Multi-Feature Unsupervised Domain Adaptation (M-FUDA) Applied to Cross Unaligned Domain-Specific Distributions in Device-Free Human Activity Classification DOI Creative Commons
Muhammad Hassan, Tom Kelsey

Sensors, Journal Year: 2025, Volume and Issue: 25(6), P. 1876 - 1876

Published: March 18, 2025

Human–computer interaction (HCI) drives innovation by bridging humans and technology, with human activity recognition (HAR) playing a key role. Traditional HAR systems require user cooperation infrastructure, raising privacy concerns. In recent years, Wi-Fi devices have leveraged channel state information (CSI) to decode movements without additional preserving privacy. However, these struggle unseen users, new environments, scalability, thereby limiting real-world applications. Recent research has also demonstrated that the impact of surroundings causes dissimilar variations in at different times day. this paper, we propose an unsupervised multi-source domain adaptation technique addresses challenges. By aligning diverse data distributions target (e.g., or atmospheric conditions), method enhances system adaptability leveraging public datasets varying samples. Experiments on three CSI using preprocessing module convert into image-like formats demonstrate significant improvements baseline methods average micro-F1 score 81% for cross-user, 76% cross-user cross-environment, 73% cross-atmospheric tasks. The approach proves effective scalable, device-free sensing realistic cross-domain scenarios.

Language: Английский

Overall architecture design of power IoT transmission scenario based on cloud-edge-end collaboration DOI Open Access

Hua Li,

Z.D. Wu,

Yunhua Liu

et al.

Journal of Physics Conference Series, Journal Year: 2025, Volume and Issue: 2935(1), P. 012021 - 012021

Published: Jan. 1, 2025

Abstract The general architecture of the conventional power IoT transmission scenario mainly uses MPCAuth (mutual proof consent authentication) mutual authentication protocol for identity authentication, which is vulnerable to impact remote deployment expansion, resulting in poor comprehensive performance indicators. Therefore, a based on cloud-side collaboration proposed. That is, collaborative classification used process data Internet Things, optimize overall response mode Things scenario, and thus complete design scenario. experimental results show that designed has good indicators, reliability, certain application value, made contributions promoting development informatization ensuring quality supply distribution.

Language: Английский

Citations

0
Multi-Feature Unsupervised Domain Adaptation (M-FUDA) Applied to Cross Unaligned Domain-Specific Distributions in Device-Free Human Activity Classification DOI Creative Commons
Muhammad Hassan, Tom Kelsey

Sensors, Journal Year: 2025, Volume and Issue: 25(6), P. 1876 - 1876

Published: March 18, 2025

Human–computer interaction (HCI) drives innovation by bridging humans and technology, with human activity recognition (HAR) playing a key role. Traditional HAR systems require user cooperation infrastructure, raising privacy concerns. In recent years, Wi-Fi devices have leveraged channel state information (CSI) to decode movements without additional preserving privacy. However, these struggle unseen users, new environments, scalability, thereby limiting real-world applications. Recent research has also demonstrated that the impact of surroundings causes dissimilar variations in at different times day. this paper, we propose an unsupervised multi-source domain adaptation technique addresses challenges. By aligning diverse data distributions target (e.g., or atmospheric conditions), method enhances system adaptability leveraging public datasets varying samples. Experiments on three CSI using preprocessing module convert into image-like formats demonstrate significant improvements baseline methods average micro-F1 score 81% for cross-user, 76% cross-user cross-environment, 73% cross-atmospheric tasks. The approach proves effective scalable, device-free sensing realistic cross-domain scenarios.

Language: Английский

Citations

0