Sensor technologies empower Industry 4.0 by enabling integration of in-field and real-time raw data into digital twins. However, sensors might be unreliable due to inherent issues and/or environmental conditions. This paper aims at detecting anomalies in measurements from sensors, identifying the faulty ones and accommodating them with appropriate estimated data, thus paving the way to reliable digital twins. More specifically, we propose a general machine-learning-based architecture for sensor validation built upon a series of neural-network estimators and a classifier. Estimators correspond to virtual sensors of all unreliable sensors (to reconstruct normal behaviour and replace the isolated faulty sensor within the system), whereas the classifier is used for detection and isolation tasks. A comprehensive statistical analysis on three different real-world data-sets is conducted and the performance of the proposed architecture validated under hard and soft synthetically-generated faults.

The recent wave of digitalization is characterized by the widespread deployment of sensors in many different environments, e.g., multi-sensor systems represent a critical enabling technology towards full autonomy in industrial scenarios. Sensors usually produce vast amounts of unlabeled data in the form of multivariate time series that may capture normal conditions or anomalies. Multivariate Time Series Anomaly Detection (MTSAD), i.e., the ability to identify normal or irregular operative conditions of a system through the analysis of data from multiple sensors, is crucial in many fields. However, MTSAD is challenging due to the need for simultaneous analysis of temporal (intra-sensor) patterns and spatial (inter-sensor) dependencies. Unfortunately, labeling massive amounts of data is practically impossible in many real-world situations of interest (e.g., the reference ground truth may not be available or the amount of data may exceed labeling capabilities); therefore, robust unsupervised MTSAD is desirable. Recently, advanced techniques in machine learning and signal processing, including deep learning methods, have been developed for unsupervised MTSAD. In this article, we provide an extensive review of the current state of the art with a theoretical background about multivariate time-series anomaly detection. A detailed numerical evaluation of 13 promising algorithms on two publicly available multivariate time-series datasets is presented, with advantages and shortcomings highlighted.

In this article, we address the problem of distributed detection of a noncooperative (unknown emitted signal) target with a wireless sensor network. When the target is present, sensors observe a (unknown) deterministic signal with attenuation depending on the unknown distance between the sensor and the target, multiplicative fading, and additive Gaussian noise. To model energy-constrained operations within Internet of Things, one-bit sensor measurement quantization is employed and two strategies for quantization are investigated. The fusion center receives sensor bits via noisy binary symmetric channels and provides a more accurate global inference. Such a model leads to a test with nuisances (i.e., the target position x _T ) observable only under H ₁ hypothesis. Davies' framework is exploited herein to design the generalized forms of Rao and locally optimum detection (LOD) tests. For our generalized Rao and LOD approaches, a heuristic approach for threshold optimization is also proposed. The simulation results confirm the promising performance of our proposed approaches.

We consider Decentralized Detection (DD) of an unknown signal corrupted by zero-mean unimodal noise via Wireless Sensor Networks (WSNs). We assume the presence of both smart and dumb sensors: the former transmit unquantized measurements, while the latter employ multi-level quantizations (before transmission through binary symmetric channels) in order to cope with energy and/or bandwidth constraints. The data are received by a Fusion Center (FC), which relies on a proposed Rao test, as a simpler alternative to the Generalized Likelihood Ratio Test (GLRT). The asymptotic performance analysis of the multi-bit Rao test is provided and exploited to propose a (signalindependent) quantizer design approach by maximizing the noncentrality parameter of the test-statistic distribution. Since the latter is a non-linear and non-convex function of the quantization thresholds, we employ the particle swarm optimization algorithm for its maximization. Numerical results are provided to show the effectiveness of Rao test in comparison to GLRT and the boost in performance obtained by (multiple) threshold optimization. Asymptotic performance is also exploited to define detection gain measures allowing to assess gain arising from (a) use of dumb sensors and (b) increasing their quantization resolution.

Sensor technologies empower Industry 4.0 by enabling integration of in-field and real-time raw data into digital twins. However, sensors might be unreliable due to inherent issues and/or environmental conditions. This paper aims at detecting anomalies instantaneously in measurements from sensors, identifying the faulty ones and accommodating them with appropriate estimated data, thus paving the way to reliable digital twins. More specifically, a real-time general machine-learning-based architecture for sensor validation is proposed, built upon a series of neural-network estimators and a classifier. Estimators correspond to virtual sensors of all unreliable sensors (to reconstruct normal behaviour and replace the isolated faulty sensor within the system), whereas the classifier is used for detection and isolation tasks. A comprehensive statistical analysis on three different real-world data-sets is conducted and the performance of the proposed architecture is validated under hard and soft synthetically-generated faults.