp. 4328 ◽

Cited By ~ 5

Author(s):

Zhan Huan ◽

Chang Wei ◽

Guang-Hui Li

Keyword(s):

Wireless Sensor Networks ◽

Sensor Networks ◽

Model Selection ◽

Outlier Detection ◽

Sensor Data ◽

Wireless Sensor ◽

Quality Data ◽

Support Vector ◽

Support Vector Data Description ◽

Vector Data

Wireless sensor networks (WSNs) are often deployed in harsh and unattended environments, which may cause the generation of abnormal or low quality data. The inaccurate and unreliable sensor data may increase generation of false alarms and erroneous decisions, so it’s very important to detect outliers in sensor data efficiently and accurately to ensure sound scientific decision-making. In this paper, an outlier detection algorithm (TSVDD) using model selection-based support vector data description (SVDD) is proposed. Firstly, the Toeplitz matrix random feature mapping is used to reduce the time and space complexity of outlier detection. Secondly, a novel model selection strategy is realized to keep the algorithm stable under the low feature dimensions, this strategy can select a relatively optimal decision model and avoid both under-fitting and overfitting phenomena. The simulation results on SensorScope and IBRL datasets demonstrate that, TSVDD achieves higher accuracy and lower time complexity for outlier detection in WSNs compared with existing methods.

Download Full-text

Bridge health anomaly detection using deep support vector data description

Neurocomputing ◽

10.1016/j.neucom.2020.08.087 ◽

2021 ◽

Author(s):

JianXi Yang ◽

Fei Yang ◽

Likai Zhang ◽

Ren Li ◽

Shixin Jiang ◽

...

Keyword(s):

Anomaly Detection ◽

Support Vector ◽

Support Vector Data Description ◽

Vector Data ◽

Data Description ◽

Deep Support

Download Full-text

Anomaly Detection for Hyperspectral Imagery Based on Incremental Support Vector Data Description

2010 International Conference on Multimedia Technology ◽

10.1109/icmult.2010.5631355 ◽

2010 ◽

Author(s):

Liyan Zhang ◽

Yonghua Sun ◽

Dan Meng ◽

Xiaojuan Li

Keyword(s):

Anomaly Detection ◽

Hyperspectral Imagery ◽

Support Vector ◽

Support Vector Data Description ◽

Vector Data ◽

Data Description

Download Full-text

Anomaly Detection in Medical Wireless Sensor Networks using SVM and Linear Regression Models

International Journal of E-Health and Medical Communications ◽

10.4018/ijehmc.2014010102 ◽

2014 ◽

Vol 5 (1) ◽

pp. 20-45 ◽

Cited By ~ 32

Author(s):

Osman Salem ◽

Alexey Guerassimov ◽

Ahmed Mehaoua ◽

Anthony Marcus ◽

Borko Furht

Keyword(s):

Wireless Sensor Networks ◽

Sensor Networks ◽

Anomaly Detection ◽

Patient Data ◽

Sensor Nodes ◽

Sensor Data ◽

Wireless Sensor ◽

Smart Devices ◽

Support Vector ◽

Processing Unit

This paper details the architecture and describes the preliminary experimentation with the proposed framework for anomaly detection in medical wireless body area networks for ubiquitous patient and healthcare monitoring. The architecture integrates novel data mining and machine learning algorithms with modern sensor fusion techniques. Knowing wireless sensor networks are prone to failures resulting from their limitations (i.e. limited energy resources and computational power), using this framework, the authors can distinguish between irregular variations in the physiological parameters of the monitored patient and faulty sensor data, to ensure reliable operations and real time global monitoring from smart devices. Sensor nodes are used to measure characteristics of the patient and the sensed data is stored on the local processing unit. Authorized users may access this patient data remotely as long as they maintain connectivity with their application enabled smart device. Anomalous or faulty measurement data resulting from damaged sensor nodes or caused by malicious external parties may lead to misdiagnosis or even death for patients. The authors' application uses a Support Vector Machine to classify abnormal instances in the incoming sensor data. If found, the authors apply a periodically rebuilt, regressive prediction model to the abnormal instance and determine if the patient is entering a critical state or if a sensor is reporting faulty readings. Using real patient data in our experiments, the results validate the robustness of our proposed framework. The authors further discuss the experimental analysis with the proposed approach which shows that it is quickly able to identify sensor anomalies and compared with several other algorithms, it maintains a higher true positive and lower false negative rate.

Download Full-text