Data-driven anomaly detection based on multi-sensor data fusion

Di Wang, Ahmad Al-Rubaie, Sandra Stincic, John Davies, Alia Aljasmi

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In the age of IoT, a huge amount of real time data is produced every second from the colossal number and different types of sensors deployed. A generic and intelligent method to monitor these large data streams from a wide range of sources without human supervision or the use of expert knowledge is a big challenge. In this paper we propose, develop, and test a generic method for anomaly detection which is completely data-driven without human supervision. The proposed method is able to detect the underlying correlations amongst multiple sensors and detect the data patterns from all correlated sensor data through time. Anomalies are detected from marginal deviations from the normal identified patterns. The proposed method is applied to Building Management System's data which include various types of sensors and proves the generality of the proposed method.

Original languageBritish English
Title of host publication2021 International Conference on Smart Applications, Communications and Networking, SmartNets 2021
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665435451
DOIs
StatePublished - 22 Sep 2021
Event2021 International Conference on Smart Applications, Communications and Networking, SmartNets 2021 - Glasgow, United Kingdom
Duration: 22 Sep 202124 Sep 2021

Publication series

Name2021 International Conference on Smart Applications, Communications and Networking, SmartNets 2021

Conference

Conference2021 International Conference on Smart Applications, Communications and Networking, SmartNets 2021
Country/TerritoryUnited Kingdom
CityGlasgow
Period22/09/2124/09/21

Keywords

  • Anomaly detection
  • Data driven
  • Data fusion
  • Multiple sources

Fingerprint

Dive into the research topics of 'Data-driven anomaly detection based on multi-sensor data fusion'. Together they form a unique fingerprint.

Cite this