Citation

Rizwan, Muhammad and Ng, Yin Hoe and Wong, Hin Yong and Tan, Chee Keong (2025) A Novel Lightweight ConvNeXt-Incorporated 3D CNN for Robust and Efficient Indoor Localization. In: 2025 Multimedia University Engineering Conference, MECON 2025, 21 July 2025 - 23 July 2025, Cyberjaya.

Text IEEE Xplore Full-Text PDF_10.pdf - Published Version Restricted to Repository staff only Download (1MB)

Abstract

Deep learning-based convolutional architectures have significantly enhanced indoor localization performance. However, their deployment in real-world environments is often hindered by signal variations caused by shadowing, fading, and multipath propagation. While capturing temporal variations alongside spatial features with 3D convolutional neural network (CNN) can improve model robustness, it often comes at the cost of increased computational complexity, restricting its deployment on resource-constrained devices. This study proposes a novel lightweight architecture that incorporates next-generation ConvNeXt blocks into a separable CNN framework to effectively learn spatio-temporal features. Compared to existing 3D CNN model, the proposed model achieves a 91.37% reduction in computational complexity while delivering a slight improvement in localization performance, thereby bridging the gap between accuracy and efficiency for real-world applications.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Deep learning
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	17 Mar 2026 02:45
Last Modified:	17 Mar 2026 02:46
URII:	http://shdl.mmu.edu.my/id/eprint/15466

Downloads

Edit (login required)