IEEE Sensors Journal · 2024 · 10 citations · 28 references
ComputingComputer EngineeringFeature EngineeringInternet Of ThingsSensorsAnomaly DetectionFault DetectionFault DiagnosisIot ReliabilitySensor NetworksSmart SensorsFault Detection In Sensor NetworksFault Diagnosis In IotKnowledge-based Expert SystemsSensor Data AnalysisSensor Fault DetectionSensor Fault DiagnosisSensor NodesSensor-based Anomaly Detection
With the rapid development of the Internet of Things (IoT), ensuring high-quality data collected by edge devices is of utmost importance for accurate decision-making and cost-effective applications. Collected data may contain errors due to various reasons, such as adverse environmental conditions, vandalism, or issues with sensors themselves (e.g., degradation of sensing material). To ensure the reliability of the IoT applications, it is essential to implement knowledge-based systems that analyze and validate collected data. However, achieving robustness and accuracy in current anomaly detection models is challenging. Faulty sensor data may not always appear as clear deviations from the norm and can mimic normal behavior. Conversely, irregular fluctuations in sensor measurements may represent normal operational variability rather than true anomalies. Moreover, the absence of labeled datasets encompassing normal and real (as opposed to synthetic) abnormal sensor measurements has limited the range of viable data analysis methods. In this article, we propose AssureSense, an advanced fault detection framework that incorporates a novel systematic, data-driven feature extraction method named TsAssure to accurately detect even subtle and hidden faults within sensor data. TsAssure effectively extracts meaningful temporal, local, and spatial features from sensor measurements and its correlated sensors, providing a precise understanding of sensor behavior. This capability enables AssureSense to detect anomalies with a shorter training phase, which is essential in real-world deployments. Testing with real-world datasets demonstrates that AssureSense outperforms established methods in effectively capturing anomalies. Furthermore, experiments reveal the potential for integrating TsAssure with established methods, leading to improv their performance in detecting faults.
28
DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · 2023 · 76.3K citations · Full text
Are Transformers Effective for Time Series Forecasting?
Ailing Zeng, Muxi Chen, Lei Zhang et al. · Proceedings of the AAAI Conference on Artificial Intelligence · 2023 · 2.6K citations · Full text
Robust Anomaly Detection for Multivariate Time Series through Stochastic Recurrent Neural Network
Ya Su, Youjian Zhao, Chenhao Niu et al. · 2019 · 1.5K citations
Time Series FeatuRe Extraction on basis of Scalable Hypothesis tests (tsfresh – A Python package)
Maximilian Christ, N. Braun, Julius Neuffer et al. · Neurocomputing · 2018 · 1.2K citations · Full text