HiFiNet: Hierarchical Fault Identification in Wireless Sensor Networks via Edge-Based Classification and Graph Aggregation

TL;DR

HiFiNet tackles fault diagnosis in Wireless Sensor Networks by jointly leveraging local temporal dynamics and network-wide spatial context. It introduces an edge-based LSTM-SAE for initial temporal feature extraction followed by an Iterative Graph Network with FiLM-based confidence modulation to refine predictions using neighboring nodes. Across synthetic fault-injected NASA MERRA-2 and Intel Lab datasets, HiFiNet delivers higher accuracy, F1-score, and precision than DBN, LSTM-AE, and SVM, while offering a tunable energy-performance trade-off through configurable network-aggregation frequency. This work advances robust, energy-aware fault detection in WSNs, enabling reliable monitoring under harsh conditions and variable data quality.

Abstract

Wireless Sensor Networks (WSN) are the backbone of essential monitoring applications, but their deployment in unfavourable conditions increases the risk to data integrity and system reliability. Traditional fault detection methods often struggle to effectively balance accuracy and energy consumption, and they may not fully leverage the complex spatio-temporal correlations inherent in WSN data. In this paper, we introduce HiFiNet, a novel hierarchical fault identification framework that addresses these challenges through a two-stage process. Firstly, edge classifiers with a Long Short-Term Memory (LSTM) stacked autoencoder perform temporal feature extraction and output initial fault class prediction for individual sensor nodes. Using these results, a Graph Attention Network (GAT) then aggregates information from neighboring nodes to refine the classification by integrating the topology context. Our method is able to produce more accurate predictions by capturing both local temporal patterns and network-wide spatial dependencies. To validate this approach, we constructed synthetic WSN datasets by introducing specific, predefined faults into the Intel Lab Dataset and NASA's MERRA-2 reanalysis data. Experimental results demonstrate that HiFiNet significantly outperforms existing methods in accuracy, F1-score, and precision, showcasing its robustness and effectiveness in identifying diverse fault types. Furthermore, the framework's design allows for a tunable trade-off between diagnostic performance and energy efficiency, making it adaptable to different operational requirements.

Figures (12)

• Figure 1: WSN Fault Taxonomy.
• Figure 2: Example of a WSN cluster with a base station. Target node is the node which data sequence needed fault classification. In HiFiNet, this sequence will be sent along the shortest path to the cluster head.
• Figure 3: Example of a fault sample and the classification objective.
• Figure 4: Proposed HiFiNet inference pipeline, illustrating the Edge Classifier processing a target node's temperature sequence and the Network Classifier integrating edge outputs with contextual network data.
• Figure 5: Illustration of the Iterative Graph Network architecture, showing the iterative process of feature modulation, graph attention convolution, and confidence update.