Aug 2026· 2026 6th International Conference on Advanced Algorithms and Neural Networks (AANN)· pp. 495-498· 0 citations· 22 references
Abstract
Industrial Internet of Things (IIoT) systems continuously produce multivariate time-series data for fault warning and predictive maintenance. Many deep anomaly detectors, however, use costly temporal encoders or dense dependency modeling. This paper presents a lightweight attention-enhanced graph neural network (LGAT) for multivariate IIoT anomaly detection. LGAT represents sensor variables as graph nodes, derives a sparse correlation graph from normal data, applies graph attention within sliding windows, and uses forecasting errors as anomaly scores. Experiments on all 28 machines of the public Server Machine Dataset (SMD) compare LGAT with classical, recurrent, autoencoding, Transformer, GCN, and MLP baselines, and report event-level and timing-aware measures in addition to point-wise metrics. LGAT uses 35,329 trainable parameters, 89.3% fewer than the dense MLP, while providing stronger point-adjusted and event-level F1. New visual comparisons and window-size results clarify the accuracy–compactness trade-off and the limitations of fixed temporal windows.
The increasing deployment of IoT sensors in smart homes and offices enables data-driven automation while exposing these systems to sensor faults, abnormal user activities, and unexpected event patterns. Graph Neural Networks (GNNs) offer a principled way to model spatial dependencies among sensors, yet existing GNN-bas...
M. Tanvir, Fahim Ahmed Irfan, Razib Iqbal· Proceedings of the 4th Inter...· 0 citations
Industrial Internet of Things (IIoT) environments generate large volumes of multivariate sensor data through distributed sensing infrastructures used for monitoring, control, and optimization of physical processes. Reliable anomaly detection within these high-dimensional sensor streams is critical for maintaining opera...
Industrial Internet of Things (IIoT) networks support mission-critical applications in manufacturing, energy, and process automation, where even a single link failure can trigger cascading failures with severe safety and operational consequences. Predicting failures before they occur requires models that can capture bo...
Rehab Alawadh, Mostafa Rahmani Ghourtani, H. Ahmadi et al.· IEEE Open Journal of the Com...· 0 citations
This work proposes Residual GRU-Attention Anomaly Detector (RGAAD), an unsupervised framework for IoT time-series anomaly detection that achieves highly competitive performance and consistently outperforms strong baseline methods.
