MobiCom 2025 Papers

Non-invasive continuous glucose monitoring for diabetes patients remains challenging despite ongoing interest. This paper presents CGMM, a novel non-invasive wireless glucose monitoring system integrated into wearable devices. It features a specially designed metasurface that couples with the wearable's antenna and tissue fluid beneath the skin, amplifying frequency response changes caused by subtle glucose concentration variations. To address individual tissue variability and optimize the passive metasurface design, we develop a tunable metasurface and a one-shot calibration method to obtain the impedance for optimal resonance in glucose sensing environments with unknown parameters. The calibrated impedance is then used for the inverse design and fabrication of an economical passive metasurface. We implement prototypes of CGMM and conduct extensive experimental evaluations. In human experiments involving ten participants using the prototype with LibreVNA, the overall performance is quantified with relative errors ranging from -5.02% to 6.93% and an RMSE of 9.65 mg/dL.

Offering high-quality immersive content is the ultimate goal of volumetric video streaming. Although point clouds and meshes are dominant volumetric representations, their limitations in depicting photo-realistic content often undermine user experience. The recent advent of neural radiance fields (NeRF) offers a promising alternative content representation with superior photo-realism. However, streaming NeRF-based volumetric videos over wireless networks to mobile headsets faces significant challenges, including substantial bandwidth usage because of the large frame size, degraded visual quality due to even a low packet loss rate, and content artifacts caused by performance optimizations (e.g., remote rendering at the network edge). To address these challenges, in this paper, we introduce NeVo, a next-generation volumetric video streaming system for efficient delivery of neural content such as NeRF. NeVo incorporates the following innovations into a holistic system: (1) a novel method to model visibility of implicitly encoded neural content, thereby avoiding non-essential transmission to drastically reduce network data usage, (2) a lightweight, learning-based model for real-time content reconstruction after packet loss with carefully chosen data, and (3) judicious identification and selective delivery of intermediate data in edge-based NeRF rendering to effectively mitigate artifacts. Our extensive experiments indicate that compared with the state-of-the-art, NeVo saves up to 68.3% of bandwidth usage, maintains high visual quality despite packet loss, and enhances user experience by reducing artifacts.

Battery-free sensing has revolutionized IoT applications, but current solutions relying on signal variations between transmitted and backscattered signals remain vulnerable to environmental dynamics and deployment variations. This paper promotes a paradigm shift: inferring targets through antenna impedance variations instead of signal fluctuations, thereby eliminating the impact of unpredictable wireless communication. We demonstrate the effectiveness of this paradigm by reimplementing three existing applications: RIO [1], Keystub [2], and RF-EATS [3]. Compared to original signal-based implementations, our approach shows significant improvements in accuracy and robustness across diverse environments. Furthermore, By integrating antenna engineering with advanced materials science, we also transform antennas into innovative sensors for pressure, temperature, and UV light sensing. This interdisciplinary methodology pushes the boundaries of battery-free sensing, opening new avenues for IoT applications.

In this paper, we propose LoRaSeek, a lightweight and reliable LoRa denoising framework that enhances signal quality and robustness for neural-enhanced LoRa decoding. LoRaSeek integrates a hybrid architecture combining Convolutional Neural Networks (CNNs), Transformers, and a hierarchical U-Net to effectively capture multi-scale, multidimensional features of LoRa chirp signals. To maintain efficiency, we integrate a lightweight Transformer block that supports various LoRa configurations while keeping computational overhead low. Additionally, we incorporate dual attention-based skip connections to preserve chirp signal properties across different scales. Experiments across diverse LoRa configurations show that LoRaSeek achieves 2.04–3.86 dB signal-to-noise ratio (SNR) gains over standard decoding methods and up to 3.03 dB improvement over state-of-the-art neural-enhanced LoRa decoding methods while reducing model storage by up to 7.4× and inference time by up to 1.6×.

Multimodal reasoning by LLMs is critical to autonomous mobile systems, but the growing diversity of input data modalities prevents incorporating all modalities into LLMs. Instead, only the useful modalities should be adaptively involved at runtime, based on the current environmental contexts and task requirements. Existing work on runtime modality adaptation uses fixed connections between data encoders and LLM's input layer, but results in high training costs and ineffective cross-modal interaction. In this paper, we present MPnP, a new modality adaptation technique that connects data encoders to a flexible set of last LLM blocks and makes such latent connections fully trainable at runtime. Evaluation results show that MPnP has high compute and data efficiency, with 3.7× FLOPs reduction and 30% memory usage reduction compared to best baselines. It requires only few hundreds of training samples at runtime, and completes modality adaptation within few minutes on weak devices.

Bluetooth Low Energy (BLE) backscatter is a promising candidate for battery-free Internet of Things (IoT) applications. We propose PassiveBLE, a backscatter system that can establish authentic and fully compatible BLE connections on data channels. The key enabling techniques include (i) a synchronization circuit that can wake up tags and activate backscatter communications with symbol-level accuracy to facilitate BLE data packet generation; (ii) a distributed coding scheme that offloads the major encoding and processing burdens from tags to the excitation source while achieving high throughput; (iii) a BLE connection scheduler to enable fully compatible BLE connection interactions, including connection establishment, maintenance and termination for multiple backscatter tags. We prototype PassiveBLE tags with off-the-shelf components and also conduct comprehensive experiments to evaluate the performance. Results demonstrate that PassiveBLE achieves a success rate of over 99.9% in establishing commodity BLE connections. PassiveBLE also achieves commodity-compatible BLE communication with a high goodput of up to 974 kbps in LE 2M PHY mode and 532 kbps in LE 1M PHY mode, which is about 63.3× higher than existing commodity-level BLE backscatter system in the same mode.

Tremor, a prevalent symptom in various neurological disorders, significantly impacts patients' quality of life. Regular and precise tremor monitoring is essential for optimizing treatment effectiveness. Existing at-home monitoring solutions have limitations in daily life due to low adherence, privacy concerns, or inevitable interference from confounding body components and overlapped motions. In this paper, we propose mmTremor, the first privacy-preserving contactless system to achieve practical tremor detection during activities of daily living (ADL) in real-world settings. We design a mmWave-depth fusion tracking algorithm to handle the high interference from confounding body components and a multimodal spatiotemporal deep learning pipeline to effectively exploit latent tremor information. Additionally, a spatial contrastive unsupervised adaptation is proposed for better discriminability and adaptation capability to unseen domains. To fully assess mmTremor in real-world scenarios, we collect a diverse dataset of 28 patients and 9 healthy subject simulations in over 20 distinct environments, including offices, hospitals, and homes. Extensive evaluations demonstrate that mmTremor achieves a high macro-F1 of 0.877 in tremor detection, showcasing the potential as a transformative contactless solution for tremor monitoring in daily life. The dataset will be open-source to facilitate future research.

On-device Large Language Models (LLMs) are transforming mobile AI, catalyzing applications like UI automation without privacy concerns. Nowadays the common practice is to deploy a single yet powerful LLM as a general task solver for multiple requests. We identify a key system challenge in this paradigm: current LLMs lack the elasticity to serve requests that have diversified Service-Level Objectives (SLOs) on inference latency. To tackle this, we present ElastiLM, an on-device LLM service that elasticizes both the model and the prompt dimension of a full LLM. It incorporates (1) a one-shot neuron-reordering method, which leverages the intrinsic permutation consistency in transformer models to generate high-quality elasticized sub-models with minimal runtime switching overhead; (2) a dual-head tiny language model, which efficiently and effectively refines the prompt and orchestrates the elastification between model and prompt. We implement such an elastic on-device LLM service on multiple COTS smartphones, and evaluate ElastiLM on both standalone NLP/mobile-agent datasets and end-to-end synthesized traces. On diverse SLOs, ElastiLM outperforms 7 strong baselines in (absolute) accuracy by up to 14.83% and 10.45% on average, with <1% TTFT switching overhead, on-par memory consumption and <100 offline GPU hours.

Conventional pulmonary function tests (PFTs) are important but costly. Hence, prior research has proposed IoT sensor-based solutions to facilitate cost-efficient, at-home PFT. However, these solutions require the subject to perform maximal exhalations, a task often challenging without supervision, compromising test accuracy. In response to this challenge, this study introduces EasySpiro that, for the first time, uses non-maximal exhalations to measure PFT indicators. This is challenging since PFT indicators are only defined for maximal exhalations, and there are no guidelines to derive them from submaximal exhalations. To address that, we observe that pulmonary deficiencies affect all types of breathing, where the underlying pulmonary deficiency should be the same under different breathing efforts. Leveraging this insight, we design a reconstruction model to predict the ideal maximal breathing patterns based on submaximal ones and utilize these reconstructions for PFT. Furthermore, since the body dynamics reflect the exhalation effort, we use self-supervised learning techniques to encode body dynamics into breathing effort representations to guide the reconstruction process. We integrate these designs into earphones with microphones to measure breathing patterns and IMUs to measure body dynamics. We collaborate with a hospital and develop a dataset from 50 patients with various diseases to evaluate EasySpiro's performance, which shows an accurate prediction of PFT indicators based on non-maximal exhalations with an error rate of 7%. In addition, we open-source the collected dataset to encourage future research.

Vehicle detection in tunnels is crucial for traffic monitoring and accident response, yet remains underexplored. In this paper, we develop mmTunnel, a millimeter-wave radar system that achieves far-range vehicle detection in tunnels. The main challenge here is coping with ghost points caused by multi-path reflections, which lead to severe localization errors and false alarms. Instead of merely removing ghost points, we propose correcting them to true vehicle positions by recovering their signal reflection paths, thus reserving more data points and improving detection performance, even in occlusion scenarios. However, recovering complex 3D reflection paths from limited 2D radar points is highly challenging. To address this problem, we develop a multi-path ray tracing algorithm that leverages the ground plane constraint and identifies the most probable reflection path based on signal path loss and spatial distance. We also introduce a curve-to-plane segmentation method to simplify tunnel surface modeling such that we can significantly reduce the computational delay and achieve real-time processing.

Photovoltaic (PV) defect detection is crucial for preventing power efficiency loss and fire hazards. The industry primarily relies on the fusion of infrared and visible images for defect localization and diagnosis. However, current detection methods exhibit poor generalizability in new site environments or with altered imaging setups. While recent infrared and vision foundation models (FM) facilitate domain-invariant feature maps extraction, directly concatenating them and fine-tuning achieves limited generalizability gain to PV defect detection, due to the asymmetric dual-modal semantics of defects. In this paper, we present the first large infrared-visible model DD-LIVM to enable cross-domain defect detection. The key innovation of DD-LIVM lies in its defect-specific three-step fine-tuning strategy, which utilizes alternating modality masking. Prior to feature fusion and joint fine-tuning, the infrared and visible FM encoders are alternately masked and optimized to enhance their individual semantic utility for defect localization visibility and classification granularity, with feature distances among different defect types regulated through contrastive learning. This approach allows for the extraction of generalizable and defect-specific feature maps. Moreover, for practical employment of DD-LIVM, we propose a domain-agnostic spatial alignment algorithm for infrared-visible images before dual-modal fusion, and develop source data augmentation and adaptive detection head selection schemes based on defects' infrared characteristics to further enhance the generalizability. Extensive experiments on 7,078 dual-modal images from 9 real-world scenarios across 4 cities' PV stations demonstrate that DD-LIVM achieves an accuracy of 87.7% for cross-domain defect detection, surpassing state-of-the-art methods by 17.3%.

For highly immersive mobile volumetric video streaming, it is essential to deliver photo-realistic full-scene content with smooth playback. Unlike traditional representations such as point clouds, 3D Gaussian Splatting (3DGS) has gained attention for its ability to represent high-quality full-scene 3D content. However, our preliminary experiments show that existing methods for 3DGS-based videos fail to achieve smooth playback on mobile devices. In this paper, we propose Vega, a 3DGS-based photo-realistic full-scene volumetric video streaming system that ensures real-time playback on mobile devices. The core idea behind Vega's real-time rendering is object-level selective computation, which allocates computational resources to visually important objects to meet strict rendering deadlines. To enable mobile streaming based on the selective computation, Vega addresses two challenges: (1) designing an encoding scheme that optimizes the data size of videos while being compatible with object-level prioritization, and (2) developing a rendering pipeline that efficiently operates on resource-constrained mobile devices. We implemented an end-to-end Vega system, consisting of a streaming server and an Android application. Experimental results on commodity smartphones show that Vega achieves 30 frames per second (FPS) for full-scene volumetric video streaming while maintaining competitive data size and visual quality compared to existing baselines.

There is a growing trend in deploying DNNs on tiny microcontroller (MCUs) to provide inference capabilities in the IoT. While prior research has explored many lightweight techniques to compress DNN models, achieving overall efficiency in model inference requires not only model optimization but also careful system resource utilization for execution. Existing studies primarily leverage arithmetic logic units (ALUs) for integer-only computations on a single CPU core. Floating-point units (FPU) and multi-core capabilities available in many existing MCUs remain underutilized. To fill this gap, we propose AdaptQNet, a novel MCU neural network system that can determine the optimal precision assignment for different layers of a DNN model. AdaptQNet models the latency of various operators in DNN models across different precisions on heterogeneous processing units. This facilitates the discovery of models that utilize FPU and multi-core capabilities to enhance capacity while adhering to stringent memory constraints. Our implementation and experiments demonstrate that AdaptQNet enables the deployment of models with better accuracy-efficiency trade-off on MCUs.

The ongoing virtualization of Radio Access Networks (vRANs) promises increased velocity for updating RAN software with new features and bug fixes. To support this, we developed SwapRAN, a live update system for both vRAN software components: the Distributed Unit (DU) and the Centralized Unit (CU). Unlike previous systems, SwapRAN operates in-place without requiring additional hardware like staging servers or a programmable switch. For DU updates, SwapRAN presents two techniques: (1) using OS thread priorities to safely initialize the new DU while overlapping with the old DU which is active, and (2) using the network interface card's embedded switch to redirect fronthaul traffic to the new DU. For CU updates, SwapRAN is the first working live update system, which we achieve by (1) decoupling the stateful CU-DU connection and transparently rerouting DU messages to the new CU, and (2) repurposing existing midhaul control plane messages to move users to the new CU. We evaluate SwapRAN on real 5G testbeds and demonstrate its practical deployability via integration with Kubernetes. Our evaluations show that SwapRAN completes DU or CU updates with just 1–2 seconds of user downtime.

With the rapid growth of Low Earth Orbit (LEO) satellite networks, satellite-IoT systems using the LoRa technique have been increasingly deployed to provide widespread Internet services to low-power and low-cost ground devices. However, the long transmission distance and adverse environments from IoT satellites to ground devices pose a huge challenge to link reliability, as evidenced by the measurement results based on our real-world setup. In this paper, we propose a blind coherent combining design named B2LoRa to boost LoRa transmission performance. The intuition behind B2LoRa is to leverage the repeated broadcasting mechanism inherent in satellite-IoT systems to achieve coherent combining under the low-power and low-cost constraints, where each re-transmission at different times is regarded as the same packet transmitted from different antenna elements within an antenna array. Then, the problem is translated into aligning these packets at a fine granularity despite the time, frequency, and phase offsets between packets in the case of frequent packet loss. To overcome this challenge, we present three designs — joint packet sniffing, frequency shift alignment, and phase drift mitigation to deal with ultra-low SNRs and Doppler shifts featured in satellite-IoT systems, respectively. Finally, experiment results based on our real-world deployments demonstrate the high efficiency of B2LoRa.

Real-time video analytics on high-resolution cameras has become a popular technology for various intelligent services like traffic control and crowd monitoring. While extensive work has been done on improving analytics accuracy with timing guarantees, virtually all of them target static viewpoint cameras. In this paper, we present Uirapuru, a novel framework for real-time, edge-based video analytics on high-resolution steerable cameras. The actuation performed by those cameras brings significant dynamism to the scene, presenting a critical challenge to existing popular approaches such as frame tiling. To address this problem, Uirapuru incorporates a comprehensive understanding of camera actuation into the system design paired with fast adaptive tiling at a per-frame level. We evaluate Uirapuru on a high-resolution video dataset, augmented by pan-tilt-zoom (PTZ) movements typical for steerable cameras and on real-world videos collected from an actual PTZ camera. Our experimental results show that Uirapuru provides up to 1.45× improvement in accuracy while respecting specified latency budgets or reaches up to 4.53× inference speedup with on-par accuracy compared to state-of-the-art static camera approaches.

Federated AIoT uses distributed data on IoT devices to train AI models. However, in practical AIoT systems, heterogeneous devices cause data heterogeneity and varying amounts of device staleness, which can reduce model performance or increase federated training time. When addressing the impact of device delays, existing FL frameworks improperly consider it as independent from data heterogeneity. In this paper, we explore a scenario where device delays and data heterogeneity are closely correlated, and propose FedDC, a new technique to mitigate the impact of device delays in such cases. Our basic idea is to use gradient inversion to learn knowledge about device's local data distribution and use such knowledge to compensate the impact of device delays on devices' model updates. Experiment results on heterogeneous IoT devices show that FedDC can improve the FL performance by 34% with high amounts of device delays, without impairing the devices' local data privacy.

The conventional wisdom has been that designing ultra-compact, battery-constrained wireless hearables with on-device speech AI models is challenging due to the high computational demands of streaming deep learning models. Speech AI models require continuous, real-time audio processing, imposing strict computational and I/O constraints.

The Logical Link Control and Adaptation Protocol (L2CAP) is a core Bluetooth component, and verifying its correctness is crucial for reliable and secure connectivity. However, verification can be challenging due to the complexity and ambiguities in its natural language (English) specification. In this paper, we present a formally verified implementation of the L2CAP state machine. Our approach introduces the Specification State Machine (SSM) to formalize the L2CAP state machine in the specification and the Operational State Machine (OSM) as an abstraction of the implementation. We then formally prove that (i) OSM refines SSM, and (ii) our implementation semantically conforms to OSM. By combining these two proofs, we verify that our implementation complies with our formalization of the specification. Furthermore, we define critical safety and liveness properties and formally prove that our implementation satisfies these guarantees. To ensure practicality, we implement the L2CAP state machine in Dafny and integrate it into Android's Fluoride Bluetooth stack. Our evaluation demonstrates that our formally verified implementation maintains competitive performance while ensuring formal correctness.