LDGCN: An Edge-End Lightweight Dual GCN Based on Single-Channel EEG for Driver Drowsiness Monitoring
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the challenge of real-time driver drowsiness monitoring using single-channel electroencephalography (EEG) signals, specifically focusing on deployment on resource-constrained edge devices. Existing methods often lack interpretability in constructing adjacency graphs for Graph Convolutional Networks (GCNs) and suffer from high computational costs, hindering their practical application in edge-end scenarios. The authors propose LDGCN, a lightweight dual GCN that integrates neurophysiological knowledge to improve both accuracy and efficiency. The methodology begins with the construction of a Baseline Drowsiness Status Adjacency Graph (BDSAG). Leveraging the neurophysiological finding that Theta (4–8 Hz) and Alpha (8–12 Hz) frequency bands increase during drowsiness, the authors calculate a Baseline Drowsiness Status Tensor (BDST) from these bands. The BDSAG is built by comparing EEG frequency signals against this BDST, ensuring biological interpretability. To reduce the time complexity of graph construction from $O(n^2)$ to $O(n)$, an adjacency connectivity coefficient $K$ limits the range of node comparisons. To enhance feature expression within limited data, an Augmented Graph-level Module (AGM) captures both global and local views of the graph structure. The core model, LDGCN, utilizes a lightweight dual graph convolution architecture employing depthwise separable convolutions to minimize computational load. Finally, Adaptive Pruning Optimization (APO) is applied to prune channels and neurons, optimizing the model for the fourth-generation Raspberry Pi. Experiments on benchmark datasets demonstrate that LDGCN achieves superior performance compared to state-of-the-art algorithms. The integration of neurophysiological knowledge into the adjacency graph construction significantly improves the interpretability and rationality of the model. The AGM effectively enhances feature representation without destroying graph structural integrity. Crucially, the APO strategy reduced inference latency by approximately half (about 44.44 ms) while simultaneously improving recognition accuracy by 0.8%. The model successfully balances high recognition accuracy with low resource consumption, making it suitable for real-time edge-end deployment. The significance of this work lies in its novel approach to combining domain-specific neurophysiological knowledge with deep learning for EEG analysis. By addressing the lack of interpretability in existing GCN-based methods and optimizing for edge deployment, LDGCN provides a robust, efficient solution for driver drowsiness monitoring. This contributes to the field by demonstrating that lightweight, interpretable models can outperform complex architectures in specific constrained environments, facilitating the practical adoption of AI-driven safety systems in vehicular applications.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-08-09 |
| archive | success | semantic_scholar | — | — | 6 | 2026-08-09 |
| extract | success | cached | — | — | 3 | 2026-08-10 |
| clean | success | clean | — | — | 1 | 2026-08-09 |
| chunk | success | chunk | — | — | 1 | 2026-08-09 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-08-09 |
| enrich | success | semantic_scholar | — | — | 1 | 2026-08-09 |
| promote | success | — | — | — | 1 | 2026-08-09 |
| summarize | success | llm | qwen3.6-27b-nvidia | summ-v5 | 2 | 2026-08-10 |
| tag | success | vector_similarity | — | — | 10 | 2026-08-11 |
| verify | success | — | — | — | 2 | 2026-08-10 |
Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: physiological data