Prior work in RF-based positioning has mainly focused on discovering the absolute location of an RF source, where state-of-the-art systems can achieve accuracy order tens centimeters using a large number antennas. However, many applications gaming and gesture based interface see more benefits knowing detailed shape motion. Such trajectory tracing requires resolution several fold higher than what existing offer.

Location services, fundamentally, rely on two components: a mapping system and positioning system. The provides the physical map of space, identifies position within map. Outdoor location services have thrived over last couple decades because well-established platforms for both these components (e.g. Google Maps mapping, GPS positioning). In contrast, indoor haven't caught up lack reliable frameworks. Wi-Fi lacks maps is also prone to environmental errors. this paper, we present DLoc, Deep...

Prior work in RF-based positioning has mainly focused on discovering the absolute location of an RF source, where state-of-the-art systems can achieve accuracy order tens centimeters using a large number antennas. However, many applications gaming and gesture based interface see more benefits knowing detailed shape motion. Such trajectory tracing requires resolution several fold higher than what existing offer. This paper shows that one provide dramatic increase accuracy, even with small The...

This paper focuses on a simple, yet fundamental question: ``Can node infer the wireless channels one frequency band by observing different band?'' question arises in cellular networks, where uplink and downlink operate frequencies. Addressing this is critical for deployment of key 5G solutions such as massive MIMO, multi-user distributed which require channel state information.

Backscatter requires zero transmission power, making it a compelling technology for in-body communication and localization. It can significantly reduce the battery requirements (and hence size) of micro-implants smart capsules, enable them to be located on-the-move inside body. The problem however is that electrical properties human tissues are very different from air vacuum. This creates new challenges both For example, signals no longer travel along straight lines, which destroys geometric...

Large constellations of Low Earth Orbit satellites promise to provide near real-time high-resolution imagery. Yet, getting this large amount data back is challenging because their low orbits and fast motion through space. Centralized architectures with few multi-million dollar ground stations incur hour-level download latency are hard scale. We propose a geographically distributed station design, L2D2, that uses low-cost commodity hardware offer robust downlink. L2D2 the first system use...

Bluetooth Low Energy (BLE) tags have become very prevalent over the last decade for tracking applications in homes as well businesses. These are used to track objects, navigate people, and deliver contextual advertisements. However, spite of wide interest BLE tags, primary methods them based on signal strength (RSSI) measurements. Past work has shown that such inaccurate, prone multipath dynamic environments. As a result, localization using Wi-Fi moved Channel State Information (CSI,...

We study the problem of active learning for multilabel classification. focus on real-world scenario where average number positive (relevant) labels per data point is small leading to label sparsity. Carrying out mutual information based near-optimal in this setting a challenging task since computational complexity involved exponential total labels. propose novel inference algorithm sparse Bayesian model [17]. The benefit alternate scheme that it enables natural approximation objective. prove...

We present a novel Bayesian framework for facial action unit recognition. The first key observation behind this work is sparsity: out of possible 45 (and more) units, only very few are active at any moment. second the strong statistical co-occurrence structure: most expressions made by common combinations so knowing presence one can act as prior inferring others. developed graphical model that encodes these two natural aspects units via compressed sensing and group-wise sparsity inducing...

Electronic toll collection transponders, e.g., E-ZPass, are a widely-used wireless technology. About 70% to 89% of the cars in US have these devices, and some states plan make them mandatory. As devices however, they lack basic function: MAC protocol that prevents collisions. Hence, today, can be queried only with directional antennas isolated spots. However, if one could interact e-toll transponders anywhere city despite collisions, it would enable many smart applications. For example,...

Digital agriculture technologies rely on sensors, drones, robots, and autonomous farm equipment to improve yields incorporate sustainability practices. However, the adoption of such is severely limited by lack broadband connectivity in rural areas. We argue that farming applications do not require permanent always-on connectivity. Instead, activity digital follow seasonal rhythms agriculture. Therefore, need for highly localized time space. introduce BYON, a new model high bandwidth...

Underwater communication is essential for both recreational and scientific activities, such as scuba diving. However, existing methods remain highly constrained by environmental challenges often require specialized hardware, driving research into more accessible underwater solutions. While recent acoustic-based systems support text messaging on mobile devices, their low data rates severely limit broader applications. We present AquaScope, the first acoustic system capable of image...

Short video streaming systems such as TikTok, Youtube Shorts, Instagram Reels, etc have reached billions of active users. At the core is a (proprietary) recommendation algorithm which recommends sequence videos to each user, in personalized way. We aim understand temporal evolution recommendations made by these algorithms and interplay between user experience. While past work has studied using textual data (e.g., titles, hashtags, etc.) well studies interviews, we add third modality analysis...

Since the start of COVID-19 pandemic, technology enthusiasts have pushed for digital contact tracing as a critical tool breaking transmission chains. Motivated by this push, many countries and companies created apps that enable with goal to identify chain from an infected individual others early quarantine. Digital applications like AarogyaSetu in India, TraceTogether Singapore, SwissCovid Switzerland, been downloaded hundreds millions times. Yet, hasn't seen impact we envisioned at...

Although past work on RF-based indoor localization has delivered important advances, it typically makes assumptions that hinder its adoption in smart home applications. Most systems assume users carry their phones them at home, an assumption been proven highly inaccurate measurements. The few do not require the user to a device her, cannot tell identity of person; yet identification is essential most This paper focuses addressing these issues so homes can benefit from recent advances...

Low Earth Orbit satellites for observation have become very popular in recent years due to their ability take high-resolution images of the at high revisit rates. These collect hundreds GigaBytes imagery during orbit. This data needs be downloaded using ground stations on Earth. However, low altitudes, move fast with respect a station, and consequently, few minutes time window download single station. We propose geographically distributed station design, DGS, that improves robustness reduces...

We present Wi-Peep - a new location-revealing privacy attack on non-cooperative Wi-Fi devices. exploits loopholes in the 802.11 protocol to elicit responses from devices network that we do not have access to. It then uses novel time-of-flight measurement scheme locate these works without any hardware or software modifications target and requiring physical space they are deployed in. Therefore, pedestrian drone carries device can estimate location of every building. Our design costs $20...

Earth observation Low Orbit (LEO) satellites collect enormous amounts of data that needs to be transferred first ground stations and then the cloud, for storage processing. Satellites today transmit greedily stations, with full utilization bandwidth during each contact period. We show due layout orbital characteristics, this approach overloads some underloads others, leading lost throughput large end-to-end latency images. present a new scheduler system called Umbra, which plans transfers...

The world needs to sustainably grow more food feed the growing population of planet. Data-driven agriculture is a promising technique that can help farmers be productive, reduce costs, and enable adoption sustainable agricultural practices. However, data-driven limited by lack affordable technologies, for broadband, sensing, imaging, insights. In this article, we present an overview Project FarmBeats, research project started in 2014 increase We provide various components FarmBeats...

The advent of radio sensing that works through walls & obstacles challenges the notion indoor privacy. An eavesdropper can deploy such to snoop on their neighbors and a smart sensor embedded with capabilities perform large scale behavioral health data mining. We present RF-Protect, new framework enables privacy by injecting fake humans in sensed data. RF-Protect consists novel hardware reflector design modifies waves create reflections at arbitrary locations environment generative mechanism...

Unmanned aerial vehicles (UAVs) rely on optical sensors such as cameras and lidar for autonomous operation. However, are error-prone in bad lighting, inclement weather conditions including fog smoke, around textureless or transparent surfaces. In this paper, we ask: is it possible to fly UAVs without relying sensors, i.e., can seeing? We present BatMobility, a lightweight mmWave radar-only perception system that eliminates the need sensors. BatMobility enables two core functionalities --...

The time-of-flight of a signal captures the time it takes to propagate from transmitter receiver. Time-of-flight is perhaps most intuitive method for localization using wireless signals. If one can accurately measure transmitter, compute transmitter's distance simply by multiplying speed light. Today, GPS, widely used outdoor system, localizes device radio signals satellites. However, applying same concept indoor has proven difficult. Systems in spaces are expected deliver high accuracy...