MobiCom 2023 Papers

On-device Deep Neural Network (DNN) inference consumes significant computing resources and development efforts. To alleviate that, we propose LUT-NN, the first system to empower inference by table lookup, to reduce inference cost. LUT-NN learns the typical features for each operator, named centroid, and precompute the results for these centroids to save in lookup tables. During inference, the results of the closest centroids with the inputs can be read directly from the table, as the approximated outputs without computations.

Existing inertial measurement unit (IMU) based human activity recognition (HAR) approaches still face a major challenge when adopted across users in practice. The severe heterogeneity in IMU data significantly undermines model generalizability in wild adoption. This paper presents UniHAR, a universal HAR framework for mobile devices. To address the challenge of data heterogeneity, we thoroughly study augmenting data with the physics of the IMU sensing process and present a novel adoption of data augmentations for exploiting both unlabeled and labeled data. We consider two application scenarios of UniHAR, which can further integrate federated learning and adversarial training for improved generalization. UniHAR is fully prototyped on the mobile platform and introduces low overhead to mobile devices. Extensive experiments demonstrate its superior performance in adapting HAR models across four open datasets.

As we move towards a future of connected and autonomous vehicles, high-speed and low-latency connectivity between vehicles is becoming increasingly important. This paper investigates enabling high data rate mmWave links in vehicle-to-vehicle (V2V) scenarios using street view images. We find that mmWave V2V links in urban settings suffer from frequent and prolonged blockages, resulting in unreliable connection and high beamforming overhead. Our work proposes mmSV, a system that creates 3D reflection profiles from street view images to assist vehicles in finding mmWave reflections from the environment in real-time. mmSV consists of two key components: material identification which identifies materials from street view images to determine their reflectivity and create 3D reflection map, and environment-driven ray-tracing and beamsearching which finds a high-SNR beam using predicted 3D material maps. Our extensive experimental results on the mmWave testbed show that mmSV can provide highly reliable V2V mmWave connectivity with low beamforming overhead.

Neglecting proper oral hygiene has proven to potentially lead to severe oral disease, resulting in complications over time. Careful brushing can mitigate the problem, but it is common for individuals to dedicate insufficient time to the various areas of their teeth. We propose LiT to monitor the brushing situation of 16 Bass technique surfaces in real-time. LiT relies on commercial toothbrushes with blue LEDs as a transmitter and requires only 2 low-cost photosensors as receivers on the toothbrush head. However, the transmission channel of light in the oral cavity is unclear. Finding the optimal deployment positions and minimizing the number of photosensors is challenging. To tackle these obstacles, we design the positioning of the 2 photosensors and create a transmission model within the oral cavity to verify the feasibility theoretically. Additionally, obstacles in implementation include separating brushing action accurately, interference of light on the outer surfaces of front teeth, and individual variability. To overcome these challenges, we develop corresponding technologies and a comprehensive framework. Experiments with 16 users show that LiT achieves a highly accurate recognition rate of 95.3% with an error estimate for brushing duration of 6.1%. Furthermore, LiT also proves resilient under user motion and environmental interference.

This paper presents the design, implementation, and evaluation of TaGroup, a fast, fine-grained, and robust grouping technique for RFIDs. It can achieve a nearly 100% accuracy in distinguishing multiple groups of closely located RFIDs, within only a few seconds. It would benefit many inventory tracking applications, such as self-checkout in retails and packaging quality control in logistics.

This paper tries to answer a question: "Can we achieve spatial-selective transmission on IoT devices?" A positive answer would enable more secure data transmission among IoT devices. The challenge, however, is how to manipulate signal propagation without relying on beamforming antenna arrays which are usually unavailable on low-end IoT devices.

We present MilliMobile: a first of its kind battery-free autonomous robot capable of operating on harvested solar and RF power. We challenge the conventional assumption that motion and actuation are beyond the capabilities of battery-free devices and demonstrate completely untethered autonomous operation in realistic indoor and outdoor lighting as well as RF power delivery scenarios. We show first that through miniaturizing a robot to gram scale, we can significantly reduce the energy required to move it. Second, we develop methods to produce intermittent motion by discharging a small capacitor (47--150 μF) to move a motor in discrete steps, enabling motion from as little as 50 μW of power or less. We further develop software defined techniques for maximizing power harvesting. MilliMobile operates in the optimal part of the charging curve by varying the charging time to achieve maximum speeds of up to 5.5 mm/s.

A severe bandwidth mismatch between incoming sensor data rate and wireless backhaul bandwidth often exists on unmanned probes when collecting new training data for machine learning (ML). To overcome this mismatch, we describe a self-improving ML-based transmission system called Hawk. Starting from a weak model that is trained on just a few examples, it seamlessly pipelines semi-supervised learning, active learning, and transfer learning, with asynchronous bandwidth-sensitive data transmission to a distant human for labeling. When a significant number of true positives (TPs) have been labeled, Hawk trains an improved model to replace the old model. This iterative workflow, called Live Learning, continues until a sufficient number of TPs have been collected. For very rare events on challenging datasets, and bandwidths as low as 12 kbps, a team of 7 probes using Hawk discovers up to 87% of the TPs that could have been discovered via full preview, transmission and labeling of all mission data. Hawk also uses diversity sampling and few-shot learning.

This paper proposes a new primitive that allows soft robots to be physically controlled in a completely non-line-of-sight context using wireless energy - a process we call wireless actuation. Soft robots, which are composed entirely of soft materials and exclude any rigid components, are highly flexible platforms that can change their shape. This paper considers a specific class of soft robots composed of liquid-crystal elastomers (LCE) that are entirely electronics-free and engineered to change shape when heated to 60 °C. Traditionally, such robotic systems must be in line-of-sight of a light source, such as infrared to be moved, or require an external power supply for Joule heating and often take several tens of seconds to heat. We present WASER, a novel RF-based heating platform that allows electronics-free robots to be actuated rapidly (within a few seconds) and potentially in non-line-of-sight. WASER achieves this through innovations in both wireless systems and material science. On the wireless front, WASER develops a new blind beamforming solution that directs high-power wireless energy at fine spatial granularity without electronics on the robot to provide feedback. On the material science front, WASER exhibits heat-responsive shape-morphing and energy-harvesting material functionalities that allow for rapid wireless heating. We implement and evaluate WASER and demonstrate diverse shape-morphing capabilities.

Continual learning (CL) trains NN models incrementally from a continuous stream of tasks. To remember previously learned knowledge, prior studies store old samples over a memory hierarchy and replay them when new tasks arrive. Edge devices that adopt CL to preserve data privacy are typically energy-sensitive and thus require high model accuracy while not compromising energy efficiency, i.e., cost-effectiveness. Our work is the first to explore the design space of hierarchical memory replay-based CL to gain insights into achieving cost-effectiveness on edge devices. We present Miro, a novel system runtime that carefully integrates our insights into the CL framework by enabling it to dynamically configure the CL system based on resource states for the best cost-effectiveness. To reach this goal, Miro also performs online profiling on parameters with clear accuracy-energy trade-offs and adapts to optimal values with low overhead. Extensive evaluations show that Miro significantly outperforms baseline systems we build for comparison, consistently achieving higher cost-effectiveness.

Given the fact that WiFi-based sensing can be realized by reusing WiFi communication facilities and communication frequency bands, integrated sensing and communication (ISAC) is considered a crucial development direction for future WiFi standards, such as IEEE 802.11bf. Traditional WiFi sensing systems extract channel state information (CSI) from customized WiFi packets to quantify the characteristics of the sensing target. This poses challenges for existing WiFi systems originally designed for communication purposes, as it requires high-quality and sufficient CSI measurements. In this paper, we propose SenCom, which extracts CSI from general WiFi packets. SenCom enables CSI calibration across different WiFi communication modes and provides unified CSI measurements for upper-layer sensing applications. We also devise a fitting-resampling scheme to derive evenly sampled CSI with consistent dimensionality, and an incentive strategy to ensure sufficient CSI measurements over time. We build a prototype of SenCom and perform extensive experiments with 15 participants. The results show that SenCom is competent for a variety of sensing tasks, while incurring little compromise to the WiFi communication performance.

As the demand for precision medicine rapidly grows, companion diagnostics is proposed to monitor and evaluate therapeutic effects for adjusting medicine plans in time. Although a set of clinical companion diagnostics tools (e.g., polymerase chain reaction) have been investigated, they are expensive and only accessible in a lab environment, which hinders the promotion to broader patients. In light of this situation, we take the first steps towards developing a real-world companion diagnostic tool by leveraging mobile technology. In this paper, we present TherapyPal, a privacy-preserving medicine effectiveness computational framework by harnessing semantic hashing-based digital symptomatic phenotyping. Specifically, sensor data captured from daily-life activities is first transformed into spectrograms. Then, we develop a hashing learning network to extract privacy-masked symptomatic phenotypes on smartphones. Afterward, symptomatic hashes at different medicine states are fed to a contrastive learning network in the cloud for treatment effectiveness detection. To evaluate the performance, we conduct a clinical study among 65 Parkinson's disease (PD) patients under dopaminergic drug treatment. The results show that TherapyPal can achieve around 84.1% medicine effectiveness detection accuracy among patients and above 0.925 privacy-masked scores for protecting each private attribute, which validates the reliability and security of TherapyPal to be used as a real-world companion diagnostics tool.

Multipath transport allows the simultaneous use of diverse paths on mobile devices to maximize mobile resource usage. Over the years, we have witnessed several mobile multipath deployment examples by network operators and mobile app providers. However, existing deployment methods require modifications to either the network infrastructure or both endpoints. To lower the bar of the deployment, we present Fleety, a mobile system service that provides the multi-path transport capability with client-only modification. To the best of our knowledge, we are the first to carry out a large-scale mobile multipath deployment that can support hundreds of mobile applications in the cross-ISP setting. This paper is a retrospective of our experience in building and deploying multipath transport for mobile applications. We reveal several practical deployment challenges and share our experience in dealing with them.

UAVs are becoming versatile and valuable platforms for various applications. However, the main limitation is their flying time. We present BEAVIS, a novel aerial robotic platform striking an unparalleled trade-off between the maneuverability of drones and the long-lasting capacity of blimps. BEAVIS scores highly in applications where drones enjoy unconstrained mobility yet suffer from limited lifetime. A nonlinear flight controller exploiting novel, unexplored, aerodynamic phenomena to regulate the ambient pressure and enable all translational and yaw degrees of freedom is proposed without direct actuation in the vertical direction. BEAVIS has built-in rotor fault detection and tolerance. We explain the design and the necessary background in detail. We verify the dynamics of BEAVIS and demonstrate its distinct advantages, such as agility, over existing platforms including the degrees of freedom akin to a drone with 11.36× increased lifetime. We exemplify the potential of BEAVIS to become an invaluable platform for many applications.

While a number of recent efforts have explored the use of "cloud offload" to enable deep learning on IoT devices, these have not assumed the use of duty-cycled radios like BLE. We argue that radio duty-cycling significantly diminishes the performance of existing cloud-offload methods. We tackle this problem by leveraging a previously unexplored opportunity to use early-exit offload enhanced with prioritized communication, dynamic pooling, and dynamic fusion of features. We show that our system, FLEET, achieves significant benefits in accuracy, latency, and compute budget compared to state-of-art local early exit, remote processing, and model partitioning schemes across a range of DNN models, datasets, and IoT platforms.

Mobile payment has achieved explosive growth in recent years due to its contactless feature, which lowers the infection risk of COVID-19. In the market, near-field communication (NFC) and barcodes have become the de facto standard technologies for mobile payment. The NFC-based payment outperforms barcode-based payment in terms of security, usability, and convenience. It is especially more user-friendly for the amblyopia group. Unfortunately, NFC functionality is unavailable in nearly half of smartphones in the market nowadays due to the shortage of NFC modules or being disabled for security reasons.

Given the central role mobile core plays in supporting mobile network operations, the efficiency, cost-effective dynamic scalability and resilience of the core control plane are paramount. Achieving these goals, however, presents two main challenges: (i) decoupling core network state from processing; (ii) decoupling control plane processing in the core from its interface to the radio access network (RAN). To overcome them, we present CoreKube, a novel message focused and cloud-native mobile core system design, which features truly stateless workers (processing units) that interface with a common database (to hold the core network state) and with the RAN through a frontend. The fully stateless and generic nature of the workers to process any control plane message enables efficient message handling. Orchestration of containerized CoreKube components using Kubernetes, allows leveraging the latter's autoscaling and self-healing properties. We develop 4G and 5G standard-compliant CoreKube implementations, exploiting the agile development methodology enabled by CoreKube's message focused design. Results from our extensive experimental evaluations over the Powder platform relative to prior art show that CoreKube efficiently processes control plane messages, scales dynamically while using minimal compute resources and recovers seamlessly from failures.

Tracking the position and orientation, or pose, of a viewing device enables AR applications to accurately embed virtual content in physical spaces. Mobile OSs track pose by matching device camera images against street-level imagery. Thus, pose tracking is often unavailable at off-street pedestrian locations. UbiPose enables pose tracking at such locations using aerial meshes, generated from satellite imagery, that are likely to be more widely available at these locations. However, matching a camera image against an aerial mesh can be error-prone, even with modern neural matchers. These neural components are also compute-intensive. UbiPose contains a novel pose tracking pipeline that runs entirely on a mobile device using fast-path optimizations designed to accept or reject pose estimates in many cases, without sacrificing accuracy. Experiments on real-world traces show that it achieves tracking accuracy comparable to AR pose tracking in iOS in places where that is available, and is able to track pose accurately in places where it is not.

Power governing is a critical component of modern mobile devices, reducing heat generation and extending device battery life. A popular technology of power governing is dynamic voltage and frequency scaling (DVFS), which adjusts the operating frequency of a processor to balance its performance and energy consumption. With the emergence of diverse workloads on mobile devices, traditional DVFS methods that do not consider workload characteristics become suboptimal. Recent application-oriented methods propose dedicated and effective DVFS governors for individual application tasks. Since their approach is only tailored to the targeted task, performance drops significantly when other tasks run concurrently, which is however common on today's mobile devices. In this paper, our key insight is that hardware meta-data, widely used in existing DVFS designs, has great potential to enable capable workload awareness and task concurrency adaptability for DVFS, but they are underexplored. We find that workload characteristics can be described in a hyperspace composed of multiple dimensions derived from these metadata to form a novel workload contextual indicator to profile task dynamics and concurrency. On this basis, we propose a meta-state metric to capture this relationship and design a new solution, GearDVFS. We evaluate it for a rich set of application tasks, and it outperforms state-of-the-art methods.