A clustering algorithm for detecting differential deviations in the multivariate time-series IoT data based on sensor relationship

Internet-of-things (IoT) applications involve a large number of sensors reporting data as a set of time series. Often these data are related to each other based on the relationship of the sensors in the actual application. Any small deviations could indicate a change in the operation of the IoT syst...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Knowledge and information systems Ročník 67; číslo 3; s. 2641 - 2690
Hlavní autoři: Idrees, Rabbia, Maiti, Ananda, Garg, Saurabh
Médium: Journal Article
Jazyk:angličtina
Vydáno: London Springer London 01.03.2025
Springer Nature B.V
Témata:
Algorithms
Anomalies
Clustering
Complexity
Computer Science
Data analysis
Data Mining and Knowledge Discovery
Data transmission
Database Management
Datasets
Deviation
Information Storage and Retrieval
Information Systems and Communication Service
Information Systems Applications (incl.Internet)
Internet of Things
IT in Business
Multivariate analysis
Outliers (statistics)
Regular Paper
Sensors
Synthetic data
Time series
Outlier detection
Multivariate time series
Sensor relationship
Clustering
Deviations
Anomaly detection
ISSN:0219-1377, 0219-3116
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Internet-of-things (IoT) applications involve a large number of sensors reporting data as a set of time series. Often these data are related to each other based on the relationship of the sensors in the actual application. Any small deviations could indicate a change in the operation of the IoT system and potential problems with the IoT application’s goals. It is often difficult to detect such deviations with respect to the relationship between the sensors. This paper presents the clustering algorithm that can efficiently detect all the deviations small or large in the complex and evolving IoT data streams with the help of sensor relationships. We have demonstrated with the help of experiments that our algorithm can efficiently handle high-dimensional data and accurately detect all the deviations. In this paper, we have presented two more algorithms, anomaly detection and outlier detection, that can efficiently categorize the deviations detected by our proposed clustering algorithm into anomalous or normal deviations. We have evaluated the performance and accuracy of our proposed algorithms on synthetic and real-world datasets. Furthermore, to check the effectiveness of our algorithms in terms of efficiency, we have prepared synthetic datasets in which we have increased the complexity of the deviations to show that our algorithm can handle complex IoT data streams efficiently.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0219-1377
0219-3116
DOI:10.1007/s10115-024-02303-3