Ke Ma, Jiaqi Tang, Bin Guo, Xueting Han, Ruonan Xu, Qingfeng He, Ziheng Wang, Xu Wang, Qifeng Chen, Zhiwen Yu, Yunhao Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Bluetooth Low Energy (BLE) is a prevalent technology in various applications due to its low power consumption and wide device compatibility. Despite its numerous advantages, the encryption methods of BLE often expose devices to potential attacks. To fortify security, we investigate the application of Physical-layer Key Generation (PKG), a promising technology that enables devices to generate a shared secret key from their shared physical environment. Although extensively investigated, PKG is generally discussed in the context of Wi-Fi, and existing solutions for BLE demonstrate significantly lower performance. To bridge this gap, we propose a distinctive approach that capitalizes on the inherent characteristics of BLE to facilitate efficient PKG. We utilize the constant tone extension within BLE protocols to extract comprehensive physical layer information and introduce an innovative method that employs Legendre polynomial quantization for PKG. This method facilitates the exchange of secret keys with a high key matching rate and a high key generation rate. The efficacy of our approach is validated through extensive experiments on a software-defined radio platform, underscoring its potential to enhance security in the rapidly expanding field of BLE applications. A pilot study on commercial off-the-shelf BLE devices further validates the system's practicality, revealing important trade-offs between performance and hardware constraints in real-world deployments.

In this paper, we propose a dynamic data transmission strategy for smart home environments that aims to optimize the Quality of Experience (QoE) by adaptively adjusting the data upload frequency based on the predicted trends in sensor data. Using the home wireless sensors monitoring dataset, we implement a deep learning model for accurate time series forecasting. In addition, an anomaly detection mechanism is used to identify critical events, requiring more frequent data uploads when important changes are detected. The QoE is quantified through a weighted average of several influencing factors, including data timeliness, timely upload of critical events, and transmission frequency. Our optimization objective is to maximize QoE while minimizing the number of transmissions, with an emphasis on reducing energy consumption through intelligent scheduling. The results demonstrate that our approach effectively balances data timeliness, transmission efficiency, and energy savings, leading to improved user satisfaction in smart home applications.

Bluetooth Low Energy (BLE) Mesh is a pivotal multi-hop self-organizing network in the Internet of Things (IoT) domain, offering low power consumption, low cost, and robustness. This paper presents a comprehensive study on the communication performance of BLE-Mesh using commercial off-the-shelf devices, focusing on the impact of key mesh parameters such as transmission power, packet interval, and network structure on performance. Through extensive indoor and outdoor experiments, we quantify the impact of these parameters and conduct a detailed study. Our findings provide insights into the actual communication range of BLE-Mesh, the effect of node design on overall network performance, and the configuration for optimal performance. The research contributes to the establishment of a BLE-Mesh network in real-world environments, answering critical questions for practitioners, and offering a reference for future BLE-Mesh deployments. This work furthers our understanding of the characteristics, challenges, and future directions of BLE-Mesh, setting the stage for advancements in IoT applications such as smart offices and homes.

Bluetooth Low Energy (BLE) is a prevalent technology in various applications due to its low power consumption and wide device compatibility. Despite its numerous advantages, the encryption methods of BLE often expose devices to potential attacks. To fortify security, we investigate the application of Physical-layer Key Generation (PKG), a promising technology that enables devices to generate a shared secret key from their shared physical environment. We propose a distinctive approach that capitalizes on the inherent characteristics of BLE to facilitate efficient PKG. We harness the constant tone extension within BLE protocols to extract comprehensive physical layer information and introduce an innovative method that employs Legendre polynomial quantization for PKG. This method facilitates the exchange of secret keys with a high key matching rate and a high key generation rate. The efficacy of our approach is validated through extensive experiments on a software-defined radio platform, underscoring its potential to enhance security in the rapidly expanding field of BLE applications.

This article introduces LinkStream, a liquidity analysis system based on multiple video streams designed and implemented for oilfield. LinkStream combines a variety of technologies to solve several problems in computing power and network latency. First, the system adopts an edge-central architecture and tailoring based on spatio-temporal correlation, which greatly reduces computing power requirements and network costs, and enables real-time analysis of large-scale video stream on limited edge devices. Second, it designed a set of liquidity information to describe the liquidity status in the oilfield. Finally, it uses object tracking technology to design a counting algorithm for the unique tubing object in the oilfield. We have deployed LinkStream in an oilfield in Iraq. LinkStream can perform real-time inference on over 200 video streams with acceptable resource overhead.

The digital transformation of factories has greatly increased the number of peripherals that need to connect to a network for sensing or control, resulting in a growing demand for a new network category known as the Equipment Area Network (EAN). The EAN is characterized by its cable-free, high-capacity, low-latency, and low-power features. To meet these expectations, we presentBEANet, a novel solution designed specifically for EAN that combines a two-stage synchronization mechanism with a time division protocol. We implemented the system using commercially available Bluetooth Low Energy (BLE) modules and evaluated its performance. Our results show that the network can support up to 150 peripherals with a packet reception rate of 95.4%, which is only 0.9% lower than collision-free BLE transmission. When the cycle time is set to 2 s, the average transmission latency for all peripherals is 0.1 s, while the power consumption is 18.9 μW, which is only half that of systems using LLDN or TSCH. Simulation results also demonstrate that BEANet has the potential to accommodate over 30,000 peripherals under certain configurations.

Industrial Identifiers (IID), such as GS1, Handle, and OID are fundamental to device identification in growing industrial networks. Appropriate resolution and addressing methods for those identifiers are designed for wide area networks (WAN). However, due to compatibility, efficiency, and security considerations, they are not suitable for local area networks (LANs) environments. Therefore, we propose a new industrial identification framework to handle LAN scenarios by industrial address. It mainly includes the Industrial Identifier Resolution Protocol (IIRP), which combines the IIRP table lookup, the IIRP request, and the response based on the data link layer frame transmission. It is implemented as a software plug-in on LAN devices without changing any network protocol or hardware, ensuring compatibility with existing network infrastructure. Our experiments also test the efficiency of the IIRP protocol.

Rapid advances in programmable network devices catalyzed the development of in-network computing, which is foreseen as a key enabler to empower the intelligence of production lines and mechanical arms in Industry 4.0. Various pioneering approaches have demonstrated the significant benefits of moving simple yet delay-sensitive industrial control tasks performed by servers to network switches. However, our detailed field study at a top-tier auto glass factory reveals that current practice fails to achieve a real-time and deterministic intelligent decision closure as leaving those complex yet essential planning tasks still on edge or cloud. In this paper, we design and implement a brand-new industrial switch, named Netopia, on a commercial Zynq platform through software and hardware co-design. Netopia enables planning and control to simultaneously perform on a network switch during communication. At the core of Netopia are three simple yet effective modules - a determinism guarantee mechanism, a computing acceleration scheme, and a packet deterministic forwarding framework that work hand-in-hand to ensure mechanical arms obtain intelligent control commands with low and deterministic latency. Comprehensive evaluations in industrial environments demonstrate that Netopia achieves an average end-to-end intelligent decision latency of 3.0ms with a jitter < 0.4ms, reduced by > 86% over existing works.

Future mobile devices are anticipated to perceive, understand and react to the world on their own by running multiple correlated deep neural networks locally on-device. Yet the complexity of these deep models needs to be trimmed down both within-model and cross-model to fit in mobile storage and memory. Previous studies squeeze the redundancy within a single model. In this work, we aim to reduce the redundancy across multiple models. We propose Multi-Task Zipping (MTZ), a framework to automatically merge correlated, pre-trained deep neural networks for cross-model compression. Central in MTZ is a layer-wise neuron sharing and incoming weight updating scheme that induces a minimal change in the error function. MTZ inherits information from each model and demands light retraining to re-boost the accuracy of individual tasks. MTZ supports typical network layers (fully-connected, convolutional and residual) and applies to inference tasks with different input domains. Evaluations show that MTZ can fully merge the hidden layers of two VGG-16 networks with a 3.18% increase in the test error averaged on ImageNet for object classification and CelebA for facial attribute classification, or share$39.61\%$parameters between the two networks with$<0.5\%$increase in the test errors. The number of iterations to retrain the combined network is at least$17.8\times$lower than that of training a single VGG-16 network. Moreover, MTZ can effectively merge nine residual networks for diverse inference tasks and models for different input domains. And with the model merged by MTZ, the latency to switch between these tasks on memory-constrained devices is reduced by$8.71{\times}$.

Accurate, real-time object detection on resource-constrained devices enables autonomous mobile vision applications such as traffic surveillance, situational awareness, and safety inspection, where it is crucial to detect both small and large objects in crowded scenes. Prior studies either perform object detection locally on-board or offload the task to the edge/cloud. Local object detection yields low accuracy on small objects since it operates on low-resolution videos to fit in mobile memory. Offloaded object detection incurs high latency due to uploading high-resolution videos to the edge/cloud. Rather than either pure local processing or offloading, we propose to detect large objects locally while offloading small object detection to the edge. The key challenge is to reduce the latency of small object detection. Accordingly, we develop EdgeDuet, the first edge-device collaborative framework for enhancing small object detection with tile-level parallelism. It optimizes the offloaded detection pipeline in tiles rather than the entire frame for high accuracy and low latency. Evaluations on drone vision datasets under LTE, WiFi 2.4GHz, WiFi 5GHz show that EdgeDuet outperforms local object detection in small object detection accuracy by 233.0%. It also improves the detection accuracy by 44.7% and latency by 34.2% over the state-of-the-art offloading schemes.

Intelligent personal and home applications demand multiple deep neural networks (DNNs) running on resourceconstrained platforms for compound inference tasks, known as multitask inference. To fit multiple DNNs into low-resource devices, emerging techniques resort to weight sharing among DNNs to reduce their storage. However, such reduction in storage fails to translate into efficient execution on common accelerators such as GPUs. Most DNN graph rewriters are blind for multi-DNN optimization, while GPU vendors provide inefficient APIs for parallel multi-DNN execution at runtime. A few prior graph rewriters suggest cross-model graph fusion for low-latency multi-DNN execution. Yet they request duplication of the shared weights, erasing the memory saving of weight-shared DNNs. In this paper, we propose MTS, a novel graph rewriter for efficient multitask inference with weight-shared DNNs. MTS adopts a model stitching algorithm which outputs a single computational graph for weight-shared DNNs without duplicating any shared weight. MTS also utilizes a model grouping strategy to avoid overwhelming the GPU when co-running tens of DNNs. Extensive experiments show that MTS accelerates multitask inference by up to 6.0× compared to sequentially executing multiple weightshared DNNs. MTS also yields up to 2.5× lower latency and 3.7× less memory usage compared with NETFUSE, a state-of-the-art multi-DNN graph rewriter.

Accurate forecast of citywide passenger demand helps online car-hailing service providers to better schedule driver supplies. Previous research either uses only passenger order history and fails to capture the deep dependency of passenger demand, or is restricted on grid region partition that loses physical context. Recent advance in mobile traffic analysis has fostered understanding of city functions. In this article, we propose FlowFlexDP, a demand prediction model that integrates regional crowd flow and applies to flexible region partition. Analysis on a cellular dataset covering 1.5 million users in a major city in China reveals strong correlation between passenger demand and crowd flow. FlowFlexDP extracts both order history and crowd flow from cellular data, and adopts Graph Convolutional Neural Network to adapt prediction for regions of arbitrary shapes and sizes in a city. Evaluation on a large scale data set of 6 online car-hailing applications from cellular data shows that FlowFlexDP accurately predicts passenger demand and outperforms the state-of-the-art demand prediction methods.

This article introduces LinkStream, a liquidity modeling system based on multiple video streams designed and implemented for oilfield. LinkStream combines a variety of technologies to solve several problems in computing power and network latency. First, the system adopts an edge-central architecture and tailoring based on spatio-temporal correlation, which greatly reduces computing power requirements and network costs, and enables real-time analysis of large-scale video stream on limited edge devices. Second, it designed a set of liquidity models to describe the liquidity status in the oilfield. Finally, it uses object tracking technology to design a counting algorithm for the unique tubing object in the oilfield. We have deployed LinkStream in an oilfield in Iraq. LinkStream can perform real-time inference on over 200 video streams with acceptable resource overhead.

Accurate, real-time object detection on resource-constrained devices enables autonomous mobile vision applications such as traffic surveillance, situational awareness, and safety inspection, where it is crucial to detect both small and large objects in crowded scenes. Prior studies either perform object detection locally on-board or offload the task to the edge/cloud. Local object detection yields low accuracy on small objects since it operates on low-resolution videos to fit in mobile memory. Offloaded object detection incurs high latency due to uploading high-resolution videos to the edge/cloud. Rather than either pure local processing or offloading, we propose to detect large objects locally while offloading small object detection to the edge. The key challenge is to reduce the latency of small object detection. Accordingly, we develop EdgeDuet, the first edge-device collaborative framework for enhancing small object detection with tile-level parallelism. It optimizes the offloaded detection pipeline in tiles rather than the entire frame for high accuracy and low latency. Evaluations on drone vision datasets under LTE, WiFi 2.4GHz, WiFi 5GHz show that EdgeDuet outperforms local object detection in small object detection accuracy by 233.0%. It also improves the detection accuracy by 44.7% and latency by 34.2% over the state-of-the-art offloading schemes.

Accurate forecast of citywide crowd flows on flexible region partition benefits urban planning, traffic management, and public safety. Previous research either fails to capture the complex spatiotemporal dependencies of crowd flows or is restricted on grid region partition that loses semantic context. In this paper, we propose DeepFlowFlex, a graph-based model to jointly predict inflows and outflows for each region of arbitrary shape and size in a city. Analysis on cellular datasets covering 2.4 million users in China reveals dependencies and distinctive patterns of crowd flows in not only the conventional space and time domains, but also the speed domain, due to the diverse transportation modes in the mobility data. DeepFlowFlex explicitly groups crowd flows with respect to speed and time, and combines graph convolutional long short-term memory networks and graph convolutional neural networks to extract complex spatiotemporal dependencies, especially long-term and long-distance inter-region dependencies. Evaluations on two big cellular datasets and public GPS trace datasets show that DeepFlowFlex outperforms the state-of-the-art deep learning and big-data-based methods on both grid and non-grid city map partition.

Understanding customer mobility patterns to commercial districts is crucial for urban planning, facility management, and business strategies. Trade areas are a widely applied measure to quantify where the visitors are from. Traditional trade area analysis is limited to small-scale or store-level studies, because information such as visits to competitor commercial entities and place of residence is collected by labour-intensive questionnaires or heavily biased location-based social media data. In this article, we propose CellTradeMap, a novel district-level trade area analysis framework using mobile flow records (MFRs), a type of fine-grained cellular network data. We show that compared to traditional cellular data and social network check-in data, MFRs can model customer mobility patterns comprehensively at urban scale. CellTradeMap extracts robust location information from the irregularly sampled, noisy MFRs, adapts the generic trade area analysis framework to incorporate cellular data, and enhances the original trade area model with cellular-based features. We evaluate CellTradeMap on two large-scale cellular network datasets covering 3.5 million and 1.8 million mobile phone users in two metropolis in China, respectively. Experimental results show that the trade areas extracted by CellTradeMap are aligned with domain knowledge and CellTradeMap can model trade areas with a high predictive accuracy.

Understanding customer mobility patterns to commercial districts is crucial for urban planning, facility management, and business strategies. Trade areas are a widely applied measure to quantity where the visitors are from. Traditional trade area analysis is limited to small-scale or store-level studies because information such as visits to competitor commercial entities and place of residence is collected by labour-intensive questionnaires or heavily biased location-based social media data. In this paper, we propose CellTradeMap, a novel district-level trade area analysis framework using mobile flow records (MFRs), a type of fine-grained cellular network data. CellTradeMap extracts robust location information from the irregularly sampled, noisy MFRs, adapts the generic trade area analysis framework to incorporate cellular data, and enhances the original trade area model with cellular-based features. We evaluate CellTradeMap on a large-scale cellular network dataset covering 3.5 million mobile phone users in a metropolis in China. Experimental results show that the trade areas extracted by CellTradeMap are aligned with domain knowledge and CellTradeMap can model trade areas with a high predictive accuracy.

Understanding and predicting cellular traffic at large-scale and fine-granularity is beneficial and valuable to mobile users, wireless carriers, and city authorities. Predicting cellular traffic in modern metropolis is particularly challenging because of the tremendous temporal and spatial dynamics introduced by diverse user Internet behaviors and frequent user mobility citywide. In this paper, we characterize and investigate the root causes of such dynamics in cellular traffic through a big cellular usage dataset covering 1.5 million users and 5,929 cell towers in a major city of China. We reveal intensive spatio-temporal dependency even among distant cell towers, which is largely overlooked in previous works. To explicitly characterize and effectively model the spatio-temporal dependency of urban cellular traffic, we propose a novel decomposition of in-cell and inter-cell data traffic, and apply a graph-based deep learning approach to accurate cellular traffic prediction. Experimental results demonstrate that our method consistently outperforms the state-of-the-art time-series based approaches and we also show through an example study how the decomposition of cellular traffic can be used for event inference.

Accurate forecast of citywide passenger demand helps online car-hailing service providers to better schedule driver supplies. Previous research either uses only passenger order history and fails to capture the deep dependency of passenger demand, or is restricted on grid region partition that loses physical context. Recent advance in mobile traffic analysis has fostered understanding of city functions. In this paper, we propose FlowFlexDP, a demand prediction model that integrates regional crowd flow and applies to flexible region partition. Analysis on a cellular dataset covering 1.5 million users in a major city in China reveals strong correlation between passenger demand and crowd flow. FlowFlexDP extracts both order history and crowd flow from cellular data, and adopts Graph Convolutional Neural Network to adapt prediction for regions of arbitrary shapes and sizes in a city. Evaluation on a large scale data set of DiDi Chuxing from cellular data shows that FlowFlexDP accurately predicts passenger demand and outperforms the state-of-the-art demand prediction methods.

Modern mobile devices are equipped with multiple antennas, which brings various wireless sensing applications such as accurate localization, contactless human detection, and wireless human-device interaction. A key enabler for these applications is phased array signal processing, especially Angle of Arrival (AoA) estimation. However, accurate AoA estimation on commodity devices is non-trivial due to limited number of antennas and uncertain phase offsets. Previous works either rely on elaborate calibration or involve contrived human interactions. In this paper, we aim to enable practical AoA measurements on commodity off-the-shelf (COTS) mobile devices. The key insight is to involve users' natural rotation to formulate a virtual spatial-temporal antenna array and conduce a relative incident signal of measurements at two orientations. Then by taking the differential phase, it is feasible to remove the phase offsets and derive the accurate AoA of the equivalent incoming signal, while the rotation angle can also be captured by built-in inertial sensors. On this basis, we propose Differential MUSIC (D-MUSIC), a relative form of the standard MUSIC algorithm that eliminates the unknown phase offsets and achieves accurate AoA estimation on COTS mobile devices with only one rotation. We further extend DMUSIC to 3-D space, integrate extra measurements during rotations for higher estimation accuracy, and fortify it in multipath-rich scenarios. We prototype D-MUSIC on commodity WiFi infrastructure and evaluate it in typical indoor environments. Experimental results demonstrate a superior performance with average AoA estimation errors of 130 with only three measurements and 50 with at most 10 measurements. Requiring no modifications or calibration, D-MUSIC is envisioned as a promising scheme for practical AoA estimation on COTS mobile devices.

Understanding and predicting cellular traffic at large-scale and fine-granularity is beneficial and valuable to mobile users, wireless carriers and city authorities. Predicting cellular traffic in modern metropolis is particularly challenging because of the tremendous temporal and spatial dynamics introduced by diverse user Internet behaviours and frequent user mobility citywide. In this paper, we characterize and investigate the root causes of such dynamics in cellular traffic through a big cellular usage dataset covering 1.5 million users and 5,929 cell towers in a major city of China. We reveal intensive spatio-temporal dependency even among distant cell towers, which is largely overlooked in previous works. To explicitly characterize and effectively model the spatio-temporal dependency of urban cellular traffic, we propose a novel decomposition of in-cell and inter-cell data traffic, and apply a graph-based deep learning approach to accurate cellular traffic prediction. Experimental results demonstrate that our method consistently outperforms the state-of-the-art time-series based approaches and we also show through an example study how the decomposition of cellular traffic can be used for event inference.

Outdoor localization is of great importance for driving navigation, attracting many research efforts in past decades. Prevailing GPS achieves meter-level localization accuracy under general outdoor conditions. Yet, GPS service performs poorly in urban canyons where skyscrapers blocks GPS signals and drain mobile phone battery quickly within few hours. In this work, we exploit common city facilities, i.e. speed humps, as an indicator for vehicle location. The key insight is that when the vehicle passes through the speed bump, it experiences significant fluctuations, causing larger acceleration in the vertical direction. On this basis, we design a localization scheme that utilizes the accelerator equipped on modern smart phones to track sequence of speed bumps, which is further transferred into sequence of moving directions of the vehicle, and adopt effective road mapping technology to derive real-time location. As we have concerned, it is the first attempt to exploit the spatiotemporal characteristics generated by the speed humps to recover the trajectory of the travelling route and infer the current position. Experimental results in typical outdoor environment (campus) demonstrate a comparable performance with GPS method, yet achieve lower energy consumption.

Modern mobile devices are equipped with multiple antennas, which brings various wireless sensing applications such as accurate localization, contactless human detection and wireless human-device interaction. A key enabler for these applications is phased array signal processing, especially Angle of Arrival (AoA) estimation. However, accurate AoA estimation on commodity devices is non-trivial due to limited number of antennas and uncertain phase offsets. Previous works either rely on elaborate calibration or involve contrived human interactions. In this paper, we aim to enable practical AoA measurements on commodity off-the-shelf (COTS) mobile devices. The key insight is to involve users' natural rotation to formulate a virtual spatial-temporal antenna array and conduce a relative incident signal of measurements at two orientations. Then by taking the differential phase, it is feasible to remove the phase offsets and derive the accurate AoA of the equivalent incoming signal, while the rotation angle can also be captured by built-in inertial sensors. On this basis, we propose Differential MUSIC (D-MUSIC), a relative form of the standard MUSIC algorithm that eliminates the unknown phase offsets and achieves accurate AoA estimation on COTS mobile devices with only one rotation. We further extend D-MUSIC to 3-D space and fortify it in multipath-rich scenarios. We prototype D-MUSIC on commodity WiFi infrastructure and evaluate it in typical indoor environments. Experimental results demonstrate a superior performance with an average AoA estimation error of 13°. Requiring no modifications or calibration, D-MUSIC is envisioned as a promising scheme for practical AoA estimation on COTS mobile devices.

In this demonstration, we proposed a hot plugging tool for the real-time motion-based system control, which is more portable and application-independent than the existing commercial motion-based sensing devices such as Kinect, Wii and PlayStation Move. This tool captures and recognizes people's real-time motions through the built-in camera of PCs, mobile phones or tablets, and automatically executes the system events which have been mapped with people's customized body motion, e.g., the head and the fist. The tool relieves people from the conventional ways to play games and use applications, and enables them to customize their preferred ways to control the systems.