MuECNet: A Lightweight Multiuser Enhanced Convolutional Architecture for Robust CSI-Based Human Activity Recognition in Real-World IoT Environments

Wi-Fi Channel State Information (CSI)-based human activity recognition (HAR) leverages rich channel propagation characteristics to enable non-intrusive, privacy-preserving, and device-free sensing. In multi-user wireless environments, however, HAR faces significant challenges, including multi-path interference, signal overlap, label ambiguity, cross-domain channel variability, and constraints imposed by real-time deployment on resource-limited edge devices. This paper presents MuECNet (Multi-user Enhanced Convolutional Network), a lightweight and modular deep learning framework designed to operate under realistic multi-user, multi-activity CSI sensing conditions. MuECNet integrates three key components: (i) an Enhanced Convolutional Encoding Module (ECEM) for fine-grained temporal–spectral feature extraction that preserves channel propagation signatures; (ii) a Branch-wise Feature Normalization (BFN) module for user-specific channel representation learning; and (iii) an adaptive Decision Module for multi-label activity inference. To improve robustness under diverse and dynamic channel conditions, we introduce MixUp-based data augmentation to emulate activity overlap and reduce label ambiguity. Evaluations on the WiMANS dataset show that MuECNet achieves 64.46% (2.4 GHz) and 64.15% (5 GHz) accuracy with only 1.19M parameters, 1.74G FLOPs, and 0.28 s inference latency, outperforming baseline models such as ABLSTM and THAT while reducing model size by up to 75%. Ablation studies confirm the contribution of each module, with accuracy drops of up to 6.41% when removed. These results demonstrate that MuECNet provides a robust and communication-efficient solution for integrating CSI-based sensing into future wireless networks and IoT systems.