บทความวารสาร
2026 (ค.ศ.)
บทความวารสาร
A Comparative Analysis of Lightweight CNN Architectures for Wi-Fi CSI-Based Fall Detection
SDG-8: งานที่มีคุณค่าและการเติบโตทางเศรษฐกิจ
SDG-9: อุตสาหกรรม นวัตกรรม และโครงสร้างพื้นฐาน
SDG-12: การบริโภคและการผลิตที่ยั่งยืน
SDG-17: ความร่วมมือเพื่อการพัฒนาที่ยั่งยืน
ผู้แต่ง (Creator)
ผู้ร่วมวิจัย (Co-Authors)
หัวเรื่อง (Subject)
คำสำคัญ (Keywords)
ปีที่เผยแพร่ (Year)
คณะ (Faculty)
ภาษา (Language)
URI (Identifier)
ชื่อวารสาร (Journal Name)
DOI
หน้า (Pages)
สิทธิ์การใช้งาน (Access Level)
สัญญาอนุญาต (License)
Dublin Core Metadata
| Element | Value |
|---|---|
| dc.title | A Comparative Analysis of Lightweight CNN Architectures for Wi-Fi CSI-Based Fall Detection |
| dc.contributor.author | Kawinthida Kingkaew (College of Computing, Prince of Songkla University,Phuket,Thailand) |
| dc.contributor.author | Jirawat Thaenthong |
| dc.contributor.author | Thawatchai Suwanapong |
| dc.date.accessioned | 2026-04-11T15:33:19+07:00 |
| dc.date.issued | 2025 |
| dc.description.abstract | Fall detection for elderly care has gained substantial research attention, with Wi-Fi Channel State Information (CSI) emerging as a promising non-intrusive sensing modality that addresses privacy concerns associated with camera-based systems. However, the adoption of deep learning in this domain requires models that balance high accuracy with computational efficiency, enabling deployment on resource-constrained devices. This study presents a comparative analysis of four lightweight Convolutional Neural Network (CNN) architectures-MobileNetV2, MobileNetV3, EfficientNet-B0, and ShuffleNetV2-that has been presented for Wi-Fi CSI-based fall detection. The models were evaluated using classification metrics, including accuracy, loss, precision, recall, and F1-score, as well as efficiency metrics such as model size, parameter count, FLOPs, and inference time. Experimental results showed that the highest classification accuracy of 98.61% was achieved by EfficientNet-B0, along with the lowest loss and best F1-score (0.986), while a near-fastest inference time of 8.53 ms was maintained. This combination of predictive performance and computational efficiency has been demonstrated to highlight strong potential for practical, real-time fall detection applications. Meanwhile, ShuffleNetV2 achieved a comparable accuracy of 98.47% and an $F 1$-score of 0.985, with the fastest inference time of $\mathbf{7. 1 9}$ ms and a small model size of only 5.21 MB. These results indicate that ShuffleNetV2 provides an excellent trade-off between accuracy and computational cost, making it a highly suitable candidate for deployment on real-time edge devices. |
| dc.identifier.uri | https://sar.trang.psu.ac.th/id/59 |
| dc.identifier.doi | 10.1109/iSAI-NLP66160.2025.11320519 |
| dc.language.iso | eng |
| dc.publisher | Prince of Songkla University, Trang Campus |
| dc.type | บทความวารสาร |
| dc.rights | Public |
| dc.rights.license | CC BY-NC-ND 4.0 - แสดงที่มา-ไม่ใช้เพื่อการค้า-ไม่ดัดแปลง |
| dc.faculty | คณะพาณิชยศาสตร์และการจัดการ |
| dc.bibliographicCitation.pages | 1-6 |
| dc.relation.journal | 2025 20th International Joint Symposium on Artificial Intelligence and Natural Language Processing (iSAI-NLP) |
| dc.subject.sdg | SDG-8: งานที่มีคุณค่าและการเติบโตทางเศรษฐกิจ |
| dc.subject.sdg | SDG-9: อุตสาหกรรม นวัตกรรม และโครงสร้างพื้นฐาน |
| dc.subject.sdg | SDG-12: การบริโภคและการผลิตที่ยั่งยืน |
| dc.subject.sdg | SDG-17: ความร่วมมือเพื่อการพัฒนาที่ยั่งยืน |
บทคัดย่อ (Abstract)
ภาษาอังกฤษ (English)
Fall detection for elderly care has gained substantial research attention, with Wi-Fi Channel State Information (CSI) emerging as a promising non-intrusive sensing modality that addresses privacy concerns associated with camera-based systems. However, the adoption of deep learning in this domain requires models that balance high accuracy with computational efficiency, enabling deployment on resource-constrained devices. This study presents a comparative analysis of four lightweight Convolutional Neural Network (CNN) architectures-MobileNetV2, MobileNetV3, EfficientNet-B0, and ShuffleNetV2-that has been presented for Wi-Fi CSI-based fall detection. The models were evaluated using classification metrics, including accuracy, loss, precision, recall, and F1-score, as well as efficiency metrics such as model size, parameter count, FLOPs, and inference time. Experimental results showed that the highest classification accuracy of 98.61% was achieved by EfficientNet-B0, along with the lowest loss and best F1-score (0.986), while a near-fastest inference time of 8.53 ms was maintained. This combination of predictive performance and computational efficiency has been demonstrated to highlight strong potential for practical, real-time fall detection applications. Meanwhile, ShuffleNetV2 achieved a comparable accuracy of 98.47% and an $F 1$-score of 0.985, with the fastest inference time of $\mathbf{7. 1 9}$ ms and a small model size of only 5.21 MB. These results indicate that ShuffleNetV2 provides an excellent trade-off between accuracy and computational cost, making it a highly suitable candidate for deployment on real-time edge devices.