Temporal and spatial outlier detection in wireless sensor networks

Abstract

Outlier detection techniques play an important role in enhancing the reliability of data communication in wireless sensor networks (WSNs). Considering the importance of outlier detection in WSNs, many outlier detection techniques have been proposed. Unfortunately, most of these techniques still have some potential limitations, that is, (a) high rate of false positives, (b) high time complexity, and (c) failure to detect outliers online. Moreover, these approaches mainly focus on either temporal outliers or spatial outliers. Therefore, this paper aims to introduce novel algorithms that successfully detect both temporal outliers and spatial outliers. Our contributions are twofold: (i) modifying the Hampel Identifier (HI) algorithm to achieve high accuracy identification rate in temporal outlier detection, (ii) combining the Gaussian process (GP) model and graph‐based outlier detection technique to improve the performance of the algorithm in spatial outlier detection. The results demonstrate that our techniques outperform the state‐of‐the‐art methods in terms of accuracy and work well with various data types.

A Literature Review on Outlier Detection in Wireless Sensor Networks

“…Combined Kernelized Outliers Detection Technique (CKODT) [5], a hybrid model, merges KFDA and One Class SVM (OCSVM) for water pipe monitoring in WSNs, while [44] presents an approach based on Optimum-Path Forest (OPF) and meta-heuristics. Techniques such as Temporal Outlier Detection (TOD) and Spacial Outlier Detection (SOD) [45], which employ statistical and graph-based approaches, enhance the detection of temporal and spatial outliers. In Ref.…”

“…Table VI shows a larger number of works using IBRL, followed by GSB; this is primarily because these datasets are public, but more importantly, they were extracted from real-world implementations. Papers such as [4,17,18,30,39,42,45] combine the use of real datasets with synthetic datasets. This latter, in most cases, is generated entirely based on certain statistical distributions, combining the injection of artificial anomalies, manual labeling, and even normalization processes.…”

“…Selecting suitable parameters for the approximate random function and the stochastic gradient descent may be a challenge and can affect the performance of the doOCSVM and Sparse doOCSVM techniques proposed in [39]. The application of TOD and SOD [45] to multivariate data could reduce its effectiveness and efficiency due to difficulties in establishing an appropriate detection threshold. ID-SVDD [23] may not be suitable for real-world applications due to its centralized approach, and there is also a lack of information regarding its efficiency in using computational resources.…”