arXiv subjects

Dian Fan

Publications and source records attributed to Dian Fan.

8 recordsLinked to original sources

arXiv2026

Over-the-Air Computation via Segmented Waveguide-Enabled Pinching-Antenna Systems

A segmented waveguide-enabled pinching-antenna system (SWAN)-assisted over-the-air computation (AirComp) framework is proposed. Three transmission architectures, namely segment selection (SS), phase-shifter-free segment aggregation (SA), and phase-shifter-enabled SA, are developed for uplink signal aggregation. For each architecture, low-complexity algorithms are developed to optimize the pinching-antenna placement and the per-segment phase shifts. Numerical results demonstrate the effectiveness of the proposed approaches and the superiority of SWAN over the conventional pinching-antenna system (PASS). It is shown that both SS and SA achieve lower computation mean-squared error than the conventional PASS, while segment-wise phase control further improves the performance of SA.

eess.SP↗

arXiv2026

The Breakthrough of Sleep: A Contactless Approach for Accurate Sleep Stage Detection Using the Sleepal AI Lamp

Sleep staging is essential for the assessment of sleep quality and the diagnosis of sleep-related disorders. Conventional polysomnography (PSG), while considered the gold standard, is intrusive, labor-intensive, and unsuitable for long-term monitoring. This study evaluates the performance of the Sleepal AI Lamp, a contactless, radar-based consumer-grade sleep tracker, in comparison with gold-standard polysomnography (PSG), using a large-scale dataset comprising 1022 overnight recordings. We extract multi-scale respiratory and motion-related features from radar signals to train a frequency-augmented deep learning model. For the binary sleep-wake classification task, experimental results demonstrated that the model achieved an accuracy of 92.8% alongside a macro-averaged F1 score of 0.895. For four-stage classification (wake, light NREM (N1 + N2), deep NREM (N3), REM), the model achieved an accuracy of 78.5% with a Cohen's kappa coefficient of 0.695 in healthy individuals and maintained a stable accuracy of 77.2% with a kappa of 0.677 in a heterogeneous population including patients with varying severities of obstructive sleep apnea (OSA). These experimental results demonstrate that the sleep staging performance of the contactless Sleepal AI Lamp is in high agreement with expert-labeled PSG sleep stages. Our findings suggest that non-contact radar sensing, combined with advanced temporal modeling, can provide reliable sleep staging performance without requiring physical contact or wearable devices. Owing to its unobtrusive nature, ease of deployment, and robustness to long-term use, the contactless Sleepal AI Lamp shows strong potential for clinical screening, home-based sleep assessment, and continuous longitudinal sleep monitoring in real-world medical and healthcare applications.

eess.SP↗

arXiv2022

Over-the-Air Federated Multi-Task Learning via Model Sparsification and Turbo Compressed Sensing

To achieve communication-efficient federated multitask learning (FMTL), we propose an over-the-air FMTL (OAFMTL) framework, where multiple learning tasks deployed on edge devices share a non-orthogonal fading channel under the coordination of an edge server (ES). In OA-FMTL, the local updates of edge devices are sparsified, compressed, and then sent over the uplink channel in a superimposed fashion. The ES employs over-the-air computation in the presence of intertask interference. More specifically, the model aggregations of all the tasks are reconstructed from the channel observations concurrently, based on a modified version of the turbo compressed sensing (Turbo-CS) algorithm (named as M-Turbo-CS). We analyze the performance of the proposed OA-FMTL framework together with the M-Turbo-CS algorithm. Furthermore, based on the analysis, we formulate a communication-learning optimization problem to improve the system performance by adjusting the power allocation among the tasks at the edge devices. Numerical simulations show that our proposed OAFMTL effectively suppresses the inter-task interference, and achieves a learning performance comparable to its counterpart with orthogonal multi-task transmission. It is also shown that the proposed inter-task power allocation optimization algorithm substantially reduces the overall communication overhead by appropriately adjusting the power allocation among the tasks.

cs.IT↗

arXiv2021

Over-the-Air Federated Multi-Task Learning

In this letter, we introduce over-the-air computation into the communication design of federated multi-task learning (FMTL), and propose an over-the-air federated multi-task learning (OA-FMTL) framework, where multiple learning tasks deployed on edge devices share a non-orthogonal fading channel under the coordination of an edge server (ES). Specifically, the model updates for all the tasks are transmitted and superimposed concurrently over a non-orthogonal uplink fading channel, and the model aggregations of all the tasks are reconstructed at the ES through a modified version of the turbo compressed sensing algorithm (Turbo-CS) that overcomes inter-task interference. Both convergence analysis and numerical results show that the OA-FMTL framework can significantly improve the system efficiency in terms of reducing the number of channel uses without causing substantial learning performance degradation.

cs.LG↗

arXiv2021

Temporal-Structure-Assisted Gradient Aggregation for Over-the-Air Federated Edge Learning

In this paper, we investigate over-the-air model aggregation in a federated edge learning (FEEL) system. We introduce a Markovian probability model to characterize the intrinsic temporal structure of the model aggregation series. With this temporal probability model, we formulate the model aggregation problem as to infer the desired aggregated update given all the past observations from a Bayesian perspective. We develop a message passing based algorithm, termed temporal-structure-assisted gradient aggregation (TSA-GA), to fulfil this estimation task with low complexity and near-optimal performance. We further establish the state evolution (SE) analysis to characterize the behaviour of the proposed TSA-GA algorithm, and derive an explicit bound of the expected loss reduction of the FEEL system under certain standard regularity conditions. In addition, we develop an expectation maximization (EM) strategy to learn the unknown parameters in the Markovian model. We show that the proposed TSAGA algorithm significantly outperforms the state-of-the-art, and is able to achieve comparable learning performance as the error-free benchmark in terms of both convergence rate and final test accuracy.

cs.IT↗

arXiv2020

Apparent Liquid Permeability in Mixed-Wet Shale Permeable Media

Apparent liquid permeability (ALP) in ultra-confined permeable media is primarily governed by the pore confinement and fluid-rock interactions. A new ALP model is required to predict the interactive effect of the above two on the flow in mixed-wet, heterogeneous nanoporous media. This study derives an ALP model and integrates the compiled results from molecular dynamics (MD) simulations, scanning electron microscopy, atomic force microscopy, and mercury injection capillary pressure. The ALP model assumes viscous forces, capillary forces, and liquid slippage in tortuous, rough pore throats. Predictions of the slippage of water and octane are validated against MD data reported in the literature. In up-scaling the proposed liquid transport model to the representative-elementary-volume scale, we integrate the geological fractals of the shale rock samples including their pore size distribution, pore throat tortuosity, and pore-surface roughness. Sensitivity results for the ALP indicate that when the pore size is below 100 nm pore confinement allows oil to slip in both hydrophobic and hydrophilic pores, yet it also restricts the ALP due to the restricted intrinsic permeability. The ALP reduces to the well-established Carman-Kozeny equation for no-slip viscous flow in a bundle of capillaries, which reveals a distinguishable liquid flow behavior in shales versus conventional rocks. Compared to the Klinkenberg equation, the proposed ALP model reveals an important insight into the similarities and differences between liquid versus gas flow in shales.

physics.flu-dyn↗

arXiv2020

Universal Particle Kinetic Distribution in Crowded Environments

We study many-particle transport in heterogeneous, crowded environments at different particle P\'{e}clet numbers ($Pe^*$). We demonstrate that a modified Nakagami-$m$ function describes particle velocity probability distributions when particle deposition occurs. We assess the universality of said function through comparison against new Lagrangian simulations of various particle types as well as experimental data from the literature. We construe the function's physical meaning as its ability to explain particle deposition in terms of $Pe^*$ and the competition between distributions of energy barriers for particle release and particles' diffusive energy.

cond-mat.soft↗

arXiv2015

Constructing Intrinsic Delaunay Triangulations from the Dual of Geodesic Voronoi Diagrams

Intrinsic Delaunay triangulation (IDT) is a fundamental data structure in computational geometry and computer graphics. However, except for some theoretical results, such as existence and uniqueness, little progress has been made towards computing IDT on simplicial surfaces. To date the only way for constructing IDTs is the edge-flipping algorithm, which iteratively flips the non-Delaunay edge to be locally Delaunay. Although the algorithm is conceptually simple and guarantees to stop in finite steps, it has no known time complexity. Moreover, the edge-flipping algorithm may produce non-regular triangulations, which contain self-loops and/or faces with only two edges. In this paper, we propose a new method for constructing IDT on manifold triangle meshes. Based on the duality of geodesic Voronoi diagrams, our method can guarantee the resultant IDTs are regular. Our method has a theoretical worst-case time complexity $O(n^2\log n)$ for a mesh with $n$ vertices. We observe that most real-world models are far from their Delaunay triangulations, thus, the edge-flipping algorithm takes many iterations to fix the non-Delaunay edges. In contrast, our method is non-iterative and insensitive to the number of non-Delaunay edges. Empirically, it runs in linear time $O(n)$ on real-world models.

cs.CG↗