A Review on Immune-Inspired Node Fault Detection in Wireless Sensor Networks with a Focus on the Danger Theory

Abstract

The use of fault detection and tolerance measures in wireless sensor networks is inevitable to ensure the reliability of the data sources. In this context, immune-inspired concepts offer suitable characteristics for developing lightweight fault detection systems, and previous works have shown promising results. In this article, we provide a literature review of immune-inspired fault detection approaches in sensor networks proposed in the last two decades. We discuss the unique properties of the human immune system and how the found approaches exploit them. With the information from the literature review extended with the findings of our previous works, we discuss the limitations of current approaches and consequent future research directions. We have found that immune-inspired techniques are well suited for lightweight fault detection, but there are still open questions concerning the effective and efficient use of those in sensor networks.

A Literature Review on Outlier Detection in Wireless Sensor Networks

“…In terms of mode of operation, online approaches identify outliers in real time or near-real time, although they may have a higher rate of false alarms [1,2,13,14,16]. Offline approaches, on the other hand, collect observations over long periods of time before identifying outliers, taking advantage of historical data and more powerful methods, but they can be unsuitable for WSNs that require online processing [1,13,16]. Hybrid offline/online techniques integrate both approaches, using offline processing for initial model training and then online processing for realtime detection [22].…”

“…In the same context, Jurdak and Wang et al [11] identifies three types of anomalies: (1) network anomalies, (2) node anomalies, and (3) data anomalies. Data anomalies refer to an observation or a subset of observations that, compared to the rest of the dataset, appear to be inconsistent [12]; hence, they are often called "outliers" [13], although the terms anomaly and outlier are commonly used interchangeably in the literature [14]. However, the latter (outlier), in the context of WSN, serves to identify unusual behavior compared to most sensor readings [15]; that is, measurements that significantly differ from the normal pattern of the detected data [16].…”

An energy‐aware software fault detection system based on hierarchical rule approach for enhancing quality of service in internet of things‐enabled wireless sensor network

“…Radio interference, multi-hop communications, and poor environmental conditions are the most typical causes of data failure in IWSNs. 12 These failures expose diverse anomalous values that signify several fault types. In general, IWSNs are vulnerable to three major faults, including hardware, link, and software faults.…”

“…Radio interference, multi‐hop communications, and poor environmental conditions are the most typical causes of data failure in IWSNs 12 . These failures expose diverse anomalous values that signify several fault types.…”

Detecting Anomaly Data for IoT Sensor Networks

“…Nevertheless, the anomaly data detection is a nontrivial task for IoT. e IoT nodes are usually resource-constrained sensors with low-cost embedded systems [21], and traditional anomaly detection solutions cannot be directly applied into IoT [22]. For this reason, it is important that a trade-off solution be found to the problem with decent accuracy while bringing minimum overhead.…”