A Comprehensive Method for Anomaly Detection in Complex Dynamic IoT Systems
Modern dynamic systems, such as transportation networks and IoT infrastructures, generate massive volumes of interrelated temporal data represented as temporal graphs. Conventional methods – like clustering, statistical thresholds, and classical time series analysis – often fail to account for the s...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Anhalt University of Applied Sciences
2025-04-01
|
| Series: | Proceedings of the International Conference on Applied Innovations in IT |
| Subjects: | |
| Online Access: | https://icaiit.org/paper.php?paper=13th_ICAIIT_1/2_4 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849682849773387776 |
|---|---|
| author | Andrii Liashenko Larysa Globa |
| author_facet | Andrii Liashenko Larysa Globa |
| author_sort | Andrii Liashenko |
| collection | DOAJ |
| description | Modern dynamic systems, such as transportation networks and IoT infrastructures, generate massive volumes of interrelated temporal data represented as temporal graphs. Conventional methods – like clustering, statistical thresholds, and classical time series analysis – often fail to account for the spatial-temporal dependencies inherent in these systems, leading to high false positive rates or missed complex anomalies. In this paper, we propose a novel anomaly detection approach that combines Temporal Graph Neural Networks (TGNN) with Autoencoders. The method utilizes TGNN to extract robust node representations by capturing both local connectivity and temporal evolution, while an autoencoder is trained to reconstruct normal node behavior. Anomalies are subsequently identified through significant reconstruction errors, which serve as indicators of deviations from typical patterns. Experimental evaluations on the real-world PeMSD7 dataset demonstrate that the proposed TGNN + Autoencoder method improves detection accuracy by 17.33% compared to traditional methods, reduces false positives by 4.71%, and achieves a 6.02% higher F1-score relative to using TGNN or autoencoder individually. These results underline the practical relevance of our approach for real-time monitoring of transportation networks, while also contributing theoretically to the integration of spatial and temporal features in anomaly detection. |
| format | Article |
| id | doaj-art-fd3af3924db04d33b178b0d3e4f8eedf |
| institution | DOAJ |
| issn | 2199-8876 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Anhalt University of Applied Sciences |
| record_format | Article |
| series | Proceedings of the International Conference on Applied Innovations in IT |
| spelling | doaj-art-fd3af3924db04d33b178b0d3e4f8eedf2025-08-20T03:24:03ZengAnhalt University of Applied SciencesProceedings of the International Conference on Applied Innovations in IT2199-88762025-04-0113110110710.25673/119221A Comprehensive Method for Anomaly Detection in Complex Dynamic IoT SystemsAndrii Liashenko0https://orcid.org/0009-0003-1714-414XLarysa Globa1https://orcid.org/0000-0003-3231-3012Institute of Telecommunication Systems, National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute", Beresteiska Avenue 37, 03056 Kyiv, Ukraine Institute of Telecommunication Systems, National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute", Beresteiska Avenue 37, 03056 Kyiv, Ukraine Modern dynamic systems, such as transportation networks and IoT infrastructures, generate massive volumes of interrelated temporal data represented as temporal graphs. Conventional methods – like clustering, statistical thresholds, and classical time series analysis – often fail to account for the spatial-temporal dependencies inherent in these systems, leading to high false positive rates or missed complex anomalies. In this paper, we propose a novel anomaly detection approach that combines Temporal Graph Neural Networks (TGNN) with Autoencoders. The method utilizes TGNN to extract robust node representations by capturing both local connectivity and temporal evolution, while an autoencoder is trained to reconstruct normal node behavior. Anomalies are subsequently identified through significant reconstruction errors, which serve as indicators of deviations from typical patterns. Experimental evaluations on the real-world PeMSD7 dataset demonstrate that the proposed TGNN + Autoencoder method improves detection accuracy by 17.33% compared to traditional methods, reduces false positives by 4.71%, and achieves a 6.02% higher F1-score relative to using TGNN or autoencoder individually. These results underline the practical relevance of our approach for real-time monitoring of transportation networks, while also contributing theoretically to the integration of spatial and temporal features in anomaly detection.https://icaiit.org/paper.php?paper=13th_ICAIIT_1/2_4anomaly detectiontemporal graphstemporal graph neural networksautoencodergraph neural networksreconstruction errordynamic systemstransportation network |
| spellingShingle | Andrii Liashenko Larysa Globa A Comprehensive Method for Anomaly Detection in Complex Dynamic IoT Systems Proceedings of the International Conference on Applied Innovations in IT anomaly detection temporal graphs temporal graph neural networks autoencoder graph neural networks reconstruction error dynamic systems transportation network |
| title | A Comprehensive Method for Anomaly Detection in Complex Dynamic IoT Systems |
| title_full | A Comprehensive Method for Anomaly Detection in Complex Dynamic IoT Systems |
| title_fullStr | A Comprehensive Method for Anomaly Detection in Complex Dynamic IoT Systems |
| title_full_unstemmed | A Comprehensive Method for Anomaly Detection in Complex Dynamic IoT Systems |
| title_short | A Comprehensive Method for Anomaly Detection in Complex Dynamic IoT Systems |
| title_sort | comprehensive method for anomaly detection in complex dynamic iot systems |
| topic | anomaly detection temporal graphs temporal graph neural networks autoencoder graph neural networks reconstruction error dynamic systems transportation network |
| url | https://icaiit.org/paper.php?paper=13th_ICAIIT_1/2_4 |
| work_keys_str_mv | AT andriiliashenko acomprehensivemethodforanomalydetectionincomplexdynamiciotsystems AT larysagloba acomprehensivemethodforanomalydetectionincomplexdynamiciotsystems AT andriiliashenko comprehensivemethodforanomalydetectionincomplexdynamiciotsystems AT larysagloba comprehensivemethodforanomalydetectionincomplexdynamiciotsystems |