With the proliferation of smart devices and widespread Internet connectivity, social sensing is advancing as a pervasive sensing paradigm where experiences shared by individuals on social platforms (e.g., Twitter and Facebook) are analyzed to interpret the physical world. In this article, we introduce CovidTrak, a vision of social intelligence-empowered contact tracing that aims to scrutinize the knowledge derived using social sensing to track Coronavirus Disease 2019 (COVID-19) infections among the general public. Contact tracing is known to be an effective technique for detecting and monitoring persons who may have been exposed to individuals infected with any communicable disease. While a good number of contact tracing schemes are existent today (e.g., in-person and phone interviews, paper forms, email and web-based questionnaires, and smartphone apps), they often require active user participation and might miss certain cases of social interactions that go off-the-records but still lead to COVID-19 transmission. By contrast, social sensing provides an alternative avenue for spontaneously determining such contacts by harnessing the rich experiences and information conveyed by people on social data platforms (e.g., a group photograph tweeted from a house party with a potential contact). As such, CovidTrak can form a powerful basis to combat the COVID-19 pandemic. The vision of CovidTrak intends to answer the following questions: 1) how to bolster the privacy and security of the online users while determining their contacts? 2) how to collect relevant social signals that indicate in-person encounters among people? 3) how to reliably process the vast amount of noisy data from social platforms to identify chains of transmission? 4) how to handle the scarcity of location metadata in the incoming data? 5) how to effectively communicate crucial contact information to concerned individuals? and 6) how to model and handle the responses of the common people toward contact information? We envision unexplored opportunities to leverage multidisciplinary techniques to address the above questions and develop effective future CovidTrak schemes.

Social airborne sensing (SAS) is emerging as a new sensing paradigm that leverages the complementary aspects of social sensing and airborne sensing (i.e., UAVs) for reliable information collection. In this paper, we present HeteroSAS, a heterogeneous resource management framework for "all-in-the-air" SAS in disaster response applications. Current SAS approaches use UAVs to only capture data, but carry out computation on ground-based processing nodes that may be unavailable in disaster scenarios and thus consider a single model of UAV along with only one type of task (i.e., data capture). In this paper, we explore the opportunity to exploit the complementary strengths of different UAV models to accomplish all stages of sensing tasks (i.e., data capturing, maneuvering, and computation) exclusively "in-the-air". However, several challenges exist in developing such a resource management framework: i) handling the uncertain social signals in presence of the heterogeneity of UAVs and tasks; and ii) adapting to constantly changing cyber-physical-social environments. The HeteroSAS framework addresses these challenges by building a novel resource management framework that observes the environment and learns the optimal strategy for each UAV using techniques from multi-agent reinforcement learning, game theory, and ensemble learning. The evaluation with a real-world case study shows that HeteroSAS outperforms the state-of-the-art in terms of detection effectiveness, deadline hit rate, and robustness on heterogeneity.

In the domain of smart city applications, anomalistic crowd investigation in public places is a crucial task for the efficient management of unforeseen crowds. Many of the current approaches for crowd monitoring utilize physical sensor-based tracking (e.g., cellular signals, cameras, RFID beacons) to detect abrupt crowds. Unfortunately, these solutions are highly application-specific and are able to only detect the mere presence of crowds without determining their causes that may reveal the actual story behind gatherings that occur suddenly. In this paper, we explore the opportunity to meld two complementary sensing paradigms: social sensing and unmanned aerial vehicle (UAV)-based physical sensing to scrutinize abrupt crowd events occurring in public places. However, two technical challenges exist in developing our solution: i) handling the data sparsity across social sensing platforms and within video footage from UAVs; and ii) the accurate inference of the cause of crowd events from the collected data. To address these challenges, we develop Unravel, an intelligent social airborne sensing (SAS) framework that leverages interdisciplinary techniques from natural language processing (NLP), object recognition, and estimation theory for a comprehensive anomalistic crowd investigation. The evaluation results with a real-world anomalous crowd instance dataset show that the Unravel framework outperforms the state-of-the-art baselines in accurately validating the existence of anomalistic crowd instances in public areas and deriving their causes.

Image-driven smart urban sensing (ISUS) has emerged as a powerful sensing paradigm to capture abundant visual information about the urban environment for intelligent city monitoring, planning, and management. In this paper, we focus on a Classification and Super-resolution Coupling (CSC) problem in ISUS applications, where the goal is to explore the interdependence between two critical tasks (i.e., classification and super-resolution) to concurrently boost the Quality of Service (QoS) of both tasks. Previous efforts often focus on solving the two tasks individually and ignore the opportunity to explore the interdependence between them. In this paper, we develop SuperClass, a deep duo-task learning framework, to effectively integrate the classification and super-resolution tasks into a holistic network design that jointly optimizes the QoS of both tasks. The evaluation results on a real-world ISUS application show that SuperClass consistently outperforms state-of-the-art baselines by simultaneously achieving better land usage classification accuracy and higher reconstructed image quality.

Forest fires cause irreversible damages worldwide every year. Monitoring wildfire propagation is thus a vital task in mitigating forest fires. While computational model-based wildfire prediction methods provide reasonable accuracy in monitoring wildfire behavior, they are often limited due to the lack of constant availability of real-time meteorological data. In contrast, social-media-driven drone sensing (SDS) is emerging as a new sensing paradigm that detects the early signs of forest fires from online social media feeds and drives the drones for reliable sensing. However, due to the scarcity of social media data in remote regions and limited flight times of drones, SDS solutions often underperform in large-scale forest fires. In this paper, we present CompDrone, a wildfire monitoring framework that exploits the collective strengths of computational wildfire modeling and SDS for reliable wildfire monitoring. Two critical challenges exist to integrate computational modeling and SDS together: i) limited availability of social signals in the regions of a forest fire; and ii) predicting the regions of fire where the drones should be dispatched to. To solve the above challenges, the CompDrone framework leverages techniques from cellular automata, constrained optimization, and game theory. The evaluation results using a real-world wildfire dataset show that CompDrone outperforms the state-of-the-art schemes in effectively predicting wildfire propagation.

Social media sensing has emerged as a new disaster response application paradigm to collect real-time observations from online social media users about the disaster status. Due to the noisy nature of social media data, the task of identifying trustworthy information (referred to as "truth discovery") has been a crucial task in social media sensing. However, existing truth discovery solutions often fall short of providing accurate results in disaster response applications due to the spread of misinformation and difficulty of an efficient verification in such scenarios. In this paper, we present SocialDrone, a novel closed-loop social-physical active sensing framework that integrates social media and unmanned aerial vehicles (UAVs) for reliable disaster response applications. In SocialDrone, signals emitted from the social media are distilled to drive the drones to target areas to verify the emergency events. The verification results are then taken back to improve the sensing and distillation process on social media. The SocialDrone framework introduces several unique challenges: i) how to drive the drones using the unreliable social media signals? ii) How to ensure the system is adaptive to the high dynamics from both the physical world and social media? iii) How to incorporate real-world constraints (e.g., the deadlines of events, limited number of drones) into the framework? The SocialDrone addresses these challenges by building a novel integrated social-physical sensing system that leverages techniques from game theory, constrained optimization, and reinforcement learning. The evaluation results on a real-world disaster response application show that SocialDrone significantly outperforms state-of-the-art truth discovery schemes and drone-only solutions by providing more effective disaster response.

Recent advances in remote sensing open up unprecedented opportunities to obtain a rich set of visual features of objects on the earth's surface. In this paper, we focus on a single-image super-resolution (SISR) problem in remote sensing, where the objective is to generate a reconstructed satellite image of high quality (i.e., a high spatial resolution) from a satellite image of relatively low quality. This problem is motivated by the lack of high quality satellite images in many remote sensing applications (e.g., due to the cost of high resolution sensors, communication bandwidth constraints, and historic hardware limitations). Two important challenges exist in solving our problem: i) it is not a trivial task to reconstruct a satellite image of high quality that meets the human perceptual requirement from a single low quality image; ii) it is challenging to rigorously quantify the uncertainty of the results of an SISR scheme in the absence of ground truth data. To address the above challenges, we develop PQA-CNN, a perceptual quality-assured conventional neural network framework, to reconstruct a high quality satellite image from a low quality one by designing novel uncertainty-driven neural network architectures and integrating an uncertainty quantification model with the framework. We evaluate PQA-CNN on a real-world remote sensing application on land usage classifications. The results show that PQA-CNN significantly outperforms the state-of-the-art super-resolution baselines in terms of accurately reconstructing high-resolution satellite images under various evaluation scenarios.

Recent advances in remote sensing provide a powerful and scalable sensing paradigm to capture abundant visual information about the urban environments. We refer to such a sensing paradigm as remote urban sensing. In this paper, we focus on a migratable satellite image super-resolution problem in remote urban sensing applications. Our goal is to reconstruct satellite images of a high resolution in a target area where the high-resolution training data is not available by transferring a super-resolution model learned in a source area where such data is available. This problem is motivated by the limitation of current solutions that primarily rely on a rich set of high-resolution satellite images in the studied area that are not always available. Two important challenges exist in solving our problem: i) the target and source areas often have very different urban characteristics that prevent the direct application of a super-resolution model learned from the source area to the target area; ii) it is not a trivial task to ensure effective model migration with desirable quality without sufficient high quality training data. To address the above challenges, we develop TransRes, a deep adversarial transfer learning framework, to effectively reconstruct high-resolution satellite images without requiring any ground-truth training data from the studied area. We evaluate the TransRes framework using the real-world satellite imagery data collected from three different cities in Europe. The results show that TransRes consistently outperforms the state-of-the-art baselines by achieving the lowest perception errors under various application scenarios.

While autonomous unmanned aerial vehicles (UAVs) have attained a reputable stance in modern disaster response applications, their practical adoptions are impeded due to various constraints (e.g., requiring manual input, battery life). In this paper, we develop a novel spatiotemporal aware drone sensing system that is driven by harnessing social media signals, a process known as social sensing. Social sensing has emerged as a new sensing paradigm where humans act as "sensors" to report their observations about the physical world. However, maneuvering drones with "social signals" introduces a new realm of challenges. The first challenge is to drive the drones by leveraging noisy and unreliable social media signals. The second challenge is to optimize the drone deployment by exploring the highly dynamic and latent correlations among event locations. In this paper, we present CollabDrone that devises a new spatiotemporal correlation inference model and game-theoretic drone dispatching mechanism to address the above challenges. The evaluation results on a real-world case study show that CollabDrone significantly outperforms current drone and social sensing baselines in terms of accuracy and deadline hit rate.

Autonomous unmanned aerial vehicles (UAVs) have become an important tool for efficient disaster response. Despite the virtues of UAVs in disaster response applications, various limitations (e.g., requiring manual input, finite battery life) hinder their mass adoption. In contrast, social sensing is emerging as a new sensing paradigm that utilizes signals provided by "human sensors" to gather awareness of the events occurring in the physical world. Despite being inherently broader in scope, a shortcoming of social sensing is the reliability of the sensing data that are contributed by humans. In this paper, we introduce the concept of jointly exploiting the reliability of drones and the scope of social sensing to efficiently uncover the truthful events during disasters. However, such a tight integration of social and physical sensing introduces several technical challenges. The first challenge is satisfying the conflicting objectives of event coverage of the application and energy conservation of drones. The second challenge is adapting to the dynamics of the physical world and social media. In this paper, we present a Social-media-driven Energy-Aware Drone (SEAD) sensing framework to address the above challenges. In particular, we develop a reinforcement learning-based drone dispatching scheme that adapts to the physical and social environments and launches an appropriate proportion of drones for event exploration. We further utilize a bottom-up game-theoretic task allocation approach to guide drones effectively to the event locations. The evaluation with a real-world disaster case study show that SEAD noticeably outperforms state-of-the-art baselines in terms of detection effectiveness and energy efficiency.

Vehicular sensor networks (VSNs) have become a reliable sensing instrument utilizing sensors built into cars. However, VSNs have limited sensing scopes since car drivers only opportunistically discover new events. Conversely, social sensing is protruding as a new sensing paradigm where measurements about the physical world are collected from humans. In this paper, we develop SocialCar, a task allocation framework for social media driven vehicular network sensing systems that exploits the collective powers of social sensing and VSNs for a robust sensing application. However, integrating VSNs with social sensing introduces a new set of challenges. The first challenge is leveraging the noisy and unreliable social signals to route vehicles to the regions of interest. The second challenge is the inconsistent availability attributed to car drivers being rational actors causing churn in the system. To address the above challenges, this paper introduces a bottom-up game-theoretic task preference model along with a top-down dynamic incentive control mechanism to allow car drivers to be selectively dispatched to event locations and verify the truthfulness of reported events. The results on a real-world disaster recovery case study show that SocialCar significantly outperforms current VSNs based solutions in detection effectiveness and deadline hit rate.

Modern Internet of Things (IoT) systems are increasingly leveraging deep neural networks (DNNs) with the goal of enabling intelligence at the edge of the network. While applying DNNs can greatly improve the accuracy of autonomous decisions and inferences, a significant challenge is that DNNs are traditionally designed and developed for advanced hardware (e.g., GPU clusters) and can not easily meet the real time requirements when deployed in a resource-constrained edge computing environment. While many systems have been proposed to facilitate deep learning at the edge, a key limitation lies in the under-utilization of the parallelizable GPU resources of edge nodes (e.g., IoT devices). In this paper, we propose EdgeBatch, a collaborative intelligent edge computing framework that minimizes the delay and energy consumption of executing DNN tasks at the edge by sharing idle GPU resources among privately owned IoT devices. EdgeBatch develops 1) a stochastic task batching mechanism that identifies the optimal batching strategy for the GPUs on IoT devices given uncertain task arrival times, and 2) a dynamic task offloading scheme that coordinates the collaboration among edge nodes to optimize the utilization of idle GPU resources in the system. We implemented EdgeBatch on a real-world edge computing testbed that consists of heterogeneous IoT devices (Jetson TX2, TX1, TK1, and Raspberry Pi3s). The results show that EdgeBatch achieved significant performance gains in terms of both the end-to-end delay and energy savings compared to the state-of-the-art baselines.

Social sensing based Edge Computing (SSEC) is an emerging application paradigm for rich context awareness in sensing applications in which people collaborate in both data acquisition and processing at the edge of the network. While keeping people in the loop can be an immense benefit, the unreliability and churn introduced can pose a dangerous stability threat for long-running SSEC applications. In the past this problem was addressed by offloading these responsibilities to more reliable server hardware in the cloud or a fog layer, however, this does not take full advantage of the power at the edge. In this paper we address the issue of reliable edge computing on dynamic high-churn edge systems. We develop a deviceless pipeline based approach (DPA) to establish workflows in which stages of the analysis pipeline are completed on edge devices, and any devices that leave the system can be replaced without data loss. We evaluate the performance of our system on a real-world edge computing system performing an object detection application and demonstrate that it can provide significant performance gains over traditional computation offloading schemes in terms of throughput and error recovery.