This paper presents DriveTrack, a new benchmark and data generation framework for long-range keypoint tracking in real-world videos. DriveTrack is motivated by the observation that the accuracy of state-of-the-art trackers depends strongly on visual attributes around the selected keypoints, such as texture and lighting. The problem is that these artifacts are especially pronounced in real-world videos, but these trackers are unable to train on such scenes due to a dearth of annotations. DriveTrack bridges this gap by building a framework to automatically annotate point tracks on autonomous driving datasets. We release a dataset consisting of 1 billion point tracks across 24 hours of video, which is seven orders of magnitude greater than prior real-world benchmarks and on par with the scale of synthetic benchmarks. DriveTrack unlocks new use cases for point tracking in real-world videos. First, we show that fine-tuning keypoint trackers on DriveTrack improves accuracy on real-world scenes by up to 7%. Second, we analyze the sensitivity of trackers to visual artifacts in real scenes and motivate the idea of running assistive keypoint selectors alongside trackers.

The rapid proliferation of commodity drones has expanded interest in building applications that acquire imagery, video, and sensor data at scale. In addition, recent work on drone programming frameworks have simplified the development of aerial computing apps that gather this data. These advancements have popularized the drones-as-a-service model, where large drone fleets serve multiple apps simultaneously. This paper proposes a marketplace for aerial computing, where apps can gather aerial data on demand and providers can offer up their drones for aerial computing. We introduce Aerialis, a drones-as-a-service platform that schedules tasks to drones by arbitrating bids submitted by apps. Aerialis allows apps with different semantics and spatiotemporal preferences to express how much they would like to pay for each aerial computing task. It then aggregates requests across apps, and schedules tasks on drones according to a marketplace policy (e.g., maximizing revenue or guaranteeing quality-of-service to apps). We build a prototype of Aerialis, and implement urban sensing apps to monitor air pollution, measure road traffic, and profile cellular throughput. We discuss operational challenges in deploying Aerialis, and show how the measurements collected from our real-world experiments offer valuable insights for engineers and city planners.

This paper studies the problem of allocating tasks from different customers to vehicles in mobility platforms, which are used for applications like food and package delivery, ridesharing, and mobile sensing. A mobility platform should allocate tasks to vehicles and schedule them in order to optimize both throughput and fairness across customers. However, existing approaches to scheduling tasks in mobility platforms ignore fairness. We introduce Mobius, a system that uses guided optimization to achieve both high throughput and fairness across customers. Mobius supports spatiotemporally diverse and dynamic customer demands. It provides a principled method to navigate inherent tradeoffs between fairness and throughput caused by shared mobility. Our evaluation demonstrates these properties, along with the versatility and scalability of Mobius, using traces gathered from ridesharing and aerial sensing applications. Our ridesharing case study shows that Mobius can schedule more than 16,000 tasks across 40 customers and 200 vehicles in an online manner.

The rapid development of small aerial drones has enabled numerous drone-based applications, e.g., geographic mapping, air pollution sensing, and search and rescue. To assist the development of these applications, we propose BeeCluster, a drone orchestration system that manages a fleet of drones. BeeCluster provides a virtual drone abstraction that enables developers to express a sequence of geographical sensing tasks, and determines how to map these tasks to the fleet efficiently. BeeCluster's core contribution is predictive optimization , in which an inferred model of the future tasks of the application is used to generate an optimized flight and sensing schedule for the drones that aims to minimize the total expected execution time. We built a prototype of BeeCluster and evaluated it on five real-world case studies with drones in outdoor environments, measuring speedups from 11.6% to 23.9%.

Video databases that enable queries with object-track predicates are useful in many applications. Such queries include selecting objects that move from one region of the camera frame to another (e.g., finding cars that turn right through a junction) and selecting objects with certain speeds (e.g., finding animals that stop to drink water from a lake). Processing such predicates efficiently is challenging because they involve the movement of an object over several video frames. We propose a novel query-driven tracking approach that integrates query processing with object tracking to efficiently process object track queries and address the computational complexity of object detection methods. By processing video at low framerates when possible, but increasing the framerate when needed to ensure high-accuracy on a query, our approach substantially speeds up query execution. We have implemented query-driven tracking in MIRIS, a video query processor, and compare MIRIS against four baselines on a diverse dataset consisting of five sources of video and nine distinct queries. We find that, at the same accuracy, MIRIS accelerates video query processing by 9x on average over the IOU tracker, an overlap-based tracking-by-detection method used in existing video database systems.

We present BurstTracker, the first tool that developers can use to detect if the LTE downlink is the bottleneck for their applications. BurstTracker is driven by our discovery that the proprietary LTE downlink schedulers running on LTE base stations allocate resources to users in a way that reveals if a user's downlink queue runs empty during a download. We demonstrate that BurstTracker works across Tier-1 cellular providers and across a variety of network conditions. We also present a case study that shows how application developers can use this tool in practice. Surprisingly, with BurstTracker, we find that the LTE downlink may not be the bottleneck for video streaming on several Tier-1 providers, even during peak hours at busy locations. Rather, transparent TCP middleboxes deployed by these providers lead to downlink underutilization, because they force Slow-Start Restart. With a simple workaround, we improve video streaming bitrate on busy LTE links by 35%.

The rapid growth in mobile connectivity is enabling phones to support a wide range of societally-important applications. In this work, we show that broad classes of popular mobile applications have distinct radio resource allocation signatures. Using this insight, we design a mobile application classifier, and demonstrate that (1) an application can infer its own type solely from its resource allocation patterns, and (2) anyone can accurately infer the type of application being served by each session on a particular cell tower. We present our findings by showing the breakdown of applications being served by an LTE base station belonging to a Tier 1 US provider in downtown Palo Alto. Our work encourages an open discussion about LTE standards, and whether they might need to be enhanced to mask features that can be exploited to infer application type from signals broadcast over the air.

Operating a distributed data stream processing workload efficiently at scale is hard. The operator of the workload must parallelize and lay out tasks of the workload with resources that match the requirement of target data rate. The challenge is that neither the operator nor the programmer is typically aware of the scaling behavior of the workload as a function of resources. An operator manually searches for a safe operating point that can handle predicted peak load and deploys with ample headroom for absorbing unpredictable spikes. Such empirical, static over-provisioning is wasteful of both compute and human resources. We show that precise performance models can be automatically learned for distributed stream processing systems that can predict the execution performance of a job even before deployment. Further, those models can be used to optimally schedule logically specified jobs onto available physical hardware. Finally, those models and the derived execution schedules can be refined online to dynamically adapt to unpredictable changes in the runtime environment or auto-scale with variations in job load.

The rapid growth in mobile connectivity is enabling phones to support a wide range of societally-important applications. In this work, we show that broad classes of popular mobile applications have distinct radio resource allocation signatures. Using this insight, we design a mobile application classifier, and demonstrate that (1) an application can infer its own type solely from its resource allocation patterns, and (2) anyone can accurately infer the type of application being served by each session on a particular cell tower. We present our findings by showing the breakdown of applications being served by an LTE base station belonging to a Tier 1 US provider in downtown Palo Alto. Our work encourages an open discussion about LTE standards, and whether they might need to be enhanced to mask features that can be exploited to infer application type from signals broadcast over the air.

Carbon nanotube field-effect transistors (CNFETs) are promising candidates for building energy-efficient digital systems at highly scaled technology nodes. However, carbon nanotubes (CNTs) are inherently subject to variations that reduce circuit yield, increase susceptibility to noise, and severely degrade their anticipated energy and speed benefits. Joint exploration and optimization of CNT processing options and CNFET circuit design are required to overcome this outstanding challenge. Unfortunately, existing approaches for such exploration and optimization are computationally expensive, and mostly rely on trial-and-error-based ad hoc techniques. In this paper, we present a framework that quickly evaluates the impact of CNT variations on circuit delay and noise margin, and systematically explores the large space of CNT processing options to derive optimized CNT processing and CNFET circuit design guidelines. We demonstrate that our framework: 1) runs over 100× faster than existing approaches and 2) accurately identifies the most important CNT processing parameters, together with CNFET circuit design parameters (e.g., for CNFET sizing and standard cell layouts), to minimize the impact of CNT variations on CNFET circuit speed with ≤5% energy cost, while simultaneously meeting circuit-level noise margin and yield constraints.