key: cord-0171106-rj44spd8
authors: Rashid, Md Tahmid; Wei, Na; Wang, Dong
title: A Survey on Social-Physical Sensing
date: 2021-04-03
journal: nan
DOI: nan
sha: 1ec6517eba4ab35aaf46b1cf1207afeced8d8184
doc_id: 171106
cord_uid: rj44spd8

Propelled by versatile data capture, communication, and computing technologies, physical sensing has revolutionized the avenue for spontaneously capturing and interpreting real-world phenomenon. Despite its virtues, various limitations (e.g., high application specificity, partial autonomy, and sparse coverage) hinder physical sensing's effectiveness in critical scenarios such as disaster response. Meanwhile, social sensing is contriving as a pervasive sensing paradigm that leverages the observations from human participants equipped with portable devices and ubiquitous Internet connectivity (i.e., through social media or crowdsensing apps) to perceive the environment. While social sensing possesses a plethora of benefits, it also inherently suffers from a few drawbacks (e.g., inconsistent reliability, uncertain data provenance, and limited sensing availability). Motivated by the complementary virtues of both physical and social sensing, social-physical sensing (SPS) is protruding as an emerging sensing paradigm that tightly integrates social and physical sensors at an unprecedented scale. The vision of SPS centers on mitigating the individual weaknesses of physical and social sensing while exploiting their collective strengths in reconstructing the"state of the world", both physically and socially. While a good amount of interesting SPS applications has been explored, several important unsolved challenges and open research questions prevail in the way of developing dependable SPS systems, which require careful study to address. In this paper, we provide a comprehensive survey of SPS, with an emphasis on its definition and key enablers, state-of-the-art applications, potential research challenges, and road-map for future work. This paper intends to bridge the knowledge gap in current literature by thoroughly examining the various aspects of SPS crucial for building potent SPS systems.

With the advent of multi-faceted data acquisition, communication, and computation technologies, physical sensing has matured into an avenue for accurate and agile information absorption from the real-world. Broadly speaking, the term physical sensing refers to a sensing paradigm that leverages hardware sensors (e.g., infrared detectors, proximity sensors, and microphones) to capture the physical world stimuli. Physical sensing systems can be predominantly classified into two variants: stationary (e.g., surveillance cameras, digital thermostats) and mobile (e.g., unmanned aerial vehicles (UAV), robots, satellites) [1] . A few notable examples of applications enabled by physical sensing include: i) environmental monitoring where arrays of sensors (e.g., temperature, pressure, and humidity sensors) are utilized to assess environmental conditions [2] ; ii) traffic surveillance in which cameras are used to identify roadside incidents such as traffic accidents [3] ; iii) industrial process monitoring where lasers and scanners are used to coordinate manufacturing processes [4] ; and iv) personal fitness monitoring where wearable fitness trackers assess individuals' daily physical activities [5] .

On the other hand, fueled by the pervasive influence of human-centric sensing and widespread prevalence of Internet connectivity across portable devices, social sensing has progressed as a new sensing paradigm where knowledge contributed by humans "sensors" connected through diverse communication technologies and protocols are acquired through social data collection platforms (e.g., Twitter, Waze) to perceive real-world occurrences [6] . Social sensing can be generally categorized into two types: social media sensing and crowdsensing [7] . In social media sensing, online users proactively report occurrences around them through online social media (e.g., Twitter, Instagram, Facebook) and form virtual relationships with other users (e.g., friend or follower) [8] . In crowdsensing, interested participants connected through crowdsensing apps or websites are engaged in carrying out specialized distributed sensing tasks through different crowdsensing platforms (e.g., mobile apps such as Citizen and Waze). Examples of social sensing applications include studying human mobility in urban areas [9] ; obtaining situation awareness in the aftermath of disasters [10] , poverty prediction and mapping [11] , locating power outages in cities [12] , urban land usage classification [13] , and contact tracing of contagious diseases such as COVID-19 [14] . Figure 1 While physical sensing has an established reputation for accurately capturing raw data from the environment and swiftly transmitting the captured data using hardware devices, it is realized that physical sensing suffers from several fundamental limitations. Such limitations arise from the facts that: i) physical sensors are designed to be application-specific and are limited by the events they can sense [15] , restricting their applicability across a wide range of events (e.g., a temperature sensor can only capture the surrounding temperature while a microphone is designed to only record sound); ii) autonomous mobile physical sensing systems such as networks of unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs) require some form of human assistance to [16] ; iii) physical sensors are typically scarce resources and require to be deployed sparingly, makes their sensing coverage sparse (e.g., a set of ground robots might not be able to cover a large forest during a wildfire) [17] ; iv) fixed physical sensors such as proximity sensors and surveillance cameras are usually mounted in particular locations cannot be relocated easily [18] ; and v) physical sensors have an initial deployment cost as well as periodic maintenance costs [19] .

In contrast to physical sensing, social sensing enjoys a certain range of benefits which includes but is not limited to: i) multifaceted sensing (e.g., people who report about traffic incidents in social media can also report about crime incidents) [20] ; ii) greater mobility (e.g., human sensors tend to spontaneously move from one location to another in contrast to stationary physical sensors) [21] ; iii) lower management costs (e.g. hardware sensors require periodic maintenance and repairs in contrast to human sensors who do not require such service from the application end) [22] ; and iv) wider sensing scope due to the pervasive nature of social signals and the active participation of individuals (e.g., any person possessing a smart device with Internet connectivity can post on the social media from any part of the world) [23] . However, despite its immense benefits, social sensing also has a number of drawbacks: i) inconsistent reliability since social sensing innately relies on noisy social signals contributed by unvetted human users (e.g., people can report observations that are biased or influenced by personal views) [24] ; ii) uncertain data provenance due to the fact that human sensors tend to be correlated and may propagate rumors or falsified facts initiated by other users [25] ; iii) limited sensing availability since social sensing relies on the participatory nature of individuals (e.g., people may be less interested in certain types of public occurrences and not report them through crowdsensing platforms) [26] ; iv) privacy concerns whereby the personal information of the participants of social sensing remains at risk of falling into the wrong hands (e.g., the whereabouts of an individual may be obtained from crowdsensing apps and used by criminals to threaten them) [27] ; and v) data sparsity whereby a majority of human sensors contribute only a small number of reports, providing insufficient evidence to draw conclusions on real-world occurrences [28] .

Motivated by the plethora of benefits from both social and physical sensing and the complementary nature of the two sensing paradigms, social-physical sensing (SPS) is emerging as an integrated networked sensing paradigm that unifies the human-wisdom derived from social sensing with the empirical sensing quality of physical sensors to reconstruct the "state of the world", both physically and socially [29] - [31] . Let us consider an SPS application called social airborne sensing (SAS) [32] as shown in Figure 2 . In SAS, social media signals are used to locate events of interest (e.g., a building on fire) and drive unmanned aerial vehicles (UAVs) for validating the authenticity of the reported events using onboard sensors (e.g., cameras, thermal scanners). The validation results from the UAVs can be further used to filter out unreliable social media users. Thus SPS-based systems capitalize on the versatile sensing potentials of networked social and physical sensors by tightly integrating them and mitigating their individual drawbacks for more effective information retrieval and interpretation. In this survey paper, we explore the current literature on SPS, with an emphasis on the enabling technologies behind SPS, state-of-the-art SPS applications, recurring challenges in SPS, and opportunities for future research along this emerging domain. A few other notable application domains empowered by SPS include urban search and rescue [33] , smart healthcare [34] , simultaneous localization and mapping [35] , human mobility modeling [9] , and anomaly detection [36] . Figure 3 highlights several recent examples of representative SPS applications which encompass: i) anomalistic crowd detection with social media and surveillance camera; ii) social media-driven vehicular sensor network (S-VSN)-based plate recognition; iii) fire monitoring with UAV and crowdsensing; iv) road damage detection with satellites and social media; v) crime reporting with wireless sensor networks (WSN) and crowdsensing; and vi) contact tracing with social media and wearable sensors.

The key design philosophy of such SPS applications is to harness the collective and complementary information from social and physical sensors and draw a complete picture of real-world occurrences that otherwise might not be possible with standalone sensors. For instance, in an anomalistic crowd detection application based solely on networked surveillance cameras, the cameras might only be able to detect crowd events of interest (e.g., election campaigns, protests) and estimate their size without deducing the key attributes of the crowds such as nature and cause. On the other hand, people might post their plans for accumulating in public places across social media platforms (e.g., Twitter, Facebook) and post realtime updates of the progress of the crowds. However, the size and exact duration of the crowds might not be attainable from just the social media reports. When the complementary information from the social and physical sensing sources are merged, it can potentially be used to infer the key attributes of the crowd (e.g., duration, nature, and cause of the crowd) and tell the complete story behind the crowd gathering in the first place (e.g., for staging a public demonstration in support of a protest).

While SPS promises the groundwork for a paradigm shift in sensing and data collection, it also brings a set of new challenges to address. Examples of such challenges include: i) how to simultaneously collect relevant data from multitudes of social and physical sensors scattered around the world and relate the collected data to each other in a reliable fashion given their diverse characteristics? ii) How to efficiently handle the complex interactions between the human, cyber, and physical components in SPS when melding social sensing with physical sensing? iii) How to handle the data and device heterogeneity originating from the two distinct sensing paradigms (e.g., text data from social media vs. image data from cameras)? iv) How to characterize the dependency and correlation between the data sources when physical and social sensors are meld together? v) How to ensure end-user privacy and security considering the diverse sets of complementary information contained in the social and physical sensing mediums (e.g., geo-location data from mobile devices can be combined with information from social media posts of users to reveal sensitive information)? vi) How to adapt to the intricate dynamics that arise when jointly exploring the physical world and the social domain (e.g. how to concurrently cope with the rapidly evolving physical world events and the escalating social media reports during an emergency response)?

While the above challenges impose difficulty in developing effective SPS systems, they also set forth opportunities to instigate future research directions. We envision the potential to incorporate techniques from multiple disciplines such as networked sensing, communication systems, estimation theory, control theory, artificial intelligence (AI), distributed systems, and cryptography to address the highlighted challenges. Several current survey papers on physical sensing have investigated the functionality and features of recent physical sensing approaches (e.g., roadside surveillance systems, wildfire monitoring systems, indoor localization using wireless networks) [37] , [38] . On the same note, several survey papers on social sensing have provided comparative studies on representative social sensing schemes (e.g., fuel availability finder using crowdsensing apps, social mediadriven interesting place discovery) [39] - [41] . While a few survey papers have explored some sensing approaches that fall at the intersection of social sensing and physical sensing and are partially related to SPS [42] - [44] , they do not focus on an extensive definition and overview of the SPS paradigm itself or present a comparative study of the existing SPS applications.

Most importantly, past studies have not fully addressed the need for highlighting the key challenges prevalent in emerging SPS systems, which is necessary for designing, implementing, and evaluating emerging SPS systems and applications. This survey paper aims to reduce this knowledge gap in the existing literature and extensively explore SPS.

The rest of the paper is organized as follows. In Section II, we articulate the definition and overview of SPS. In Section III, we outline the key enabling technologies for SPS. In Sections IV, we identify the different applications propelled by SPS and discuss the corresponding state-of-the-art solutions. In Section V, we elucidate the key potential research challenges in constructing reliable and pervasive SPS. In Section VI, we highlight a set of research directions and opportunities for future work in SPS to mitigate the identified challenges. Lastly, in Section VII we manifest a reflection of our findings and conclude our survey of SPS.

In this section, we provide an explicable definition of socialphysical sensing (SPS). We first discuss the deficiency of earlier literature in defining SPS and underscore the need for a comprehensive definition of SPS. Afterward, we provide an overview of the SPS paradigm.

Before delving into framing a concrete definition of SPS, it is important to highlight why prior studies have not entirely acknowledged the need for a generalized definition of SPS. First, depending on the application context, often the lines between social and physical sensors tend to be blurred. For example, an urban air quality monitoring application that uses a crowdsourcing app and social media to take user inputs for assessing the air quality might appear to be a purely social sensing application at first glance. However, if the application utilizes the GPS and accelerometers of the users' smartphones to determine the location and position of the users or relies on images taken by the users through the crowdsensing app (e.g., pictures of the sky or surroundings), the application also involves physical sensors and hence can be categorized as SPS. Due to the fact that there are diverse ways of intertwining the plethora of social and physical sensors in applications that can be classified as SPS, there is no single widely accepted definition of SPS. Second, while SPS is a versatile sensing paradigm, it is a relatively new sensing paradigm that has not been extensively explored by existing literature. A few early survey papers have attempted to discuss sensing approaches that incorporate social and physical sensors such as cyberphysical-social systems (CPSS) [44] and cyber-social systems (CSS) [45] . However, such papers solely discuss the mapping of physical and social sensors to the cyberspace by considering the social and physical sensors as black-box information retrieval tools. Moreover, survey papers on CPSS and CSS primarily focus on the controlling or monitoring of physical processes through feedback loops, without explicitly defining SPS. We believe a concrete and comprehensive definition of SPS is crucial for the analysis, evaluation, and future development of SPS systems.

The term social sensing refers to sensing approaches that utilize human "sensors" to collect real-time data from the physical world through social data platforms such as social media, crowdsourcing apps, or a combination of the two [20] . On the other hand, physical sensing refers to sensing approaches where physical sensors capture the real-world phenomenon using hardware sensors that can be stationary (e.g., roadside traffic measurement units), mobile (e.g., UAVs, robots), or a combination of both types of sensors [46] . SPS consists of an umbrella of sensing systems that combine the complementary virtues of human observations expressed through social sensing and the accurate sensing capabilities of hardware sensors constituting physical sensing to capture the real-world phenomenon. In a holistic sense, SPS refers to any sensing scheme which leverages a combination of social sensors alongside any combination of physical sensors. An overview of the SPS paradigm is shown in Figure 4 . Figure 3 , SPS encompasses several diverse domains of applications based on the application requirements and the data acquisition tools involved. While there are no strict classification criteria for SPS schemes, the applications in SPS may be broadly classified into a few major types. The first major type of SPS involves information acquisition from reports obtained from social media platforms combined with sensing data from fixed physical sensors installed across various locations. A few examples of this form of SPS include: i) anomaly detection using surveillance cameras and social media posts [47] ; and ii) traffic accident detection based on social media and roadside traffic measurement sensors [48] . The second major variant of SPS is integrated social media and mobile physical sensor-based sensing systems where social media data are analyzed to locate probable event locations and mobile agents are dispatched to further scrutinize the event reports. An example of this type of SPS is anomaly detection with social airborne sensing (SAS) where social media signals are used to drive UAVs to locations involved with critical events such as natural disasters as illustrated in Figure 3 [32] . The third major type of SPS involves crowdsourcing integrated with mobile physical sensors. A few examples of this format of SPS are: i) environmental sensors and crowdsensing-based air-quality monitoring systems [49] ; ii) noise mapping in urban areas using mobile crowdsensing and acoustic sensor networks [50] ; iii) automatic license plate recognition (ALPR) using vehicular sensors and reports from drivers on roads as shown in Figure 3 [51] ; and iv) smart water quality monitoring based on crowdsourcing and IoTenabled water quality sensors [52] . The fourth major variant of SPS uses a combination of crowdsourcing with fixed physical sensors to sense the environment. Some examples of such type of SPS are: i) collaborative disaster damage assessment (DDA) using surveillance camera footage and crowdsourcing website such as MTurk [21] ; and ii) crime detection with heterogeneous sensor networks (e.g., cameras, microphones, proximity sensors) and crowdsensing apps [53] . While the above four categories represent the major formats of SPS applications, different variants of SPS can be integrated since there are often no absolute strict boundaries across various SPS application types. Based on the application criteria, various data collection platforms cay be combined in SPS. For example, in a search and rescue application in the aftermath of an earthquake, locations of potential victims can be collectively gathered from social media posts and crisis reporting apps. Subsequently, ground robots might be dispatched to the reported locations to validate the information from the social data platforms.

With the interplay of social and physical sensors, SPS aims to deliver a more comprehensive, ubiquitous, and effective information acquisition paradigm compared to standalone social or physical sensing paradigms. By leveraging the collective wisdom from the social and physical domains, SPS can not only sense the real-world but also help to control and actuate critical real-world processes. Examples of such control processes include mitigating traffic accidents, reducing the spread of diseases, and preventing crimes in high-risk areas. While traditional social and physical sensing systems focus on acquiring stimuli from the environment, SPS applications aim to bridge the gap between the social and physical worlds by establishing a closed-loop system connecting the human, cyber, and physical worlds. To accomplish the above objectives, SPS requires careful coordination and interaction between a set of essential enabling technologies, which are discussed in the following section.

At the forefront of SPS lies several key enabling technologies encompassing the data acquisition platforms, communication technologies and protocols, and computing paradigms. In this section, we discuss three major enabling technologies behind SPS as shown in Figure 5 .

A crucial component of the sensing process in SPS is the data collection. The key drivers for data acquisition in SPS can be classified broadly into two categories: social data platforms and physical data platforms. The details of the platforms are discussed below.

1) Social Data Platforms: Intuitively, social data platforms embody the mediums of information retrieval where human sensors are directly involved in the contribution of the raw data. Social data platforms can be further subdivided into two types. The first type of social data platform is social media sensing where any individual in the possession of a smart device (e.g., a smartphone) with Internet connectivity may voluntarily report an occurrence happening nearby through social media websites or apps [54] - [56] . Common forms of social media include social networking services such as Twitter, Facebook, Instagram, Pinterest, and Snapchat [57] . Within each such social networking channel, people develop networks and relationships with other individuals who typically share similar personality traits, mutual goals, activities, ethnicity, community, or are personally known to each other [58] . Conscious individuals tend to report or share the incidents happening around them in the real-world on social networking websites. This serves as the starting point for vital information in SPS which can be further utilized to detect the onset of critical occurrences (e.g., floods, traffic accidents, gas explosions). Another variant of social media is social news aggregation websites that gather news contributed by multiple individuals from different online sources and accumulate them into one platform [59] . News contents in these aggregation websites are typically ranked based on factors such as popularity, credibility, and urgency of the news. Examples of popular social news aggregation websites include Digg, Reddit, and Medium [60] .

The second type of social data platfors is crowdsensing which generally involves large groups of participants engaged to carry out specialized distributed sensing tasks (e.g., traffic condition reporting, crisis reporting, smart water sensing) through individual devices (e.g., smartphones, portable sensors) [61] . A few recent examples of representative crowdsensing applications include: i) interesting place locator [62] ; ii) risky traffic zone identification [63] ; and iii) urban air quality monitoring [49] . Crowdsensing can be further divided into two subcategories. One variant of crowdsensing is non-monetized crowdsensing where individuals perform small sensing tasks on a "pay it forward mentality" with the mutual incentive of obtaining information from the platform in return. For example, traffic apps, such as Waze, drivers proactively report roadside occurrences to provide real-time traffic information, in exchange for traffic updates from other users. Gas price reporting apps, such as GasBuddy, allow users to report gas station availability and prices, in return for obtaining prices at other gas stations. The other variant of crowdsensing is monetized crowdsensing where dedicated individuals perform a set of incentivized sensing tasks as paid freelancers. Compared to non-monetized crowdsensing, monetized crowdsensing typically attracts a larger number of participants and is known to generate denser data [64] . Several monetized crowdsensing platforms utilize the Internet to allocate sensing tasks between participants in different parts of the world (e.g., tasks involving urban anomaly detection in a region) [65] . A few examples of monetized crowdsensing applications include crisis reporting [66] , gas emission monitoring in urban areas [67] , and health monitoring [68] .

2) Physical Data Platforms: As the name implies, physical data platforms consist of methods that rely on physical sensors (e.g., cameras, thermal scanners) for data capture [15] . The first form of physical data platforms involves fixed sensors where a collection of dedicated sensors installed across an area or region are used to gather sensing data. Arrays of sensing devices (e.g., weather sensors, infrared sensors) are usually installed ahead of time within a region of interest (e.g., a city, building, facility, or neighborhood) which are typically parts of a large sensor network (e.g., a wireless sensor network (WSN), radio frequency identification (RFID) network) [69] , [70] . A few examples of these sensors are surveillance cameras, air quality monitors, roadside units (RSU), metal detectors, and millimeter-wave scanners [71] , [72] . In recent times, there is a surge of fixed sensors in the realm of smart home applications such as motion sensors, smoke sensors, water and gas leakage detectors, indoor positioning beacons [73] .

The second form of physical data platforms is based on mobile sensors where the sensors are not confined to a specific location and may be transported to a different position as required by the application. Mobile sensors can be further divided into several sub-classes. The first sub-class of mobile sensors are in the form of sensor-fitted autonomous unmanned ground vehicles (UGVs) and unmanned aerial vehicles (UAVs). Given the high agility and mobility of UGVs and UAVs along with the ability to carry on-board processing components, these dedicated mobile sensors are generally deployed for delay-sensitive and critical SPS applications in areas unreachable to humans (e.g., locating forest fires, monitoring flood progress, locating survivors in a wreckage site). Remote sensors are another form of mobile sensors that obtain detailed visual information of constituents on the earth's surface using optical sensors installed on aircraft or satellites in space [74] . The second sub-class of mobile sensors are smartphone-based sensing where the sensors built into smartphones (e.g., microphones, camera, GPS, accelerometer, gyroscope, ambient light sensor) are used to perceive the environment around the users of the smartphones. Smartphonebased sensing enables an opportunistic sensing approach that often does not require dedicated sensors to be installed. Thus, this technique provides a more economical and scalable sensing approach in contrast to UAVs and UGVs. For example, the vibrations picked up by a phone's accelerometer inside a car may be utilized to locate road damage, discover potholes, or detect accidents [75] . Recently, there is an emergence of wearable devices, health and fitness trackers, and smart tags used by individuals which are being increasingly used in smart city applications such as urban mobility modeling [9] .

In SPS, the confluence of the social and physical data platforms helps to collect a more extensive and comprehensive representation of the physical world. As an example of how the complementary information from social and physical data platforms in SPS can be leveraged to retrieve knowledge from the real-world, let us consider a post-disaster resource monitoring application based on social vehicular sensor networks (S-VSN). Following a disaster (e.g., hurricane, flood), it is critical to locate vital resources such as fuel and pharmacy. Often people report information about such resources on social media websites such as Twitter. However, the availability of fuel at gas stations or the chances of a pharmacy being open might change any time during the period following the disaster. Car drivers driving nearby can be dispatched to the reported locations of the vital resources where the onboard sensors of the cars (e.g., dashboard cameras) can be used to assert the availability of the resources. Thus, the mutual information exchange between the social and physical data acquisition platforms enables SPS applications to perceive and interpret real-world phenomena with greater fidelity.

The ensemble of physical and social sensors in SPS exchange data with each other and the backend servers through diverse communication technologies and associated protocols [76] . Depending on the application context (e.g., critical vs. non-critical), nature of the environment (e.g., outdoor vs. indoor), and energy profiles of the data sources (e.g., battery-powered UAVs vs. fixed surveillance cameras), various communication technologies and protocols enable the diverse devices in SPS applications to establish communication among themselves. A collection of key state-of-the-art communication technologies and protocols in SPS are discussed below.

1) Ubiquitous Local Wireless Connectivity: One variant of ubiquitous communication technology in SPS is WiFi which utilizes radio waves to transfer data among connected devices within close proximity [77] . WiFi is appropriate for indoor environments where both social data platforms (e.g., Twitter users equipped with laptops) and physical data platforms (e.g., IP cameras and smartphones) coexist. However, due to the limited range of WiFi networks, they are not an ideal fit for long distant operations, outdoor operations, and applications where the sensors move at high speeds (e.g., vehicular sensors). Bluetooth is a form of communication technology utilized to transfer data between devices over short distances (e.g., about 10 meters) using short-wavelength radio with the intent to reduce power usage [78] . Bluetooth is typically used by battery-powered sensing devices in close proximity where energy savings is a key challenge such as wearable devices like smartwatches, health and activity trackers, localization beacons, and smartphones. Figure 6 provides an overview of the state-of-the-art wireless connectivity standards enabling SPS. Figure 6 : Overview of wireless connectivity that enables SPS 2) 4/5G Cellular Network: The second widely-used communication technology in SPS is cellular networks. In the domain of cellular technologies, a wireless data transmission standard known as LTE (Long-Term Evolution) has established itself as an accepted norm for high-speed data transfer between cellular devices [79] . A newer more advanced standard in cellular communication is 5G which has superseded the older generation of LTE by delivering a plethora of additional features (e.g., greater network bandwidth, decentralized networking, reduced latency, lower congestion and interference, capacity to support larger numbers of connected devices, and better energy utilization) [80] . 5G LTE has the capability of providing network coverage to widely scattered social sensors or fast-traveling mobile physical sensors (e.g., cars, UAVs, smartphone sensors) alongside providing multi-casting and broadcasting services [76] . LTE Advanced (LTE-A) is an enhanced iteration of 5G LTE that enables bandwidth extension (i.e., up to 100 MHz), spatial multiplexing, extended coverage, higher throughput, and lower latency. These features are essential for real-time sensing tasks (e.g., offloading image/video processing tasks between UAVs) in SPS [81] .

3) Delay Tolerant Networks (DTN): While existing communication technology is expected to operate at full capacity in ideal conditions, there could be circumstances where the network becomes intermittent, sparse, or have delays due to disrupted infrastructure especially after a disaster, making it challenging for sensing devices to transmit data in an endto-end path [82] . For such scenarios, one solution is delaytolerant networking (DTN) which allocates hardware with large persistent storage (e.g., flash memory or hard drives) at the edge of the network that are resilient to extended power losses for storing incoming sensing data from nearby connected sensors [83] . When queried by processing nodes at the edge or backend servers, the stored sensing data is propagated through each connected storage node and relayed to the processing nodes [84] . DTN aims to facilitate smooth data transmission in networks that may lack continuous network connectivity. Non-time-critical SPS applications such as urban water quality monitoring and invasive species finding can benefit most from DTN. DTN is designed to operate effectively during extreme situations such as those encountered in hostile environments (e.g., during wars or natural calamities) or extreme climates (e.g., at Earth's poles or in a desert). Two variants of DTNs are vehicular DTN that uses cars [82] and UAV-based DTN that uses drones [83] .

The interconnection of the diverse sensing devices in SPS is facilitated by the evolution of diverse Internet of Things (IoT) messaging protocols [76] . IoT facilitates the coexistence of the social and physical sensors in SPS by harnessing the multiple communication technology and protocols to unify the social and physical realms. While the topic of IoT deserves an elaborate discussion of its own, it is important to realize the need for the IoT messaging protocols that drive the communication between entities within SPS applications. At one end, social sensors respond to environmental stimuli and post their real-world observations online using their owned connected devices (e.g., smartwatches, laptops, or smartphones with local wireless networks). On the other end, physical sensors perceive the environment using transducers, convert their signals into a digital format, and transmit the information through a particular communication technology (e.g., LTE, Bluetooth). A wide range of IoT standards such as Message Queue Telemetry Transport (MQTT), Constrained Application Protocol (CoAP), Data Distribution Service (DDS), and Extensible Messaging and Presence Protocol (XMPP), enable the seamless interchange of data between the social and physical data platforms [85] . 5) Sensor Networks: Traditionally sensor networks have formed the foundation for intercommunication separately within networked physical sensors (e.g., internet-connected vehicular sensors or wireless surveillance cameras) [86] . However, in recent times sensor networks have become an integral component of communication for social sensors given the ubiquity of smart devices with internet connectivity (e.g., large groups of individuals equipped with smartphones for measuring air quality through crowdsensing apps or drivers reporting traffic incidents in Waze through dashboard infotainment systems) [49] . Given these bases, sensor networks are crucial for the operation of SPS which closely interconnects diverse physical and social sensors for capturing the real-world phenomenon at large scales. A few examples of sensor network-enabled SPS applications include: i) social vehicular sensor network (S-VSN) which integrates social sensing with existing ground-based vehicular sensor network (VSN) for monitoring events of interest along roadsides [87] ; ii) flood inundation measurement using crowdsensing and wireless stream gauges to measure flood levels [88] ; and iii) social media and networked surveillance camera-driven anomalistic crowd investigation where video footage from connected surveillance cameras are meld with social media reports to infer abnormal crowd events [89] .

6) Vehicular-to-Vehicle Communication (V2V): With the growing need for inter-vehicular communication, Vehicular-to-Vehicle Communication (V2V) has gained traction in recent years [37] . V2V incorporates cars equipped with arrays of on-board sensors (e.g. dashboard cameras, proximity sensors) coupled with sufficient computation power to opportunistically identify event occurrences on the roadsides and transmit the information to other cars through a dedicated network [90] . Due to a large number of road networks and vehicles operating on the roads, it is inherently difficult to monitor the occurrences on the roads. V2Vs represent an important facet in roadside monitoring by allowing vehicular sensors to identify roadside events (e.g., road damage, objects on roads, stopped vehicles, traffic congestion, and accidents) and communicate the information to other vehicles, allowing them to make important decisions ahead of time. V2V predominantly leverages communication standards such as Dedicated short-range communications (DSRC), C-V2X, IEEE 802.11p, and ARIB STD-T109 [91] .

In SPS, the sheer volumes of input data generated by the social and physical sensors require extensive processing and computation. To cater to the ever-increasing data processing and analytics requirement, several computing paradigms coexist. In this paper, we mainly focus on two major categories: cloud computing and edge computing. The mechanisms of the two computation paradigms are discussed below.

1) Cloud Computing: A widely used computing paradigm for carrying out processing of sensor data is cloud computing [92] . Given the sheer volumes of raw data generated every instant in SPS applications by a plethora of sensors distributed across the world, it is imperative to systematically process the sensing data and interpret valuable information in a scalable and efficient manner [93] . In cloud computing, powerful computers in the form of clustered compute nodes allow for a distributed computing environment to process the incoming sensor data from a wide range of SPS applications (e.g. health monitoring wearable sensors, social media reports indicating disaster situations). The back-end servers in cloud computing provide a global service interface to all users interested in the SPS applications and its associated services. A few more recent advances in the cloud computing paradigm include cloud execution models such as: i) serverless computing, where the cloud provider allocates machine resources ondemand for sensing applications [94] ; ii) ThingSpeak, an opensource cloud framework for processing, analyzing, storing, and visualizing real-time data from IoT devices [95] ; and iii) Apache Spark, which is an open-source unified cloud-driven analytics engine for big data processing [96] .

The computing resources in cloud computing can be either dedicated and application-owned (i.e., private cloud) or commissioned to commercial cloud platforms (i.e., public cloud) [93] . Notable examples of public cloud platforms include Amazon Web Services (AWS), Google Cloud Platform (GCP), and Microsoft Azure. These cloud platforms provide the venue for end-user devices such as IoT devices and smartphones to transmit their data to and allows for the vast amount of sensor data to be processed in real-time, enabling the SPS applications to interpret knowledge extracted from the generated sensing data.

2) Edge Computing: With the surge of time-critical SPS applications (e.g., urban search and rescue missions), edge computing has protruded as an efficient computing paradigm to conduct localized data processing [97] , [98] . Edge computing is also referred to as fog computing in some literature. In contrast to cloud computing which performs data processing on dedicated distributed clusters spread across different regions or centralized servers, edge computing administers computation at the "edge" of the network, closer to the social and physical data source and the end users. In edge computing, devices at the edge of the network (e.g., UAVs, smartphones, wearable devices) are used as computational resource to carry out the processing and data analytic tasks generated in SPS applications [99] , [100] .

One key facet of edge computing is computation offloading, where an edge device can decide to offload the data processing tasks to any other device or service within the network of connected nodes. Thus, by delegating computation tasks from resource constrained devices (e.g, UAVs with limited flight times) to devices with greater resource headroom or more specialized hardware (e.g, a vehicle fitted with dedicated GPU), the computation can be executed more efficiently, enhancing the quality of experience (QoE) for end users. In recent times, a subset of edge computing known as multiaccess edge computing or mobile edge computing (MEC) has received substantial attention which extends the capabilities of cloud computing by abstracting "cloud-like" resources (e.g., virtualized storage containers, virtual machines) and bringing them closer to mobile devices at the end points of the network. MEC allows computation processes to take place in base stations, central offices, and other aggregation points on the network. While edge computing can serve as a standalone computing paradigm, MEC can enable SPS applications to incorporate a combination of computing at the edge with abstracted cloud resources to accomplish diverse computation tasks [97] . Depending on the nature of the SPS application, edge nodes might also be comprised of local servers, cloudlets, smart routers, or gateways to facilitate both computation and network requirements. Thus, the edge computing paradigm enables a distribution of decentralized computation nodes which can potentially offload the burden of data processing from backend servers to eliminate single points of failure, reduce network overhead, curb transmission latency between devices, and eventually improve response times in sensing.

In Figure 7 , we present an abstraction model comprising the three enabling technologies that enabled SPS. The bottommost layer is the data acquisition layer that is made up of the various data acquisition platforms which actively assist in capturing raw sensor data from the social and physical sensors involved in SPS. Above the data acquisition layer resides the communication layer that contains the various communication technologies and protocols which enable the process of information exchange between all the devices comprising SPS. Above the communication layer is the computation layer. This layer is further divided into two sub-layers: cloud platforms and edge platforms which collectively perform all the computation tasks in SPS. At the top-most position resides the applications layer which represents the SPS applications that holistically coordinate the data acquisition, communication and computation to interpret the real-world phenomenon. In the following subsection, we discuss a collection of representative SPS applications existent today.

In this section, we review a set of exciting real-world SPS applications. We categorically present the applications and discuss representative schemes from recent literature for each application scenarios. Figure 3 in Section I illustrates a summary of some of the applications.

A. Social sensing and IoT-driven healthcare applications 1) Contact tracing of infectious diseases using crowdsensing and smartphone sensors: In the field of epidemiology, contact tracing is a mechanism of identifying and monitoring individuals who may have come in close contact with people having any infectious disease in order to circumvent further disease spread [101] . Pinpointing and quarantining sources of an infectious disease restricts their ability to "contact" the disease, thereby minimizing community spread [102] . Recently, with the pandemic of the Coronavirus disease 2019 (COVID- 19) , there is a surge of contact tracing applications that combine the power of crowdsensing with smartphone sensors distributed around the world to study the physical footprints of users [102] - [107] . Figure 8 illustrates an overview of contact tracing applications based on crowdsensing and smartphone sensors.

Several recent studies have attempted to meld nonmonetized crowdsensing with Bluetooth and WiFi radios found in smartphones for COVID-19 contact tracing applications [102] , [106] , [107] . For example, Google and Apple launched a decentralized COVID-19 contact tracing framework called Exposure Notification System (ENS) that logs interactions with other ENS users using their smartphones' Bluetooth radio [103] and augments it with crowdsensed data provided through mobile apps [104] . MIT Media Lab further enhanced the ENS framework by developing a privacy-preserving location extrapolation mechanism with smartphone's GPS to deduce the approximate geographical location of a contacted person [105] . The scheme also allows healthy users to determine if they have "crossed paths" with any infected person [106] .

The Singaporean government launched BlueTrace, a privacy-aware open-source COVID-19 contact tracing application based on Bluetooth-based localization and voluntary crowdsensing application [107] that logs Bluetooth interactions between participating devices. When two devices "meet", they trade encrypted messages with temporary identifiers and anyone suspected for infection will be requested to share their contact history with the concerned authority. Altuwaiyan et al. proposed a contact tracing scheme with integrated WiFi and Bluetooth-based localization technology from smartphones combined with crowdsensing through a mobile app [102] . Once users are tested positive, they are presented with a questionnaire through the app to input their memory of historical contacts. A contact tracing project called A-Turf was undertaken to accurately detect "encounters" between users within close proximity (e.g. less then six feet) using user feedback reported through a crowdsensing app and acoustic signals emitted by smartphones [108] . By determining the "footprint" of infected individuals, crowdsensing and smartphone sensordriven contact tracing systems help to test, isolate, and treat potential contacts of infected people.

2) IoT-enabled health monitoring integrated with social media: An emerging healthcare application within the SPS domain is monitoring the health of people using IoT-based health surveillance devices integrated with social media. Figure 9 shows an overview of IoT-enabled health monitoring systems integrated with social media. Internet-connected wearable and handheld devices equipped with various transducers (e.g, blood-pressure monitor, pulse-oximeter) are widely used to monitor health among individuals [109] . Meanwhile, people tend to often share their health experiences on social media which include medical symptoms, cases of illnesses, and feelings of mental disorders such as anxiety and depression [110] . Relying on the matching patterns between such individual health observations on social media and the data from physical sensors, recent literature have attempted to fuse IoT-enabled health monitoring systems with personalized data shared by social media users to provide accurate and real-time health surveillance of individuals [5] , [111] . One particular branch of research have focused on developing SPS-based health monitoring schemes that obtain the spatial and temporal correlations between the reports of health conditions shared through social media and the telemetry data generated by physical health monitoring devices [110] , [112] , [113] . For example, a person expressing about having recurring headaches on social media with increased pulse rate readings on a wearable sensor could indicate potential signs of hypertension. A recent concept called "health social networks" has been proposed which combines physiological data from wearable sensors (e.g., fitness trackers, smart watches) with the moods of online users expressed through social media to determine the mental health and well-being of the online users [114] , [115] . On the other hand, several studies have introduced the concept of remote health monitoring systems which analyze a patient's physiological information from IoTbased health monitoring devices and report the the findings through social media services using privacy-preserving techniques in cases of emergencies [5] , [116] , [117] . Such approaches may potentially help relatives or caregivers to respond to critical needs of elderly patients, patients with chronic and critical conditions such as tendency of heart attacks, or patients who require constant supervision and fast medical response. Thus, integrated social media and IoTbased health monitoring approaches can be used to analyze both the physical and mental health of people and provide urgent medical support to the ones subject to alarming health conditions.

Several recent studies in SPS have focused on applications integrating satellite-based remote sensing with social media and crowdsensing for capturing a wide range of visual features of the objects residing on the earth's surface. Examples of such applications include urban land usage classification [118] , [119] , predicting the poverty in underdeveloped areas [120] , post-disaster damage assessment [121] , risky traffic location identification [122] , and flood inundation mapping [123] . Figure 10 provides an overview of integrated social sensing and satellite-based remote sensing schemes. Harnessing the mutual efforts of human sensors and physical sensors installed on satellites results in: i) a more pervasive and fine-grained representation of the objects residing on earth's surface [122] , ii) a reduction of their individual weaknesses (e.g., slow update interval of satellites, poor location accuracy of social sensing) [28] , iii) localized and real-time information for closely monitoring the environment, which is useful for applications involving emergency response, smart cities, and environmental hazards [124] , and iv) a greater spatial resolution, which is crucial for applications like land cover classification, distinguishing urban-rural regions, damage assessment, target identification, and geological mapping [119] .

The fusion of social sensing with the empirical measurements from satellite-based remote sensing has opened opportunities for various interesting SPS applications. For example, Zhang et al. developed RiskSens, a multi-view learning approach to identify locations with high traffic risk by combining social media data with satellite imagery data [122] . Chi et al. proposed Crowd4RS, a land usage and land cover classification scheme that combines satellite images in urban areas with geo-tagged social media photos for a more localized and finegrained analysis [119] . Rosse et al. designed a framework to infer flood inundation levels on different terrains by melding geo-tagged images from social media, optical satellite imagery, and high-resolution terrain mapping using a Bayesian statistical model [123] . Wang et al. presented an early warning system that fuses Twitter data with historic remote sensing data for detecting and predicting weather-driven natural disasters in near real-time [118] . A Twitter-driven remote sensing approach has been developed to convert geo-tagged tweets into high resolution raster images and integrate them with satellitebased nighttime lights to infer socioeconomic activities [120] .

Another study has presented a framework to incorporate multisourced data from social media, remote sensing, and online databases through spatial data mining and text mining for postdisaster damage assessment [121] . More recently, an integrated crowdsensing and remote sensing scheme has been proposed that combines remote sensing imagery and mobile phone positioning data for urban land usage mapping [125] . By exploiting the collective benefits of social sensing and satellitebased remote sensing, the above schemes hence facilitate in obtaining a fine-grained interpretation of earth's geological features. In recent times, social airborne sensing (SAS) is progressing as a new SPS application domain where social signals are used to drive unmanned aerial vehicles (UAVs) for perceiving anomalous occurrences in time-sensitive applications (e.g., disaster response, wildfire monitoring) [32] , [126] . Figure 2 in Section I illustrates the architecture of representative SAS schemes. SAS is motivated by the agility and empirical sensing capabilities of UAVs fitted with physical sensors (e.g., camera, LiDAR, thermal scanner) [17] and the pervasive nature of social data platforms (e.g., social media, crowdsensing). Thus, SAS attempts to leverage the collective benefits of UAVs and social data platforms to provide a more rapid response and wider sensing scope compared to other SPS approaches (e.g., approaches that use satellite imagery or fixed sensors like surveillance cameras). Specifically, a more rapid and timely data acquisition is delivered by SAS, especially in critical scenarios such as search and rescue missions, post disaster response and recovery, and tracking potential suspects around crime scenes.

An SAS system initially parses and analyzes data collected from social media and crowdsensing platforms to locate probable events of interest (e.g., a person injured on a roadside, an area getting flooded, buildings damaged by an earthquake) [16] , [127] . Afterward UAVs are selectively dispatched to the extracted locations using various resource management policies (e.g., game theory, supply chain management, reinforcement learning) to verify the authenticity of the event reports using their on-board physical sensors and augment the knowledge acquisition. One of the first attempts at constructing an SAS framework is CollabDrone which identifies latent correlations among reported event locations on social media to drive UAVs to regions of interest [16] . By incorporating spatiotemporal-aware inference models to compute the temporal, contemporaneous and lagged spatial correlations of prior and concurrent event locations, Collab-Drone attempts to achieve better efficiency by selectively dispatching only a small number of UAVs. More recently, an SAS scheme called SocialDrone has been proposed which leverages a closed-loop source selection mechanism to harness the validation results from the UAVs to dynamically filter out unreliable users on social media [32] . This approach potentially helps in improving the event detection from social media data by minimizing the chance of falsified claims. The collective benefits of UAVs and social sensing empower SAS to function as an agile and responsive information acquisition tool for acquiring knowledge from the real-world.

2) Social Vehicular Sensor Network (S-VSN): While SAS schemes offer pervasive and accurate information retrieval in critical scenarios, they require dedicated UAVs, which are not only expensive resources but also limited in number. Regardless of the efficient management of UAVs by SAS schemes, UAVs have limited flight times dues to their reliance on batteries. To address these issues, an integrated SPS paradigm called social vehicular sensor network (S-VSN) has been explored in recent times that integrates social sensing with existing ground-based vehicular sensor network (VSN) to provide a more scalable and widespread anomaly detection service [87] , [128] . Figure 11 shows the architecture of representative S-VSN schemes. VSNs have matured into a dependable networked sensing paradigm for vigilance and situational awareness along roadways that uses car equipped with physical sensors (e.g. dashboard cameras, smartphones) to opportunistically identify event occurrences (e.g., accidents on roads) [86] . Harnessing existing vehicular infrastructure does not require additional dedicated sensing equipment, which in contrast to UAVs, is more unobtrusive and reduces deployment cost and time since dedicated agents are not required. However, one limitation of traditional VSNs is that the information collected by vehicles is restricted to only those regions traversed by car drivers, limiting the scope of sensing for VSNs and their adaptability in unraveling new events. Thus, by exploiting the complementary nature and collective strengths of VSNs and social sensing, S-VSN attempts to provide widespread sensing coverage and greater sensing accuracy compared to standalone VSN.

In certain scenarios such as identifying risky traffic regions or discovering important resources in the aftermath of a disaster in large areas (e.g., locating gas stations with available gas or finding available pharmacies in a city), an S-VSN might be more feasible than an SAS. An S-VSN scheme called SocialCar has been developed which uses an incentivized social media-driven task allocation approach for VSNs in smart city applications such as anomaly detection, risky traffic location identification, and environmental monitoring [128] . SocialCar leverages a top-down incentive control mechanism to dynamically adjust the incentives for exploration of the event locations based on the historic performance of the car drivers in successfully performing sensing tasks (i.e., explore events of interest). A road damage-aware S-VSN scheme called DASC has been proposed which uses a Markov Decision Process (MDP)-based damage discovery scheme to locate roads affected by damage after a disaster such as hurricane. DASC uses the knowledge obtained by car drivers to make optimal route planning decisions [87] by avoiding the damaged roads. Thus, S-VSN augments the outreach of vehicular sensors with the ubiquity of social sensors to explore events in critical scenarios.

One recent SPS application domain is automatic license plate recognition (ALPR) based on crowdsensing (e.g., smartphone apps) and physical sensors (e.g., roadside units, vehicular sensors and smartphone sensors). Figure 12 provides an overview of ALPR applications based on SPS. In SPSdriven ALPR applications, information from traffic monitoring devices (e.g., CCTV cameras, GPS in vehicles, and IoT devices such as dashboard cameras) are meld with human inputs from crowdsensing apps (e.g., Citizen, Waze, Neighbors) to track down the plate number of a potential suspect's vehicle evading from a crime scene using AI based object detection algorithms [51] , [129] . In particular, the human observations contributed by drivers, passengers, and commuters on roads (e.g., a report of a person's suspicious behavior of speeding abruptly) might be integrated with knowledge from hardware sensors to narrow down searches by law enforcement personnel and swiftly locate the whereabouts of a perpetrator.

One important concern of SPS-based ALPR applications lies in their real-time requirements where plate detection tasks are expected to be accomplished within certain time bounds in resource-constrained environments (e.g., the devices might have limited network bandwidth). Existing standalone ALPR approaches primarily focus on analyzing large volumes of video footage data collected from surveillance cameras and stored in the cloud platforms [130] . However, such schemes often introduce a non-trivial amount of data transmission delay Figure 12 : Overview of Automatic license plate recognition using crowdsensing and physical sensors to offload the videos to the cloud, which is not favorable for the real time car plate detection application. More recently, there is a growing development of ALPR schemes that harness crowdsensing combined with existing vehicular sensors and IoT devices (e.g., vehicles equipped with dash cameras and smart devices owned by citizens) to form a city-wide video surveillance network that tracks moving vehicles using the automatic license plate recognition (ALPR) technique [131] . Zhang et al. developed EdgeBatch, an SPS-based ALPR task management framework where reports about license plates of probable suspects from concerned citizens in crowdsensing apps are combined with inputs from IoT sensors (e.g. surveillance cameras) using collaborative edge computing resources to detect the license plates [132] . Trottier et al. presented the concept of a dashboard camera and crowdsensing platformdriven ALPR scheme for smart cities where video footage from dashboard cameras are analyzed by image processing algorithms and further augmented with inputs from crowdsensing participants through an app to recognize the number plates [133] .

Despite their usefulness, ALPR approaches also instill the privacy concern in the collaborative sensing context of SPS applications. For example, the car drivers might not be willing to share the metadata from their devices to the cloud in fear that such data may reveal their private information (e.g., location, speed, and driving behavior). With concerns of user privacy, Alcaide et al. proposed a privacy-aware ALPR scheme that maintains confidentiality of the users data and prevents unauthorized usage of private devices that are used for capturing and recognizing images of plate numbers [134] . A privacy-aware ALPR scheme has been proposed that masks and protects the identity of the owners of license plates recognized using data from crowdsensing apps and roadside monitors [135] . By exploiting the knowledge from crowdsensing and physical sensors, SPS-based ALPR applications aid in tracking down potential criminals on roads [51] .

The prevalence of IoT alongside social media and crowdsensing has opened new domains for situational awareness in SPS. A few examples of such applications include real-time crowd density measurement, search and rescue operations, and urban anomaly detection [136] - [139] . By integrating social media and crowdsensing with the IoT paradigm, the emerging areas of SocialIoT and CrowdIoT respectively are achieving results beyond what is possible with traditional standalone situational awareness approaches. Figure 13 exemplifies a few situational awareness applications based on SocialIoT and CrowdIoT. In the following subsections, we discuss a few variants of SocialIoT and CrowdIoT. 

In recent times, there is surge in SPS applications that focus on indoor localization based on contextual information provided on social media and raw signals from IoT devices. While GPS provides fairly accurate outdoor location tracking, the applicability of GPS for indoor tracking is limited primarily due to inaccessibility of satellite signals inside confined spaces and lower degrees of precision. As such, accurate indoor localization schemes require either additional infrastructure support (e.g., ranging devices) or extensive training before system deployment (e.g., WiFi signal fingerprinting). In indoor localization, networks of IoT devices are used to track people or objects in confined placed where GPS and other satellite technologies usually lack precision or fail entirely, such as inside multistory buildings, airports, alleys, parking garages, and underground locations. Location-based services, such as targeted advertisement, geosocial networking and emergency services, are becoming increasingly popular for mobile SPS applications [140] , [141] .

In order to help existing localization systems to overcome their limitations or to further improve their accuracy, approaches have been developed that combines social media sensing with IoT for accurate location tracking indoors. For example, a scheme called Social-Loc has been proposed that integrates the physical traces of an individual posted through social media (e.g., check-ins to a particular shop in a shopping mall) with RSSI signals from WiFI routers to potentially derive the exact location of individual users within a building [140] . Chu et al. designed SBOT, a social media and sensor networkdriven indoor localization scheme which combines user statuses and updates posted though social media (using text mining techniques) with telemetry data from smartphone sensors (e.g., altitude, speed, and heading of the users) to pinpoint the location of the users [142] . Liu et al. proposed a socialdriven IoT system consisting of backpacks equipped with 2D laser scanners and inertial measurement units augmented with historical social network traces of the users to perform indoor localization and visualization of complex environments such as a staircases or corridors [141] . However, with the increasing facilities for geo-locating people using their digital footprints, the concerns for the individuals' privacy also prevail. As we will discuss later in Section V-E, the metadata obtained from the social and physical sensors in SPS for locating people exposes risks for revealing their private information. A few privacy preserving SocialIoT schemes have been developed which aim to protect identities of people while geo-locating them indoors [112] , [143] .

2) CrowdIoT-based Context Awareness: Several exciting stationary CrowdIoT applications have emerged that are crucial to the well-being of the society which include criminal identification and disaster response [144] . Dunphy et al.

proposed an integrated crowdsensing and CCTV-based video surveillance framework where surveillance footage collected from CCTVs spread across a city are assigned to Amazon MTurk participants to tag instances of abnormal occurrences in real-time (e.g., traffic accidents, crimes) [144] . Abu et al. designed an integrated risk assessment framework using crowdsensing apps and fixed urban IoT sensors (e.g., proximity sensors, acoustic sensors, radars, air quality monitors, etc.) that predicts the possibility of crisis such as multivehicle accidents, major weather events, and large fires [145] . Vital information from the framework might potentially assist emergency personnel such as firefighters and first responders. Beyond surveillance-centric context awareness applications, stationary CrowdIoT-based SPS schemes are also used for locating regions of adverse weather and climatic conditions. For example, Horita et al. developed a flood inundation mapping (FIM) system that integrates crowdsourcing data with data from in situ weather radars to infer probable locations with flood [146] . Thus, building upon the tight integration of crowdsensing and fixed-sensor IoT devices, stationary Crow-dIoT solutions (like the ones discussed above) facilitate in providing rich context-aware SPS applications.

Another emerging context awareness sub-domain within SPS involves the integration of crowdsensing with mobile devices and portable IoT devices, otherwise known as mobile crowdsensing (MCS). Applications integrating mobile sensors with crowdsensing in MCS utilize users with mobile devices capable of data capturing, computation, and communication to collectively share data and extract information to measure, assess, estimate or predict processes of shared interest [148]- [107] Exchange encrypted messages between participating devices and query suspected individuals through an app to input their contact history. [102] Monitor whereabouts of infected individuals with WiFi and Bluetooth-based indoor localization and present questionnaire through app to input their memory of historical contacts. [108] Analyze acoustic signals from cellular devices to measure social distance and integrate with user feedback from app to detect infected individuals.

[122] Identify locations with high traffic risk by multi-view learning from social media and satellite imagery data [119] Classify land usage and land cover by melding satellite images in urban areas with localized geo-tagged social media photos. [123] Infer flood inundation levels on different terrains by applying Bayesian statistical model on geo-tagged images from social media, optical satellite imagery, and highresolution terrain maps.

Integrated social sensing and satellite-based remote sensing Social Media + Mobile Physical Sensors [118] Detect and predict weather-driven natural disasters in near real-time by fusing Twitter data with historic remote sensing data. [120] Infer socioeconomic activities by convert geo-tagged tweets into high resolution raster images and integrating them with satellite-based nighttime lights. [121] Incorporate multi-sourced data from social media, remote sensing, and online databases through spatial data mining and text mining for post-disaster damage assessment [125] Combines remote sensing imagery and mobile phone positioning data for urban land usage mapping.

[16] Identify latent correlations among reported event locations on social media to drive UAVs to regions of interest.

Anomaly detection using SAS and S-VSN Social Media + Mobile Physical Sensors [32] Leverage closed-loop source selection to harness the validation results from social media-driven UAVs for filtering out unreliable social media users. [128] Allocate incentivized sensing tasks to car drivers based on social media reports in smart city environments. [87] Locate roads affected by damage after a disaster such as hurricane and route cars avoiding damaged roads for performing sensing tasks based on social media reports. [132] Combine reports about license plates of probable suspects from concerned citizens in crowdsensing apps with inputs from IoT sensors (e.g. surveillance cameras) to detect the license plates. License plate recognition using crowdsensing and physical sensors Crowdsensing + Fixed & Mobile Physical Sensors [133] Perform image processing on dashboard camera footage and combine with crowdsensed feedback to recognize the number plates. [134] Obtain privacy-preserved anonymous inputs from crowdsensing participants and integrate with image processing techniques to locate suspects' number plates. [135] Mask and protect the identity of the owners of license plates recognized using data from crowdsensing apps and roadside monitors. [140] Integrate physical traces of an individual posted through social media with RSSI signals from WiFI routers to derive their location inside a building. [142] Combines user statuses and updates posted though social media using text mining techniques with telemetry data from smartphone sensors to pinpoint the location of users. [141] Perform indoor localization and visualization of complex environments such as a staircases or corridors by using backpacks equipped with 2D laser scanners and inertial measurement units augmented with historical social network traces of users. Situational awareness using social media and crowdsensing melded with IoT (Social/CrowdIoT) [112] Geo-locate users indoors using privacy preserving approach to protect their identities. [144] Process frames from CCTV surveillance footage using AI and combine with perception from Amazon MTurk participants to tag instances of abnormal occurrences in real-time. [145] Predict the possibility of crisis in smart cities using crowdsensing apps and fixed urban IoT sensors (e.g., proximity sensors, acoustic sensors). [146] Infer probable locations with flood by integrating crowdsourcing data with data from in situ weather radars. [147] Provide rapid disaster response by using vital metrics derived from both crowdsensing apps and portable devices equipped with RFID technology. [151] . Examples of such mobile devices include smartphones, wearables, and tablet computers equipped with both hardware sensors (e.g., GPS, microphones, heart rate monitors) and sufficiently powerful processing units (e.g., CPU, FPGA, GPU). The ubiquity of such "all-in-one" data acquisition, computation, and communication devices has motivated a good amount of work in developing a wide range of SPS-based urban sensing tools [152] - [154] . A few important applications fueled by mobile crowdsensing include: i) real-time urban crisis reporting where inputs from concerned citizens through smartphone apps and signals from IoT sensors (e.g., proximity sensors) are correlated to located urban crisis [66] ; ii) risky traffic zone identification where crowdsensed traffic data from dedicated websites are combined with roadside sensor units to locate traffic risks [63] ; iii) gas leakage detection in urban areas in which gas sensors are used to measure unusual gas concentrations and further integrated with knowledge from citizens acquired though crowdsensing apps to identify gas leakage [155] ; and iv) simultaneous localization and mapping for rescue missions in which reports of potential survivors from smartphone apps are augmented with received signal strength indicator (RSSI) values from WiFi routers to locate potential survivors in the aftermath of disasters [136] .

In addition to the above critical mobile crowdsensing schemes, there has been significant works on utilizing smartphones and wearable sensors (e.g., sociometric badges, smart glass, fitness trackers, and smart watches) for less critical applications such as: i) monitoring environmental conditions like noise [152] and air quality [156] ; ii) assessing infrastructural conditions such as traffic congestion [153] and road damage [154] ; and iii) determining most fuel efficient travel routes [157] . The integration of crowdsensing and mobile sensors has also opened up new possibilities for exciting applications in the domain of disaster response. A crowdsensing and mobile-IoT integration model has been proposed by Han et al. [147] that aims to improve disaster response by using important metrics such as weather conditions, damage reports, and infrastructure accessibility derived from using both crowdsensing apps and portable devices equipped with RFID technology. Driven by the unification of crowdsensing with sensors contained in mobile devices, mobile crowdsensing schemes, hence, aim to provide a more holistic representation of the environment in SPS applications. Table I provides a comprehensive summary of representative SPS applications. In the following section, we discuss a set of key research challenges prevalent in current SPS applications.

In this section, we highlight a few fundamental open challenges that lie in the interaction and integration between social and physical sensing in SPS.

Before valuable knowledge can be interpreted in SPS, the relevant data first needs to be located, extracted, and organized. Thus, one of the key challenges in SPS lies in simultaneously harvesting the raw sensor data from myriads of social and physical sensors [8] , [158] , [159] .

The first obstacle in data collection is to systematically locate useful data from the inherently noisy social and physical signals. In knowledge discovery from social data platforms (e.g., social media websites), a traditional search technique is to use keywords to query for the related data [8] . However, such searches might return a considerable amount of reports of unrelated incidents (i.e., noisy data) alongside the relevant ones. On the other hand, hardware sensors are susceptible to several types of characteristic noise that cause deviation in the data capture (e.g., satellites images might have low resolutions, drift in GPS data might cause incorrect location information) [160] , [161] . When combined together in an SPS setting, the noises originating from the social and physical sensors develop a degree of interdependence among each other, causing difficulty in collecting useful data. For example, let us consider a structural damage assessment application where damage locations reported by users through crowdsourcing apps are leveraged to dispatch ground robots and afterward the observations collected by the robots' cameras are used to assert the credibility of the reports and trustworthiness of the users. Unreliable human sensors might incorrectly report the sites of damage, sending the robots to the wrong locations and conversely, if the images captured by the robots are obscure (e.g., due to low light, faraway position, or motions), they might not reveal the true state of the damage and the reliability of the users might not be validated correctly (i.e., reliable users might be mistrusted or unreliable users might be incorrectly trusted). Figure 14 shows an example of such an application where events A and B are true reports of damage sites but event C is a false report (i.e., the person captures a photo to make it appear as though the building is on fire which in reality is not). The robot at event C captures a poor resolution image due to its location and image sensor limitation which yields misclassification of event C as true, thereby placing more trust on the unreliable source.

Existing literature on social sensing has proposed methods to overcome the noise from social data platforms with techniques such as machine learning (ML) [162] , artificial neural networks (ANNs) [163] , estimation theory [20] , and adaptive sampling [164] . Studies on physical sensors have proposed methods to reduce sensor noise using approaches like image enhancement with super-resolution [165] , deep learning driven noise reduction [166] , and graph neural network-based data extrapolation [167] . However, such standalone approaches fail to address the intrinsic interdependence between the noise from social and physical signals in SPS, which is non-trivial to quantify and model.

The second obstacle is gaining access to the sensing data from devices owned by individuals. While there is an abundance of connected devices that are able to perform a wide range of data capture, computation, and communication tasks, a significant number of them are privately-owned (e.g., smartphones, IoT devices, surveillance cameras) [168] . Consequently, gaining access to such sensors' data is difficult primarily because the individual entities might not be willing to share their personal devices due to reasons such as inconvenience, draining of battery on mobile devices, usage of cellular data, and privacy concerns [169] .

Recent literature has presented several solutions like: i) privacy-aware schemes such as game-theoretic task allocation [170] and non-invasive distributed private data collection [171] ; ii) energy-preserving data transmission schemes such as Bluetooth low energy [172] and ultra-low power sensor networks [173] ; and iii) bandwidth-conserving data sharing tools such as signal compression [168] and hop-byhop flow control [174] . These approaches might potentially help to convince people to provide access to their devices for obtaining sensor data. However, beyond the willingness of people to share their personal devices, the devices in SPS themselves might be unavailable for capturing data or providing access to the data. For example, an user might be using her smartphone to play video games or watch videos, making the device unavailable for capturing images and processing them efficiently. Therefore, collecting data from the social and physical realms that direct to the appropriate information remains an open challenge in SPS.

In SPS, one elemental challenge is handling the complex interactions between the human, cyber, and physical (HCP) components when integrating social sensing with physical sensing. As events in the real world play out, human and physical sensors are expected to spontaneously contribute knowledge through the social and physical data platforms to recover the truthful states of the real-world occurrences. Given this basis, developing a closed-loop system that seamlessly integrates the social and physical sensing paradigm is crucial.

In such a closed-loop system, the social and physical sensors effectively communicate and complement each other to jointly accomplish the assigned sensing tasks. Existing research on human computer interactions (HCI) have explored the need for designing effective interfaces to connect the human and cyber worlds which include examples such as web interfaces, mobile applications, online forms and survey questionnaires, virtual reality (VR), and motion capture [175] . In recent times, there is a surge in research on cyber-physical systems (CPS), which explores the interactions between the cyber and physical worlds with a focus on the problems in sensing, computation, and control of a CPS system [43] . Recent studies in CPS have proposed techniques such as embedding human intelligence into the cyber space and augmented reality-driven assistive technology for humans to reduce the gap between the human and cyber worlds [176] , [177] . However, handling the HCP interactions in SPS is much more complex and challenging than the problems studied by existing HCI and CPS research.

While human participants typically act as sensors in SPS applications, it is imperative to also carefully consider their roles as actuators. For example, let us consider a smart water monitoring application where human participants periodically report the quality of water in their neighborhood through a crowdsensing app which is combined with measurements from physical water quality sensors placed throughout a city [178] . Figure 15 shows a scenario of the role of humans as actuators in SPS applications. If the humans participants do not contribute data of sufficient quality (i.e., not enough reliable data or low participation level), incentives can be applied to encourage them to provide better quality data. The incentives serve as control signals and the human participants act as actuators. Upon receiving higher incentives, the humans might potentially take a response/action in the physical world by: i) collaborating to contribute more data; ii) validating the data of their peers; iii) or encouraging more people to participate by referring them to use the app [179] . Thus, the incentive serves a signal from the cyber world (i.e., through smartphone apps) to control response in the human world (i.e., the human participants) and when the humans receive the incentives, they respond in the physical world (i.e., collect and contribute higher quality data). Such an adaptive closed-loop system requires careful design that systematically models the complex HCP interactions. Figure 15 : Example of human's role as actuators in SPS Current literature has proposed methods to develop closedloop systems encompassing various sensors (e.g., cameras, GPS sensors), actuators (e.g., robotic arms, motorized doors), and controllers (e.g., proportional-integral-derivative (PID) controller, fuzzy logic controllers, reinforcement learning) for establishing effective cooperation between them using techniques such as linear quadratic Gaussian (LQG) control [180] , supply chain theory [181] , and blockchain-based smart contracts [182] . However, the closed-loop challenge at the intersection of human, cyber, and physical spaces in SPS has not been fully addressed by existing research due to several reasons. First, current solutions often do not explicitly model the human participants as actuators, which is a crucial feature of SPS applications. Second, current literature on incentive design in crowdsensing frameworks has not addressed the concern of how to use the physical sensors to validate the information contributed by human sensors. Third, existing approaches have not fully explored measures to leverage the social signals to effectively control the performance of the physical sensors. Last, current solutions have not explicitly considered the joint dynamic nature of the human, cyber, and physical worlds to tightly coordinate their interactions. As such, it remains an open challenge to address the HCP interaction prevalent in SPS systems.

While the abundance of physical and social sensors in SPS provides a rich influx of knowledge across various sensing applications, an inherent challenge in SPS lies in managing the diverse range of devices involved in the sensing process along with the different types of data they generate. We deem this challenge as device and data heterogeneity. Figure 16 demonstrates a scenario of the data and device heterogeneity challenge prevalent in SPS Applications. As identified in Section III, SPS applications are centered around a diverse collection of devices that encompasses data acquisition, communication, and computation. In particular, the physical sensing applications rely on the capabilities of hardware sensing devices (e.g., cameras, UAVs, and robots), while the social sensing applications obtain observations from human sensors through crowdsensing and social media by implicitly leveraging user devices (e.g., connected tablets, laptops, and smartphones). Such devices have distinct characteristics in terms of sensing and computation capabilities, sensitivity, power requirements, frequency of data capture, communication protocols, access control and authentication methods, and runtime environments [25] , [181] , [183] - [185] , which often presents a unique difficulty in managing them in SPS applications. For example, in the SAS application of Figure 2 in Section I [16] , smartphones capture human observations and send them to social media platforms which are then used to dispatch UAVs to recover the veracity of the reports. Standalone social or physical sensing applications are unlikely to have such diverse devices working together. As such, little work has been done in earlier research to bridge the knowledge gap in SPS and construct a unified framework that can efficiently manage such diverse devices.

A few efforts have attempted to mitigate device heterogeneity in sensor networks and distributed systems primarily using abstraction-based approaches such as: i) sensor emulation [186] , device clustering [187] , and sensor abstraction layer [188] for data acquisition devices; ii) containerization [189] and dynamic binary translation [190] for computation devices; and iii) software defined networking [191] and sensor network virtualization [192] for communication devices. However, in the context of SPS, existing solutions are inadequate in addressing the device heterogeneity challenge due to several reasons: i) the devices in SPS ares mostly privately-owned (i.e., smartphones, IoT devices) which makes it hard for an SPS application to apply global policies and control the devices from a central authority perspective [181] (e.g., it might not be possible to install a middleware application on a personal device); ii) the extent of heterogeneity of the devices in SPS is more evident due to the added heterogeneity of tasks and architectures which current solutions overlook [181] ; and iii) the devices in SPS often have complex interdependence of the tasks [193] , which existing solutions might not preserve [194] .

Beyond the diversity of the devices, the social and physical sensors in SPS typically generate data that widely vary across modality and formats. For example, the type of input data can range across text, image, location, audio, and video [195] and each type can further encompass different dimensionality which make the data heterogeneity even more pronounced [196] . For example, for image data the dimensionality can be edges, corners, blobs and ridges while for text data the dimensionality can be document frequency and n-grams [197] . Existing methods for mitigating data heterogeneity includes data fusion schemes such as bagging and boosting [198] , deep learning (DL)-driven data fusion [199] , covariance intersection [200] as well as other statistical and machine learning methods such as dimensionality reduction [201] , multi-view learning [122] , [202] , and feature concatenation [199] . Despite their effectiveness in standalone sensing applications, current approaches fall short in addressing the data heterogeneity issue in SPS due to the inherent complexity injected by the different rates of data generation by the social and physical sensors in SPS applications. The social and physical sensors in SPS are known to produce data at different frequencies, which renders existing solutions infeasible [203] , [204] . Consequently, versatile data management schemes need to be developed which can withstand the heterogeneity of data in SPS and interpret knowledge from the social and physical signals.

One fundamental challenge in SPS lies in characterizing the dependencies between the social and physical data sources and correlating the collected data across the two sensing paradigms [8] . While this challenge has been studied in social and physical sensing independently, it is more pronounced in the context of SPS applications and more challenging to solve due to several hurdles.

The first hurdle is how to build a unified analytical framework to model the source dependency and data provenance in SPS, given the diversity of source dependencies in social and physical sensing. For example, human sensors tend to be naturally correlated through social networks (e.g., Twitter followers tend to re-tweet their friends' tweets). In contrast, physical sensors do not typically inherit any such social correlations and are more likely to be correlated through the underlying physical phenomena or geographic locations (e.g., two air quality monitors are likely to report similar measurements if they are in close proximity). Such disparity in source dependency and data correlation make it non-trivial to seamlessly integrate the diverse social and physical sensor measurements under a principled framework [205] . Current knowledge discovery and data mining approaches in social and physical sensing such as semantic pattern recognition [206] , trust and influence modeling [207] , covariance intersection [208] model the dependencies across social and physical sources independently. However, due to the distinct source dependencies within social sensors and physical sensors, such approaches are largely inadequate for SPS applications. A unified source dependency modeling framework to meld the social and physical sensors in SPS is yet to be developed.

The second hurdle is imposed by the presence of strong causal relationships between the physical and social sensor data in SPS applications [209] . For instance, during a traffic accident as illustrated in Figure 17 , people might report the accident along with its location and timestamps through crowdsensing apps (e.g., Waze), while at the same time traffic flow monitoring units placed at a different segment on the same road might detect unusually slow traffic movements indicating a traffic congestion [209] . While the traffic accident and congestion reported by different sensing platforms might be seemingly unrelated at first glance, aligning the temporal and spatial information from the input signals (e.g., geolocation information and timestamps of the events) might reveal that there is an inherent causality between them (i.e., the traffic congestion was probably caused due to the traffic accident) [210] . Thus, even though there might not be any direct relation between the reported events across social and physical data platforms, the sensors across the two paradigms might be reporting about the same chain of occurrences or the same context but in different formats. While these context information might help to explain the cause of anomalous incidents, it is a non-trivial task to explore such causality across the social and physical data platforms.

Given the diverse source dependency profiles and the potential presence of causality across the physical and social sensors in SPS, it is challenging to design a holistic framework that can effectively connect the disparate social and physical sensors for interpreting real-world event occurrences. Consequently, extensively exploration and modeling of the dependency and correlation within the social and physical domains remains an outstanding challenge in SPS research.

Due to the integrated nature of the social and physical sensors in SPS, one critical challenge in SPS applications is Figure 17 : Example of causality among sensors in SPS to efficiently address the privacy issues of end users of SPS applications [211] . Existing literature has proposed several privacy-aware sensing approaches for social sensing which includes source identity obfuscation [212] , blind signatures and data shuffling [211] , ring signatures [213] , and data perturbation [214] . In a similar fashion, for alleviating privacy issues in physical sensing, current approaches have developed schemes such as slice-mixed aggregation [215] , isolated virtual networks [76] , trace-free location tracking [212] , and routing with random walk [215] . Despite the effectiveness of the above approaches in preserving user privacy in social and physical sensing separately, several unique difficulties in SPS restrict their usefulness in solving the privacy challenge in SPS systems.

First, social and physical sensors in a connected environment often deliver complementary information that can be exploited for exposing the personal information of the users. Figure 18 illustrates a scenario with privacy challenge in SPS. For example, in a fitness tracker application using social media and wearable sensors, reports of daily exercise activities posted by people through social media (e.g., jogging in a park) might be correlated with user-shared historical health data from wearable sensors (e.g., blood pressure, pulse rate, body temperature) to potentially infer the medical history of an individual (e.g., whether a person has a chronic illness). Conversely, in SPS applications the data from physical sensors might also be exploited to maliciously extract the private information of individuals when augmented with social signals. For example, in an anomalous crowd detection application that combines images captured by surveillance cameras with reports of crowd gatherings posted on social media to infer the onset of sudden crowds, the surveillance cameras can only capture the image of a person at a specific location without further details of that person. However, if that particular person periodically shares their shopping history alongside geo-location information on social media, the image data from the cameras might be correlated with the additional data to unravel the socioeconomic status of the individual [216] .

Second, due to the inherent heterogeneity of the devices and data in SPS, it is a non-trivial task to apply unified privacy preserving policies in SPS applications. As discussed in the device and data heterogeneity challenge, SPS applications involve diverse sets of devices. With such a wide range of devices, it is difficult to keep track of the data transmission and security protocols of all the devices. As such, device vulnerabilities such as unprotected APIs, outdated firmware, or defunct authentication mechanisms [217] might be exploited by hackers to steal personal data from user devices. Moreover, since social and physical sensors in SPS generate a wide variety of data (e.g., text, image, video, audio, location data), the capture, transmission, and processing of the data require different energy profiles which often leaves the devices in SPS vulnerable to exploits such as side-channel attack, an attack intended to steal user data [218] . For example, when a device is processing video frames, the patterns of power usage within the device might be analyzed by an attacker to recover the raw video data [219] . Current privacy-preserving approaches are not designed to withstand the intrinsic and pronounced data and device heterogeneity prevalent in SPS applications, which might lead to vulnerability of user privacy. Thus, it is yet to be determined how to design unified privacy-aware SPS platforms that can concurrently consider the data and device heterogeneity and protect sensitive user information to address the privacy challenge in SPS. 

A pivotal challenge in SPS is handling the interrelated dynamics induced by the fusion of the social and physical domains. SPS applications innately rely upon the tight integration between social and physical realms, both of which are dynamic in nature and exert impact over one another.

The dynamics arising from the social domain tend to directly influence the performance of the physical sensing. For example, in a crowd monitoring application using SAS as illustrated in Figure 19 , if social events related to public protests are initiated and organized on social media, dynamics on the social domain (e.g., more people tweeting, different locations being targeted, people publicizing the activities to a greater level) might cause dynamics in the physical world (e.g., more new activities related to protests such as speeches and concerts, more people joining, events taking place in locations far away from one another). Given the fact that mobile physical sensors such as UAVs and robots often suffer from constraints such as energy, communication, and speed, such physical sensors might not be able to explore or investigate all the events reported by social sensors within set deadlines. Such a scenario is also illustrated in Figure 19 where the initiated crowd events are located at various locations with different deadlines. Due to the presence of the social domain dynamics, the UAVs with their physical constraints might not be able to sense all the crowd events before their deadlines. Thus, careful choices need to be made on which subset of reports from the social data platforms to prioritize for the physical sensors, which has not been addressed by existing solutions. Figure 19 : Example of how social domain dynamics affect physical sensors in SPS On the other hand, the dynamics from the physical world might affect the performance of both the social sensors and the physical sensors [176] . For example, let us consider a socialmedia-driven vehicular sensing application in the aftermath of a disaster [128] as shown in Figure 20 . The disaster might cause road-damage around the affected area. Such damage might potentially restrict the travel of cars across certain roads, which might cause car drivers to be unable to locate and report events of interest on social media (e.g., gas station availability in the vicinity of a road). Moreover, a disaster might also cause damage to network infrastructure, leading the car drivers to lose access to network connectivity and unable to post any event reports [220] . Eventually, less number of observations might be reported by the car drivers across the social media, yielding poor coverage from the human sensors. In the physical world, the event occurrences might also be accompanied with unforeseeable circumstances such as unfavorable weathers (e.g., extreme temperatures) or damaged infrastructure (e.g., disconnected power lines), which might impede the physical sensing. For example, strong wind or cloud might impact the readings from different sensors such as cameras or gyroscopes on UAVs and bumpy roads might negate the performance of vehicular sensors on cars (e.g., shaky image captured by dashcams) [22] , [221] . Therefore, careful consideration must be given to adapt the SPS systems to accommodate such physical world dynamics on-the-fly, which has not been extensively explored by current literature. Figure 21 provides an overview of the fundamental challenges in SPS. Table II summarizes the schemes targeting the challenges as well as the open research questions to address the challenges. In the subsequent section, we explore possible directions for future research to potentially address the above challenges in SPS. In this section, we present several exciting avenues for future work along the domain of SPS. As we outline each avenue, we enlist a few potential directions of research to pursue.

Since SPS applications often rely on the noisy social and physical signals contributed by a diversified set of human and physical sensors, one potential direction for future work lies in quantifying the uncertainty generated by the diverse sensors in SPS applications. As discussed in the data collection challenge in Section V, the intrinsic interdependence between the noise generated by the social and physical sensors in SPS is hard to quantify and model. As such, a degree of uncertainty is induced in the collected social and physical signals. Without carefully determining the level of uncertainty in the input data, the performance of SPS applications might be unpredictable [10] , [225] , [226] . Current social sensing analytics tools such as truth discovery algorithms primarily focus on deducing the data veracity or source reliability from the social sensing data [24] , [227] , [228] . In a similar fashion, current physical sensor data processing schemes have focused on inferring the information contained in the physical signals using techniques such as fuzzy logic [229] , autoregressive models [230] , arbitrary polynomial chaos [231] , perturbation theory [232] , and full factorial numerical integration [233] . However, the existing schemes do not explicitly focus on quantifying the interconnected uncertainty between the social and physical sensors, which is important to ensure stable performance of SPS systems. As illustrated through Figure 14 in Section V-A with the example of a structural damage assessment application, unreliable online users might incorrectly report the sites of damage, which might dispatch robots to the wrong locations. Likewise, if the images captured by the robots are obscure, they might not correctly validate the reliability of the users. Thus, the characteristic noise originating from social and physical sensors might adversely affect the sensing performance of the SPS applications. As such, it is imperative to develop methods for quantifying the uncertainty signals in SPS applications.

It is important to first realize why existing literature has not extensively explored the domain of uncertainty quantification in SPS. There are several disparities between social and physical sensors that lead to the difficulty in rigorously quantifying the uncertainty of their signals. First, the social and physical sensors in SPS generate dissimilar types of data (e.g., social sensors typically generate text data while physical sensors usually generate numeric signal data) [1] , [20] . Second, the dependencies between the sensors in social data platforms are different from that within the sensors in physical data platforms [8] . Third, the dynamics in the social domain are characteristically contrasting from the dynamics in the physical world [43] , [176] . Fourth, the rates of data generation by social and sensors are different from physical sensors (e.g., the speed at which UAVs capture images is different from the frequency at which people report incidents on Twitter) [203] , [204] . In addition, factors such as biased opinions from human sensors in social sensing and the failure cases of physical sensors (e.g., out of battery, affected by bad weathers) implicitly aggravates the uncertainty quantification in SPS [20] , [234] .

One direction for further research in SPS is to focus on rigorously quantifying the uncertainty of social and physical signals and leverage the quantification results to jointly improve the social and physical sensing component of the system. For example, in an anomaly detection with SAS application, if the uncertainty from the social signals can be determined, it may help to better dispatch the UAVs. Similarly, if the uncertainty in the captured UAV data can be measured, it can be used as feedback signals to potentially improve reliable source selection in the social sensing. Another probable research direction in SPS can be to design schemes that can deduce the uncertainty in the social and physical sensing data while simultaneously considering the SPS challenges such as the data heterogeneity, the diverse source dependencies, the social and physical world dynamics, and the contrasting social and physical sensor data generation rates. Existing studies on statistical analysis have proposed principled approaches based on estimation theory. Examples of uncertainty quantification approaches include maximum likelihood estimation (MLE), Cramer-Rao lower bounds (CRLB) [235] - [239] . Alongside quantifying the uncertainty of estimation results, future SPS schemes can focus on incorporating the accompanying factors (e.g., human bias, physical constraints) in the uncertainty propagation models. We envision that techniques from multiple disciplines might be applicable for alleviating the above hurdles and modeling the uncertainty in SPS applications which includes Bayesian networks [240] , Monte Carlo methods [241] , evidence theory [242] , Markov Chain formulation [243] , mixed integer linear programming (MILP) [244] , and polynomial chaos expansion [245] . 

As identified in Section V-E, mitigating privacy issues is a critical challenge in SPS applications. However, in attempts to ensure user privacy, often current SPS schemes have to compromise the sensing quality. For example, metadata such as geo-location information might be concealed from privatelyowned devices to protect the identity of human sensors on social media. However, the location information might be critical for mobile sensors such as robots to be dispatched to events of interest. Figure 22 shows an example scenario where concealing private data affects sensing quality. Thus, SPS applications often require the knowledge of supporting information such as locations, timestamps, and contextual information from reported social sensing data, which is often conflicting with privacy requirements of the end users. Since ensuring user-level privacy and maximizing sensing quality often turn out to be two potentially conflicting objectives in SPS [246] , it is imperative to design schemes that carefully strike trade-offs between privacy and sensing quality for an optimized SPS system.

Existing approaches in data-driven social and physical sensing schemes have proposed techniques to manage user privacy by obfuscating identifying information such as geolocation tags from the raw data from personal devices (e.g., laptops, smartphones) [247] , [248] . However, existing privacy preserving schemes have not addressed the effective handling of the trade-off between privacy and sensing quality in SPS. There are several reasons why it is difficult to simultaneously establish privacy and sensing quality in SPS. First, the unpredictable nature of the human users in SPS applications makes it difficult to ensure that the users will strictly abide Figure 22 : Example of trading-off between privacy and sensing quality in SPS by policies to protect their privacy. Second, given the unique data and device heterogeneity in SPS applications, designing a unified framework to enforce individual privacy policies across all the devices is a challenging task [249] . Third, regardless of the robustness of privacy-preserving schemes, the complementary aspects of the contributed data through social and physical data platforms can be exploited to steal sensitive user information [213] .

One future research direction to pursue for optimizing the privacy and sensing quality in SPS applications is to design multi-faceted cryptographic techniques such as blockchain technology [250] , smart contracts [251] , and ring signatures [213] . While existing cryptographic approaches have come a long way in balancing privacy and sensing quality individually in social sensing and physical sensing [252] , it is difficult to apply unified cryptography-based solutions in an SPS setting where a diverse range of devices are associ- Locating raw sensor data from numerous social and physical sensors [20] , [162] - [168] , [170] - [174] • How to systematically locate useful data from inherently noisy social and physical signals? • How to gain access to sensing data from privately-owned devices?

Human-Cyber-Physical Interactions Challenge

Handling the complex interactions between the human, cyber, and physical domains [87] , [128] , [180] - [182] , [222] • How to develop a closed-loop system that seamlessly integrates social and physical sensors? • How to explicitly model the roles of human participants as actuators? • How to use physical sensors to validate knowledge contributed by human sensors? • How to leverage social signals to effectively control physical sensors' performance?

Device and Data Heterogeneity Challenge

Managing the diversity of the devices and data associated with the social and physical sensors [122] , [186] - [192] , [198] , [200] , [202] • How to apply global policies and control privately-owned devices from a central authority perspective? • How to explicitly consider the heterogeneity of tasks and architectures for devices? • How to manage the complex interdependence of tasks distributed across multiple devices? • How to analyze the different types of data that vary across dimensionality? • How to handle the different rates of data generation by social and physical sensors?

Characterizing the dependencies between sources and correlating the collected data [127] , [206] - [210] • How to model source dependency and data provenance, given diverse source dependency nature of social and physical sensing? • How to identify and incorporate implicit correlations within events obtained from social and physical sensors? • How to explore strong causal relationships between physical and social sensor data?

Mitigating privacy issues arising from the integration of social and physical sensors [76] , [211] , [212] , [212] - [215] , [215] • How to develop robust privacy conserving schemes to prevent the malicious exploitation of complementary information from social and physical sensors? • How to design integrated privacy-aware SPS platforms to concurrently consider the data heterogeneity and protect sensitive user information?

Adapting to the interrelated dynamics from the social and physical realms [22] , [128] , [221] , [223] , [224] • How to handle interrelated dynamics induced by the fusion of social and physical domains? • How to adapt to the dynamics from the social domain which impacts the performance of physical sensing? • How to adapt to the physical world dynamics which affect the performance of both the social and physical sensors? ated [253] . Thus, further research can focus on constructing cryptographic SPS approaches that can cater to the heterogeneity of the devices in SPS and effectively trade-off privacy and sensing quality. Another potential direction for future work on quality-aware privacy preservation in SPS is to explore and incorporate approaches like differential privacy, where noise is deliberately added to the user data to conceal the sensitive information of users [254] , [255] . While current differential privacy techniques have been applied in participatory and crowdsensing, such approaches have not considered the data heterogeneity issue prevalent in SPS. Due to the wide variety of the data generated by the social and physical sensors in SPS (e.g., text, image, audio, and location data), injecting deliberate noise for concealing user identity into different data might be computationally intensive and resource demanding. As such, further work can concentrate on alleviating the data heterogeneity in SPS applications by developing efficient differential privacy techniques.

While SPS applications deliver a multifaceted sensing package using a combination of social and physical sensors, one remaining issue is ensuring fairness alongside accuracy for the data obtained from diverse demographics [256] , [257] . With the advent of numerous data acquisition platforms as well as processing techniques, there is a heightening concern from various civil rights organizations, governments, and analysts regarding the fairness of the detection process in SPS applica-tions and their prevalent algorithmic bias towards specific demographic groups [258] . For example, in a contact tracing SPS application as illustrated in Figure 23 , overrepresented classes of data might cause a certain age of people (e.g., teenagers) to be incorrectly represented as the prime sources of the disease. One issue that arises when trying to ensure fairness and accuracy in the distribution of input data is the loss of model accuracy. Specifically, in order to reduce the bias, it is crucial to incorporate a wider distribution of data from different classes (e.g., race, ethnicity, nationality) of data contributors. However, to reduce bias by incorporating wider distribution of data, the inference models in SPS need to train over a larger sample of data, causing the overfitting/underfitting problem, which often leads to the reduction in model accuracy [259] . The fairness and accuracy issue in SPS is further exacerbated by the fact that certain demographics might be more inclined in using smart devices more often than others. For example, in an anomalistic crowd investigation application using SAS, younger people might use their mobile devices to post crowdrelated events more frequently on social media while senior people might not report their observations so often on social media. As such, a crowd inference model might be overfitted with a younger demographic. Current fairness and accuracy optimizing schemes are limited in addressing such diverse device usage scenarios present in SPS applications. Existing schemes have attempted to reduce algorithmic bias by using heuristic approaches such as genetic algorithm [260] , optimizing the model's loss function [261] , or ensuring that the model training process satisfies given fairness constraints [262] . The problem of such approaches is that they have been originally designed for input data that are of fairly good quality. However, in the context of SPS applications, both social and physical sensors are prone to systematic noise, which is hard to quantify and model due to their complex interdependence with each other [263] . Thus, further research can concentrate on optimizing the fairness and accuracy in SPS applications while concurrently offsetting the noise generated from the social and physical sensors.

Techniques such as deep learning (DL)-based collaborative filtering [264] , discrimination-aware channel pruning [265] , and selective adversarial networks [266] could be explored to develop such fairness and accuracy optimizing methods. In addition to mitigating the noise contained in the input signals in SPS, one strand of research can be focused on developing user-friendly and accessible interactive interfaces (e.g., interactive kiosks, smartphone applications, responsive websites, augmented reality experiences) for collecting fair data samples in SPS given the potential demographic bias in the participants.

One route for future research in SPS can be focused on addressing the dynamics challenge in SPS. As discussed in Section V-F, a critical task in SPS applications is handling the interrelated dynamics caused by the constantly transitioning social and physical environments. Adaptive Artificial Intelligence (AI) algorithms are known to adjust their parameters to cater to changing stimuli [222] . As such, AI algorithms might help to potentially adapt to the constantly changing social and physical environments. However, several difficulties prevent off-the-shelf AI algorithms from being directly applied to SPS applications to address the dynamics challenge.

First, as identified in Section V-B, one recurring issue stemming from the human-cyber-physical interactions challenge in SPS applications is the inconsistent availability of the social and physical sensors, known as churn [267] . Many AI algorithms heavily rely on the participation of the sensing devices for the training phase, which requires multiple iterations to converge to a global optima. Given the churn involved in SPS applications, it is often difficult for AI algorithms to accurately classify the incoming sensing measurements. As a result, these AI algorithms might end up with failure scenarios in SPS applications with significant dynamics. Second, SPS applications typically involve a large number of privately-owned devices and often users do not provide access to their devices with concerns of privacy or excessive bandwidth usage. Traditional distributed AI algorithms often tend to assume unrestricted access to local datasets from individuals' devices [268] , which may not always be true in SPS applications. Given the inaccessibility of user data across privately-owned devices, existing distributed AI algorithms fall short in addressing the dynamics challenge in SPS applications. Third, SPS applications involve a diverse set of devices along with a wide range of data types (i.e., data and device heterogeneity). However, while certain existing distributed AI algorithms are designed to handle heterogeneous data, the computational complexity of such algorithms tend to be relatively high, which might overload resource-constrained devices (e.g., smartphones, UAVs) used in SPS applications [269] .

Several future avenues for research can be explored to tackle the above difficulties. One potential realm of further work can focus on using deviceless pipelining techniques to offload and distribute AI model training subtasks in SPS applications across devices equipped with specialized hardware [267] . For example, in a disaster response application with SAS, an UAV fitted with a GPU having large video RAM can be used to execute grid search for hyperparameters in AI model training while another UAV fitted with an FPGA can be used for pooling and flattening subtasks. A second emerging model training technique is to incorporate human intelligence (HI) for augmenting AI algorithms and enhancing their performance [270] . HI platforms such as Amazon MTurk have allowed human participants to provide their inputs for labels or features which might be potentially leveraged to retrain the AI models, address their innate flaws [21] . Thus, further research in SPS can focus on incorporating HI with AI to develop robust human-AI algorithms for SPS applications. A third probable future avenue of research can focus on designing decentralized model training algorithms for collaboratively acquiring local model updates from privately-owned devices. With the intent for preserving privacy and reducing network bandwidth requirements, federated learning (FL) is gaining traction as a decentralized AI training paradigm [271] , [272] , where a shared global AI model is trained from a collection of edge devices owned by end users [273] . Future research can focus on constructing FL solutions that can consider the data and device heterogeneity originating from the social and physical sensors in SPS.

In this paper, we present a comprehensive study of SPS, an emerging integrated networked sensing paradigm that exploits the collective strengths of physical and social sensing to acquire and interpret measurements from the environment. Empowered by the ubiquity of versatile data capture, communication, and computing technologies, SPS melds the human wisdom-driven data acquisition from social sensors with the multifaceted sensing capabilities of physical sensors to deliver a deeper perception of the real-world, both physically and socially. In particular, this paper surveys the various aspects that are important for constructing compelling SPS systems, which includes an explicable definition and overview of SPS discussing, the key motivation behind its origin, the crucial technologies and protocols that enable SPS, the real-world SPS applications and state-of-the-art solutions, the key challenges prevalent in SPS, and the potential avenues for further work to address the challenges. We hope this paper will bridge the knowledge gap from current literature in SPS and motivate future studies to design novel SPS systems for more accurate and comprehensive perception of the real-world phenomena.

A survey of intrusion detection techniques for cyber-physical systems

Array of things: a scientific research instrument in the public way: platform design and early lessons learned

Distributed embedded smart cameras for surveillance applications

Annbased soft-sensor for real-time process monitoring and control of an industrial polymerization process

Physiodroid: Combining wearable health sensors and mobile devices for a ubiquitous, continuous, and personal monitoring

Social sensing: building reliable systems on unreliable data

Using humans as sensors: An estimationtheoretic perspective

Social media analytics-challenges in topic discovery, data collection, and data preparation

A tale of many cities: universal patterns in human urban mobility

Constraint-aware dynamic truth discovery in big data social media sensing

Interpretable poverty mapping using social media data, satellite images, and geospatial information

Using nightlight remote sensing imagery and twitter data to study power outages

Social sensing of urban land use based on analysis of twitter users' mobility patterns

Covidsens: a vision on reliable social sensing for covid-19

Wireless sensors networks for internet of things

Collabdrone: A collaborative spatiotemporal-aware drone sensing system driven by social sensing signals

Forest fire monitoring with multiple small uavs

Tasks allocation for real-time applications in heterogeneous sensor networks for energy minimization

Using transmit-only sensors to reduce deployment cost of wireless sensor networks

The age of social sensing

Crowdlearn: A crowd-ai hybrid system for deep learning-based damage assessment applications

Introduction to social sensing and big data computing for disaster management

On truth discovery in social sensing: A maximum likelihood estimation approach

ACM/IEEE 11th Int Information Processing in Sensor Networks (IPSN) Conf

Towards reliable missing truth discovery in online social media sensing applications

Towards reliable online clickbait video detection: A content-agnostic approach

On opinion characterization in social sensing: A multi-view subspace learning approach

Participant privacy in mobile crowd sensing task management: A survey of methods and challenges

On robust truth discovery in sparse social media sensing

Integrating physical and social sensing to enable smart city mobility services

Cyberphysical-social frameworks for urban big data systems: A survey

Recursive fact-finding: A streaming approach to truth estimation in crowdsourcing applications

An integrated social media and drone sensing system for reliable disaster response

Developing an integration framework for crowdsourcing and internet of things with applications for disaster response

Urban healthcare big data system based on crowdsourced and cloud-based air quality indicators

Start from scratch: A crowdsourcing-based data fusion approach to support location-aware applications

An improved scheme for privacy-preserving collaborative anomaly detection

A survey of urban vehicular sensing platforms

A survey of indoor localization systems and technologies

Profiling the mobility of tourists exploring social sensing

Brief survey of crowdsourcing for data mining

A survey on truth discovery

A survey of cyber-physical systems

A survey: Cyberphysical-social systems and their system-level design methodology

Cyber physical social systems: Towards deeply integrated hybridized systems

Data fusion in cyberphysical-social systems: State-of-the-art and perspectives

Wireless sensor networks: a survey

Cyclostationary statistical models and algorithms for anomaly detection using multimodal data

On the application of social internet of things with fog computing: a new paradigm for traffic information sharing system

Secondnose: an air quality mobile crowdsensing system

Internet of things for noise mapping in smart cities: State of the art and future directions

Edgebatch: Towards ai-empowered optimal task batching in intelligent edge systems

A survey on mobile crowd-sensing and its applications in the iot era

The sensable city: A survey on the deployment and management for smart city monitoring

Social media in disaster response: How experience architects can build for participation. Routledge

Crowdsourcing-based copyright infringement detection in live video streams

Exfaux: A weakly supervised approach to explainable fauxtography detection

Uses and gratifications of social networking sites for bridging and bonding social capital: A comparison of facebook, twitter, instagram, and snapchat

Social media? get serious! understanding the functional building blocks of social media

Social networks and social information filtering on digg

Framing social news sites: An analysis of the top ranked stories on reddit and digg

Applications of crowdsourcing in health: an overview

Automatically characterizing places with opportunistic crowdsensing using smartphones

Enhancing traffic safety with the implementation of crowdsensing solutions in the mobile era

An investigation of unpaid crowdsourcing

Safestreet: An automated road anomaly detection and early-warning system using mobile crowdsensing

Crowd sensing for disaster response and preparedness

Citysee: Not only a wireless sensor network

Leveraging mobile health applications for biomedical research and citizen science: a scoping review

Distributed event-triggered estimation over sensor networks: A survey

A novel solution to the reader collision problem in rfid system

Outdoor air quality level inference via surveillance cameras

Dualsurface dielectric depth detector for holographic millimeter-wave security scanners

Power sources for the internet of things

Recent advances in space-borne optical remote sensing systems for monitoring global terrestrial ecosystems

Kalman filtered gps accelerometer-based accident detection and location system: A low-cost approach

Internet of things: A survey on enabling technologies, protocols, and applications

Emerging wireless standardswifi, zigbee and wimax

An overview of the bluetooth wireless technology

LTE-the UMTS long term evolution: from theory to practice

A survey of 5g network: Architecture and emerging technologies

Lte-advanced: next-generation wireless broadband technology

A layered architecture for vehicular delay-tolerant networks

Improving message delivery in uav-based delay tolerant networks

A delay-tolerant network architecture for challenged internets

Investigating messaging protocols for the internet of things (iot)

An efficient identitybased batch verification scheme for vehicular sensor networks

Dasc: Towards a road damage-aware social-media-driven car sensing framework for disaster response applications

Emerging flood model validation frameworks for street-level inundation modeling with stormsense

Crowd and event detection by fusion of camera images and micro blogs

Motives and concerns of dashcam video sharing

Performance of dsrc and wifi for intelligent transport systems in vanet

A taxonomy and survey of cloud computing systems

The rise of "big data" on cloud computing: Review and open research issues

Serverless computation with openlambda

Thingspeak-an api and web service for the internet of things

Apache spark

A survey on the edge computing for the internet of things

When social sensing meets edge computing: Vision and challenges

Learning iot in edge: Deep learning for the internet of things with edge computing

A realtime and non-cooperative task allocation framework for social sensing applications in edge computing systems

Contact tracing and disease control

Epic: efficient privacypreserving contact tracing for infection detection

Exposure notifications: Using technology to help public health authorities fight covid-19

Getting behind covid-19 contact trace apps: The google-apple partnership

Adding location and other context to the google/apple exposure notification bluetooth api

Government access to mobile phone data for contact tracing

Bluetrace: A privacy-preserving protocol for communitydriven contact tracing across borders

Acoustic-turf: Acoustic-based privacy-preserving covid-19 contact tracing

The agenda of wearable healthcare

Social media use in healthcare: a systematic review of effects on patients and on their relationship with healthcare professionals

In-patient alert system in hospital using twitter-bot

A privacy-preserving cryptosystem for iot e-healthcare

Family burden in long-term diseases: a comparative study in schizophrenia vs. physical disorders

Analysis of an online health social network

A low-cost iot-based health monitoring platform enriched with social networking facilities

Integrating social networks and remote patient monitoring systems to disseminate notifications

A pervasive health system integrating patient monitoring, status logging, and social sharing

Fusing heterogeneous data: A case for remote sensing and social media

A novel methodology to label urban remote sensing images based on locationbased social media photos

Remote sensing and social sensing for socioeconomic systems: A comparison study between nighttime lights and location-based social media at the 500 m spatial resolution

A cloud-enabled automatic disaster analysis system of multi-sourced data streams: An example synthesizing social media, remote sensing and wikipedia data

Risksens: A multi-view learning approach to identifying risky traffic locations in intelligent transportation systems using social and remote sensing

Rapid flood inundation mapping using social media, remote sensing and topographic data

Multisource and multitemporal data fusion in remote sensing

Multisource and multitemporal data fusion in remote sensing: A comprehensive review of the state of the art

Towards a social-media driven multi-drone tasking platform

Sead: Towards a social-media-driven energy-aware drone sensing framework

Socialcar: A task allocation framework for social media driven vehicular network sensing systems

Firework: Big data sharing and processing in collaborative edge environment

Enhancing amber alert using collaborative edges: Poster

Automatic license plate recognition (alpr): A state-of-the-art review

Edgebatch: Towards ai-empowered optimal task batching in intelligent edge systems

Crowdsourcing cctv surveillance on the internet

Privacy-aware average speed monitoring system for vehicular ad-hoc networks

Crowdpark: A crowdsourcing-based parking reservation system for mobile phones

Crowd sensing aware disaster framework design with iot technologies

The internet of things: A survey

Internet of things for smart cities

Social-loc: Improving indoor localization with social sensing

Indoor localization and visualization using a human-operated backpack system

Sbot: A social media based object tracking system

Big data privacy in the internet of things era

Crowdsourcing and CCTV: The effect of interface, financial bonus and video type

Enhancing emergency response systems through leveraging crowdsensing and heterogeneous data

Determining flooded areas using crowd sensing data and weather radar precipitation: a case study in brazil

Harnessing the power of crowdsourcing and internet of things in disaster response

Mobile crowdsensing: current state and future challenges

Opportunities in mobile crowd sensing

Mobile crowd sensing and computing: The review of an emerging human-powered sensing paradigm

Surrogate mobile sensing

Using mobile crowd sensing for noise monitoring in smart cities

Cloudassisted mobile crowd sensing for traffic congestion control

Toward a mobile crowdsensing system for road surface assessment

Location-aware task assignment and routing in mobile crowd sensing

Architecture design of the air pollution mapping system by mobile crowd sensing

Greengps: A participatory sensing fuel-efficient maps application

Optimizing quality-of-information in cost-sensitive sensor data fusion

Sparse user check-in venue prediction by exploring latent decision contexts from location-based social networks

A framework for fusion of human sensor and physical sensor data

Fault detection using redundant navigation modules

Combination of singular value decomposition and k-means clustering methods for topic detection on twitter

Structured prediction models for rnn based sequence labeling in clinical text

A light-weight and quality-aware online adaptive sampling approach for streaming social sensing in cloud computing

Transres: a deep transfer learning approach to migratable image super-resolution in remote urban sensing

Deep learning-based noise reduction approach to improve speech intelligibility for cochlear implant recipients

Biggerpicture: data-driven image extrapolation using graph matching

Application of iot in military operations in a smart city

A survey on privacy preserving data mining

Toward privacy-aware task allocation in social sensing-based edge computing systems

Sedic: privacyaware data intensive computing on hybrid clouds

Bluetooth low energy

Ultralow power data storage for sensor networks

Bandwidth management in wireless sensor networks

Enhanced human-computer interaction for business applications on mobile devices: a design-oriented development of a usability evaluation questionnaire

Review of cyber-physical system architecture

Crowd-assisted disaster scene assessment with human-ai interactive attention

Smart water quality monitoring system

Pay as how well you do: A quality based incentive mechanism for crowdsensing

Optimal path tracking control of autonomous vehicle: Adaptive full-state linear quadratic gaussian (lqg) control

Heteroedge: taming the heterogeneity of edge computing system in social sensing

Understanding the emergence and importance of blockchain-based cyber-physical social machines: A proposal

Data cleaning: Overview and emerging challenges

Transland: An adversarial transfer learning approach for migratable urban land usage classification using remote sensing

Transres: A deep transfer learning approach to migratable image super-resolution in remote urban sensing

A system for simulation, emulation, and deployment of heterogeneous sensor networks

Dynamic iot device clustering and energy management with hybrid noma systems

Sensor abstraction layer: a unique software interface to effectively manage sensor networks

Virtualization and containerization of application infrastructure: A comparison

A dynamic and static combined register mapping method in binary translation

Software-defined networking

Wireless sensor network virtualization: A survey

Delay minimization offloading for interdependent tasks in energy-aware cooperative mec networks

Computational offloading strategy based on dynamic particle swarm for multi-user mobile edge computing

(big) data in a virtualized world: volume, velocity, and variety in cloud datacenters

The emerging" big dimensionality

Scale-space color blob and ridge detection

Online bagging and boosting

A survey on deep learning for multimodal data fusion

Comparison of two measurement fusion methods for kalman-filter-based multisensor data fusion

Dimensionality reduction based on tensor modeling for classification methods

Riskcast: social sensing based traffic risk forecasting via inductive multi-view learning

Iot, big data and artificial intelligence in agriculture and food industry

Industrial big data as a result of iot adoption in manufacturing

Provenance-assisted classification in social networks

Fake news pattern recognition using linguistic analysis

A trust model for analysis of trust, influence and their relationship in social network communities

A new mathematical formulation and a heuristic for the maximum disjoint set covers problem to improve the lifetime of the wireless sensor network

Clarisense+: An enhanced traffic anomaly explanation service using social network feeds

Non-parametric causality detection: An application to social media and financial data

Privacypreserving raw data collection without a trusted authority for iot

Personalization and privacy: a survey of privacy risks and remedies in personalization-based systems

Privacy-aware edge computing in social sensing applications using ring signatures

Poolview: stream privacy for grassroots participatory sensing

Privacy preservation in wireless sensor networks: A state-of-the-art survey

A personalized privacy protection framework for mobile crowdsensing in iiot

Dtaint: detecting the taint-style vulnerability in embedded device firmware

Side channel attack: an approach based on machine learning

Side channel attacks on smart home systems: A short overview

On the routing problem in disconnected vehicular ad-hoc networks

Exploitation of physical constraints for reliable social sensing

An integrated top-down and bottomup task allocation approach in social sensing based edge computing systems

Compdrone: Towards integrated computational model and social drone based wildfire monitoring

Towards scalable and dynamic social sensing using a distributed computing framework

Service composition in cyber-physical-social systems

On scalable and robust truth discovery in big data social media sensing applications

Social sensing: when users become monitors

Topic-aware social sensing with arbitrary source dependency graphs

Fuzzy logic from the viewpoint of machine intelligence

Quantifying uncertainty in remote sensing-based urban land-use mapping

Uncertainty quantification in digital image correlation for experimental evaluation of deep learning based damage diagnostic

Quantifying uncertainties of cloud microphysical property retrievals with a perturbation method

A comparative study of uncertainty propagation methods for black-box-type problems

Local wireless sensor networks positioning reliability under sensor failure

On credibility estimation tradeoffs in assured social sensing

On quantifying the accuracy of maximum likelihood estimation of participant reliability in social sensing

On bayesian interpretation of fact-finding in information networks

On scalability and robustness limitations of real and asymptotic confidence bounds in social sensing

Reliable social sensing with physical constraints: analytic bounds and performance evaluation

Streamguard: A bayesian network approach to copyright infringement detection problem in large-scale live video sharing systems

On a monte carlo method for measurement uncertainty evaluation and its implementation

An approximation approach for uncertainty quantification using evidence theory

A review of uncertainty quantification in deep learning: Techniques, applications and challenges

A computational framework for uncertainty quantification and stochastic optimization in unit commitment with wind power generation

Review of polynomial chaosbased methods for uncertainty quantification in modern integrated circuits

The conflict between big data and individual privacy

Protecting user privacy: Obfuscating discriminative spatio-temporal footprints

Privacy-preserving machine learning through data obfuscation

Towards reliability in online high-churn edge computing: A deviceless pipelining approach

Iot data privacy via blockchains and ipfs

Blockchains and smart contracts for the internet of things

Blockchain access privacy: Challenges and directions

Optimized ecc implementation for secure communication between heterogeneous iot devices

Deep learning with differential privacy

The composition theorem for differential privacy

Advances and open problems in federated learning

Fairfl: A fair federated learning approach to reducing demographic bias in privacy-sensitive classification models

Managing bias in ai

The accuracy, fairness, and limits of predicting recidivism

A generic algorithm for reducing bias in parametric estimation

Adafair: Cumulative fairness adaptive boosting

Faht: an adaptive fairness-aware decision tree classifier

Gan-driven personalized spatialtemporal private data sharing in cyber-physical social systems

Deepfair: deep learning for improving fairness in recommender systems

Discrimination-aware channel pruning for deep neural networks

One-network adversarial fairness

Towards reliability in online high-churn edge computing: A deviceless pipelining approach

Advances and open problems in federated learning

Performance-effective and low-complexity task scheduling for heterogeneous computing

Guidelines for human-ai interaction

Federated learning: Strategies for improving communication efficiency

Fedsens: A federated learning approach for smart health sensing with class imbalance in resource constrained edge computing

Social edge intelligence: Integrating human and artificial intelligence at the edge