arXiv subjects

Wei-Hsiang Wang

Publications and source records attributed to Wei-Hsiang Wang.

11 recordsLinked to original sources

arXiv2026

Scaling WiFi Sensing for Ubiquitous Home Monitoring: Lessons from Real-World Deployment on Millions of Devices

WiFi-based home monitoring offers compelling advantages over traditional camera and sensor solutions by leveraging existing wireless infrastructure for contactless, privacy-preserving and through-the-wall detection. This paper presents insights from developing and deploying a WiFi-based human monitoring system across real-world residential environments, addressing the gap between academic research and practical deployment. Through a two-year study involving 280 edge devices across 15 homes in 11 U.S. states, collecting over 4 million motion samples, we identify and address four critical deployment challenges previously underexplored in academic settings: (1) false positives from non-human motion sources (pets, robots) that degrade system reliability, (2) hardware heterogeneity in commercial IoT devices causing inconsistent CSI quality, (3) signal interference in multi-user environments limiting individual tracking capabilities, and (4) computational and bandwidth constraints preventing real-time edge processing. The deployed system integrates a biomechanics-based classifier that reduces non-human false alarms from 63.1\% to 8.4\%, a multi-layer sensing quality metric validating device suitability without environment-specific calibration, proximity-based multi-user detection leveraging distributed IoT devices, and a hybrid edge-cloud architecture with ACF-based compression achieving 99.72\% data reduction. The integrated system achieves 92.61\% human motion detection accuracy across diverse uncontrolled home environments. We have successfully deployed home monitoring technology on millions of WiFi routers nationwide and smart IoT devices (e.g., bulbs, plugs) worldwide, demonstrating its viability for large-scale real-world applications. We share these findings to guide future research toward deployable WiFi sensing systems.

eess.SP↗

arXiv2026

AM-FM: A Foundation Model for Ambient Intelligence Through WiFi

Ambient intelligence, continuously understanding human presence, activity, and physiology in physical spaces, is fundamental to smart environments, health monitoring, and human-computer interaction. WiFi infrastructure provides a ubiquitous, always-on, privacy-preserving substrate for this capability across billions of IoT devices. Yet this potential remains largely untapped, as wireless sensing has typically relied on task-specific models that require substantial labeled data and limit practical deployment. We present AM-FM, the first foundation model for ambient intelligence and sensing through WiFi. AM-FM is pre-trained on 9.2 million unlabeled Channel State Information (CSI) samples collected over 439 days from 20 commercial device types deployed worldwide, learning general-purpose representations via contrastive learning, masked reconstruction, and physics-informed objectives tailored to wireless signals. Evaluated on public benchmarks spanning nine downstream tasks, AM-FM shows strong cross-task performance with improved data efficiency, demonstrating that foundation models can enable scalable ambient intelligence using existing wireless infrastructure.

cs.LG↗

arXiv2025

CSI-Bench: A Large-Scale In-the-Wild Dataset for Multi-task WiFi Sensing

WiFi sensing has emerged as a compelling contactless modality for human activity monitoring by capturing fine-grained variations in Channel State Information (CSI). Its ability to operate continuously and non-intrusively while preserving user privacy makes it particularly suitable for health monitoring. However, existing WiFi sensing systems struggle to generalize in real-world settings, largely due to datasets collected in controlled environments with homogeneous hardware and fragmented, session-based recordings that fail to reflect continuous daily activity. We present CSI-Bench, a large-scale, in-the-wild benchmark dataset collected using commercial WiFi edge devices across 26 diverse indoor environments with 35 real users. Spanning over 461 hours of effective data, CSI-Bench captures realistic signal variability under natural conditions. It includes task-specific datasets for fall detection, breathing monitoring, localization, and motion source recognition, as well as a co-labeled multitask dataset with joint annotations for user identity, activity, and proximity. To support the development of robust and generalizable models, CSI-Bench provides standardized evaluation splits and baseline results for both single-task and multi-task learning. CSI-Bench offers a foundation for scalable, privacy-preserving WiFi sensing systems in health and broader human-centric applications.

eess.SP↗

arXiv2025

DeepCPD: Deep Learning Based In-Car Child Presence Detection Using WiFi

Child presence detection (CPD) is a vital technology for vehicles to prevent heat-related fatalities or injuries by detecting the presence of a child left unattended. Regulatory agencies around the world are planning to mandate CPD systems in the near future. However, existing solutions have limitations in terms of accuracy, coverage, and additional device requirements. While WiFi-based solutions can overcome the limitations, existing approaches struggle to reliably distinguish between adult and child presence, leading to frequent false alarms, and are often sensitive to environmental variations. In this paper, we present DeepCPD, a novel deep learning framework designed for accurate child presence detection in smart vehicles. DeepCPD utilizes an environment-independent feature-the auto-correlation function (ACF) derived from WiFi channel state information (CSI)-to capture human-related signatures while mitigating environmental distortions. A Transformer-based architecture, followed by a multilayer perceptron (MLP), is employed to differentiate adults from children by modeling motion patterns and subtle body size differences. To address the limited availability of in-vehicle child and adult data, we introduce a two-stage learning strategy that significantly enhances model generalization. Extensive experiments conducted across more than 25 car models and over 500 hours of data collection demonstrate that DeepCPD achieves an overall accuracy of 92.86%, outperforming a CNN baseline by a substantial margin (79.55%). Additionally, the model attains a 91.45% detection rate for children while maintaining a low false alarm rate of 6.14%.

eess.SP↗

arXiv2022

Chirality-Induced Noncollinear Magnetization and Asymmetric Domain-Wall Propagation in Hydrogenated CoPd Thin Films

Array-patterned CoPd-based heterostructures are created through e-beam lithography and plasma pretreatment that induces oxidation with depth gradient in the CoPd alloy films, breaking the central symmetry of the structure. Effects on the magnetic properties of the follow-up hydrogenation of the thin film are observed via magneto-optic Kerr effect microscopy. The system exhibits strong vertical and lateral antiferromagnetic coupling in the perpendicular component between the areas with and without plasma pretreatment, and asymmetric domain-wall propagation in the plasma-pretreated areas during magnetization reversal. These phenomenon exhibit evident magnetic chirality and can be interpreted with the Ruderman-Kittel-Kasuya-Yosida coupling and the Dzyaloshinskii-Moriya interaction (DMI). The sample processing demonstrated in this study allows easy incorporation of lithography techniques that can define areas with or without DMI to create intricate magnetic patterns on the sample, which provides an avenue towards more sophisticate control of canted spin textures in future spintronic devices.

cond-mat.mes-hall↗

arXiv2021

Exchange-bias dependent diffusion rate of hydrogen discovered from evolution of hydrogen-induced noncollinear magnetic anisotropy in FePd thin films

Hydrogenation-induced noncollinear magnetic anisotropy is observed from the evolution of the magnetic domains in FePd alloy thin films using magneto-optic Kerr effect (MOKE) microscopy. MOKE images reveal complicated competitions between different magnetic anisotropies during hydrogen diffusion into the film. An intriguing enhancement of the hydrogen diffusion rate due to the presence of an initial exchange bias induced by a high magnet field is thereby discovered, pointing to an additional scope of controllability of magnetic metal hydrides as potential future hydrogen sensing and storage materials.

cond-mat.mtrl-sci↗

arXiv2021

Dependence of magnetic domain patterns on plasma-induced differential oxidation of CoPd thin films

We demonstrate the evolution of the micro-patterned magnetic domains in CoPd thin films pretreated with e-beam lithography and O2 plasma. During the days-long oxidation, significantly different behaviors of the patterned magnetic domains under magnetization reversal are observed via magneto-optic Kerr effect microscopy on different days. The evolution of the magnetic behaviors indicate critical changes in the local magnetic anisotropy energies due to the Co oxides that evolve into different oxide forms, which are characterized by micro-area X-ray absorption spectroscopy and X-ray photoelectron spectroscopy. The coercive field of the area pre-exposed to plasma can decrease to a value 10 Oe smaller than that unexposed to plasma, whereas after a longer duration of oxidation the coercive field can instead become larger in the area pre-exposed to plasma than that unexposed, leading to an opposite magnetic pattern. Various forms of oxidation can therefore provide an additional dimension for magnetic-domain engineering to the current conventional lithographies.

cond-mat.mtrl-sci↗

arXiv2020

Plasma-induced magnetic patterning of FePd thin films without and with exchange bias

We demonstrate control of magnetic domain structures in continuous FePd thin films by patterning their surfaces with plasma treatment. The Fe-oxide layer formed on the surface upon ambient exposure of the FePd alloy thin film grown on an Al$_2$O$_3$(0001) substrate was patterned into microstructures by e-beam lithography followed by O$_2$- or Ar-plasma treatment. Microscopic pinning of magnetic domain walls in the thin films is then observed by magneto-optic Kerr effect microscopy, with the magnetic field needed to reverse the magnetization of the plasma-treated areas being larger than that for the untreated areas. An intriguing competition between the uniaxial anisotropy and the exchange bias is also observed in the system. This study demonstrates that patterning of the film surface with plasma treatment can be an easy and efficient method for sophisticated engineering of magnetic structures in thin films, and therefore has potential application in developing future data-storage and spintronic devices.

cond-mat.mes-hall↗

arXiv2017

Weak Localization and Weak Antilocalization in Double-Gate a-InGaZnO Thin-Film Transistors

We demonstrate manipulation of quantum interference by controlling the competitions between weak localization (WL) and weak antilocalization (WAL) via variation of the gate voltages of double- gate amorphous InGaZnO thin-film transistors. Our study unveils the full profile of an intriguing universal dependence of the respective WL and WAL contributions on the channel conductivity. This universality is discovered to be robust against interface disorder.

cond-mat.mes-hall↗

arXiv2017

Universal dependence on the channel conductivity of the competing weak localization and antilocalization in amorphous InGaZnO$_4$ thin-film transistors

We investigate the gate-voltage dependence of the magnetoconductivity of several amorphous InGaZnO$_4$ (a-IGZO) thin-film transistors (TFTs). The magnetoconductivity exhibits gate-voltage- controlled competitions between weak localization (WL) and weak antilocalization (WAL), and the respective weights of WL and WAL contributions demonstrate an intriguing universal dependence on the channel conductivity regardless of the difference in the electrical characteristics of the a-IGZO TFTs. Our findings help build a theoretical interpretation of the competing WL and WAL observed in the electron systems in a-IGZO TFTs.

cond-mat.mes-hall↗

arXiv2017

Competing weak localization and weak antilocalization in amorphous indium-gallium-zinc-oxide thin-film transistors

We have investigated the gate-voltage dependence and the temperature dependence of the magnetoconductivity of amorphous indium-gallium-zinc-oxide thin-film transistors. A weak-localization feature is observed at small magnetic fields on top of an overall negative magnetoconductivity at higher fields. An intriguing controllable competition between weak localization and weak antilocalization is observed by tuning the gate voltage or varying the temperature. Our findings reflect controllable quantum interference competition in the electron systems in amorphous indium-gallium-zinc-oxide thin-film transistors.

cond-mat.mes-hall↗