Crowdsensing Data Research Articles

An experimental analysis of outdoor UAV localisation through diverse estimators and crowd-sensed data fusion

Motivated by the challenge of achieving precise 3D outdoor localisation for unmanned aerial vehicles (UAVs) in global navigation satellite system (GNSS)-denied environments, this paper introduces an innovative technique. Integrating crowd-sensed data fusion to counter inertial navigation system (INS) drift during GNSS signal outages, the proposed method exploits diverse estimators to enhance its efficacy. A micro lightweight frequency modulated continuous wave (FMCW) radar mounted on the UAV captures ground scatterer reflections, processed via fast Fourier transform (FFT) to generate a range-Doppler map. This map facilitates forward velocity estimation during GNSS signal loss. This approach employs adaptive thresholding, image binarisation, and connected components-based techniques for target detection from a computer vision standpoint. The derived radar-based velocity fuses with magnetometer, barometer, and inertial measurement unit (IMU) data using diverse estimators like extended Kalman filter (EKF) and particle filter (PF). Real-time flight data evaluation and simulated outage periods using EKF and PF validate the outdoor localisation system. Experimental analyses demonstrate substantial improvements, enhancing 3D positioning accuracy by 99.89% and 99.83% for the initial and subsequent flights, respectively, leveraging PF to fortify the standalone INS mode during GNSS signal loss. This approach significantly enhances UAV localisation precision, particularly in challenging GNSS-denied scenarios, showcasing the potential for real-world applications.

FedCrow: Federated-Learning-Based Data Privacy Preservation in Crowd Sensing

In the process of completing large-scale and fine-grained sensing tasks for the new generation of crowd-sensing systems, the role of analysis, reasoning, and decision making based on artificial intelligence has become indispensable. Mobile crowd sensing, which is an open system reliant on the broad participation of mobile intelligent terminal devices in data sensing and computation, poses a significant risk of user privacy data leakage. To mitigate the data security threats that arise from malicious users in federated learning and the constraints of end devices in crowd-sensing applications, which are unsuitable for high computational overheads associated with traditional cryptographic security mechanisms, we propose FedCrow, which is a federated-learning-based approach for protecting crowd-sensing data that integrates federated learning with crowd sensing. FedCrow enables the training of artificial intelligence models on multiple user devices without the need to upload user data to a central server, thus mitigating the risk of crowd-sensing user data leakage. To address security vulnerabilities in the model data during the interaction process in federated learning, the system employs encryption methods suitable for crowd-sensing applications to ensure secure data transmission during the training process, thereby establishing a secure federated-learning framework for protecting crowd-sensing data. To combat potential malicious users in federated learning, a legitimate user identification method based on the user contribution level was designed using the gradient similarity principle. By filtering out malicious users, the system reduces the threat of attacks, thereby enhancing the system accuracy and security. Through various attack experiments, the system’s ability to defend against malicious user attacks was validated. The experimental results demonstrate the method’s effectiveness in countering common attacks in federated learning. Additionally, through comparative experiments, suitable encryption methods based on the size of the data in crowd-sensing applications were identified to effectively protect the data security during transmission.

IPREDICT: AI enabled proactive pandemic prediction using biosensing wearable devices

The emergence of pandemics poses a persistent threat to both global health and economic stability. While zoonotic spillovers and local outbreaks may not be fully preventable, early detection of infections in individuals before they spread to communities can make a major difference in containing an infectious disease and stopping it from becoming an epidemic and then a pandemic.In this paper, we propose a novel Artificial Intelligence (AI)-based pandemic prediction framework called iPREDICT—a concept framework designed to leverage the power of AI and crowd-sensed data for accurate and timely pandemic prediction. The core idea of iPREDICT is to leverage the deluge of data that can be harnessed from connected and wearable biosensing devices. iPREDICT system then works by monitoring anomalies in the biomarkers at the individual level and correlating them with similar anomalies observed in other members of the community. Using AI-based anomaly detection in conjunction with analysis of the spatiotemporal growth of the correlated anomalies, iPREDICT thus can potentially detect and monitor the emergence of a local outbreak in near real-time to predict a potential pandemic.However, not every outbreak has the potential to become a pandemic. We illustrate how tools like graph neural networks can be leveraged to determine optimal thresholds as a function of a large number of demographical, social, and geographical factors that determine the spatiotemporal spread of an outbreak, thus quantifying the risk of it becoming an epidemic or pandemic.We also identify essential technological and social challenges that require attention to transform iPREDICT from an idea into a globally deployable solution for future pandemic prediction and management. To provide deeper insights into iPREDICT design challenges and trigger research towards possible solutions we present a COVID-19 based case study. The results signify the impact of variation in biosensing hardware, data sampling rate, and compression rate on the performance of AI models that underpin various components of the iPREDICT system.

Analyzing the impact of fare-free public transport policies on crowding patterns at stations using crowdsensing data

Fully or partially fare-free public transport (FFPT) is a measure to make public transport (PT) more attractive and affordable. Cities worldwide are experimenting with a variety of fare discount policies, which will lead to spatiotemporal changes in mobility patterns, including crowding in PT stations. However, the response of PT stations to these policies can vary due to the heterogeneity in their surrounding built environment and other factors. Non-conventional data sources could be used to better model and understand these spatiotemporal dynamics. In this study, we propose a three-step methodological framework to understand the impact of FFPT or extreme fare discount interventions on PT demand patterns, specifically focusing on crowding patterns in PT stations, using crowdsensing data. First, we design a busyness-based similarity measure that leverages the histogram method to capture changes in crowding patterns. Then, we employ a Gaussian Mixture Model (GMM) to cluster PT stations based on their crowding pattern deviations at different stages of policy implementation. This clustering step enables the identification of distinct station types based on their response to the policy. Finally, we train a LightGBM model to learn the relationship between crowding pattern changes and the spatial–temporal characteristics of PT stations, using the busyness-based station types identified by GMM as labels. We apply our methodology to a public transport experiment in Germany during the summer of 2022 when the country introduced a monthly “9-EUR” ticket valid on local and regional PT nationwide. The clustering results show three station types: unaffected, mildly stimulated, and intensely stimulated stations. Furthermore, the classifier indicates that the station’s location, activity options near the station, and population within the adjacent area of the station, and the crowding patterns under normal operations (before policy implementation) play a significant role in the heterogeneity of the 9-EUR ticket’s impact. The insights gained from our study can help planners better understand and manage the crowding at PT stations during fare interventions.

An improved indoor positioning based on crowd-sensing data fusion and particle filter

Due to the lack of global positioning system (GPS) signals in some enclosed areas, indoor localisation has recently gained significant importance for academics. However, indoor localisation has a number of challenges and defects, including accuracy, cost, coverage, and ease of use. This paper explores the integration between the inertial measurement unit (IMU) and Wi-Fi-based received signal strength indicator (RSSI) measurements, demonstrating their combined potential for robust indoor localisation. IMUs excel at capturing precise short-term motion dynamics, offering insights into an object’s acceleration and orientation. Conversely, RSSI measurements serve as valuable indicators for relative positioning within indoor environments. By fusing data from these sources, our approach compensates for the inherent weaknesses of each sensor type. To achieve accurate indoor positioning, we employ techniques such as sensor fusion, Wi-Fi fingerprinting, and dead reckoning. Wi-Fi fingerprinting allows us to create a database that maps RSSI measurements to specific locations, while dead reckoning helps mitigate drift and inaccuracies. By combining these methods, we estimate a device’s position with increased precision. Through experimental evaluation, we assess the performance and efficiency of our integrated approach, comparing the estimated path or new location with a predefined reference path. The findings emphasise a significant improvement in accuracy, with the integration of crowd-sensing, particle filtering, and magnetic fingerprinting techniques resulting in a notable increase from 80.49% to 96.32% accuracy.

GENII: A graph neural network-based model for citywide litter prediction leveraging crowdsensing data

Plastic litter and its associated environmental hazards have garnered global attention in recent years, highlighting the need for effective management. Improper handling of plastic litter can lead to environmental degradation, making it crucial to address this issue. In this paper, we propose an innovative approach to predict the spatial distribution and quantity of plastic litter at the city level by leveraging crowd-sensing data and designing a graph neural network-based model. Meteorological data is specifically used to enhance temporal correlation, while the spatial distribution of litter is clustered using the K-means method to capture spatial correlation. For each cluster, multiple graphs are constructed based on the points of interest (POIs) and litter distribution within the cluster. Graph attention neural networks and heterogeneous graph attention networks are then utilized to aggregate the adjacency information and overall structural information of these graphs, respectively. Finally, the litter prediction results are obtained through multiple fully connected layers. Real-world experiments convincingly demonstrate the high effectiveness of our proposed model in predicting city-wide litter, surpassing multiple comparative models. All the code and datasets can be accessed from the GitHub repository at https://github.com/ZJUDataIntelligence/Genii.

Use of Mobile Crowdsensing in Disaster Management: A Systematic Review, Challenges, and Open Issues.

With the increasing efforts to utilize information and communication technologies (ICT) in disaster management, the massive amount of heterogeneous data that is generated through ubiquitous sensors paves the way for fast and informed decisions in the case of disasters. Utilization of the big "sensed" data leads to an effective and efficient management of disaster situations so as to prevent human and economic losses. The advancement of built-in sensing technologies in smart mobile devices enables crowdsourcing of sensed data, which is known as mobile crowdsensing (MCS). This systematic literature review investigates the use of mobile crowdsensing in disaster management on the basis of the built-in sensor types in smart mobile devices, disaster management categories, and the disaster management cycle phases (i.e., mitigation, preparedness, response, and recovery activities). Additionally, this work seeks to unveil the frameworks or models that can potentially guide disaster management authorities towards integrating crowd-sensed data with their existing decision-support systems. The vast majority of the existing studies are conceptual as they highlight a challenge in experimental testing of the disaster management solutions in real-life settings, and there is little emphasis on the use cases of crowdsensing through smartphone sensors in disaster incidents. In light of a thorough review, we provide and discuss future directions and open issues for mobile crowdsensing-aided disaster management.

Efficient and Privacy-Preserving Aggregated Reverse kNN Query Over Crowd-Sensed Data

Efficient and Privacy-Preserving Aggregated Reverse kNN Query Over Crowd-Sensed Data

A crowdsensing cloud-based microservice for weather forecast

We present a cloud-based microservice enabling individual users to have access to local air pressure measurements and a proof-of-concept web application providing shot-term weather forecasts based only on local crowdsensed data. Our microservice has been developed using state-of-the-art web technologies and cutting-edge technologies related to cloud computing. Apart from its individual interest and potential usefulness, our microservice and proof-of-concept web application show how modern cloud-based technologies and tools can be nicely combined to improve the community daily routine and wellbeing.

Driving Behavior Identification and Real-World Fuel Consumption Estimation With Crowdsensing Data

It is now well established that car driving behavior impacts gas pollutant emission, fuel consumption and safety. With the rise of low-cost and easily accessible mobility data (measured by GPS, accelerometer, gyroscope, etc), through OBD or smartphones devices for instance, the community is increasingly interested in characterizing these driving behaviors. In this context, we used the IFPEN’s application named GECO AIR to access GPS data on which we derived statistical descriptors to perform unsupervised classifications and highlight different trip- and driver-related behaviors (dynamic, slow, traffic jam, etc.), on different road types (urban, highway, etc). Using Markov chain, we then generate for each characterized behavior a representative velocity profile and combine them according to a given use to estimate the associated real-world fuel consumption. Promising performances have been obtained with more than half of the recorded trips for which the real-world fuel consumption is estimated with less than 10% of error. The added value of the proposed work is the capability to estimate fuel consumption in a posterior way without any GPS records, as information obtained through a questionnaire for instance could be sufficient.

Human as a Service: Towards Resilient Parking Search System With Sensorless Sensing

The high demand for ubiquitous availability of reliable parking spaces in cities faces challenges on timely information sharing and low-cost infrastructure deployment. In this paper, we propose a mobile crowdsensing system, namely ParkHop, to aggregate on-street and roadside parking space information through sensorless sensing, and disseminate this information to urban drivers in a resilient manner. ParkHop targets special social groups that have stable work routines to serve as crowd workers. We propose a crowdsensing algorithm employing a joint estimator to process crowdsensed data, and evaluate the reliability of crowd workers based on the veracity of their answers to a series of control questions. In addition, the specific worker selection method to speed up the crowdsensing process and incentive scheme to achieve fair reward distribution have been carefully designed in ParkHop. Our system disseminates the availability of parking spaces and their up-to-date price information to drivers with on-demand needs via a publish-subscribe messaging pattern. The efficacy of ParkHop for aggregation and dissemination of parking space information has been evaluated in both real-world tests and simulations. Our results show the system is robust and agile enough to cope with different crowdsensing scenarios.

BigEye: Detection and Summarization of Key Global Events From Distributed Crowdsensed Data

Social media postings using smartphones (referred to as crowd-sensed data) can often facilitate real-time detection of key physical events in applications like disaster recovery or in smart cities. These postings also often contain visual content (e.g., images) that can be used to obtain zoomed-in views of such events. These crowd-sensed data are likely to be of large volume and distributed across a plurality of producers (e.g., cloudlets). Blindly transferring this large volume of raw data from the producers to a consumer will induce information overload and consume very high bandwidth. The problem is exacerbated in scenarios with limited bandwidth (e.g., after a disaster). In this article, we design <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BigEye</monospace> , a novel framework that only transfers very limited data from the distributed producers to a central summarizer, and yet supports: 1) highly accurate detection and 2) concise visual summarization of key events of global interest. In realizing <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BigEye</monospace> , we address several challenges including: 1) identifying events that have the highest global interest via the transfer of appropriate limited metadata from the producers to the summarizer; 2) reconciling metadata that could be inconsistent across the producers; and 3) the timely retrieval of visual summaries of the key events given bandwidth constraints. We show that <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BigEye</monospace> achieves the same accuracy in detecting key events, as a system, where all data are available centrally while transferring only 1% of the raw data volume. Compared to the baseline approaches, <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BigEye</monospace> ’s parallelized transfer of visual content reduces the average delay by 67%.

Traffic Regulation Recognition using Crowd-Sensed GPS and Map Data: a Hybrid Approach

Abstract. This article presents a method for traffic control recognition at junctions (traffic lights, stop, priority and right of way rule) using crowd-sensed GPS data (vehicle trajectories), as well as features extracted from OpenStreetMap. Traffic regulators are not mapped in most maps, although the way they regulate traffic at intersections affects the traffic flow and therefore the vehicle idle time at intersections, the fuel consumption, the CO2 emissions, and the arrival time at a destination. Because of the controlled interaction that road users have with each other at intersections, driving safety or assistance applications can be enabled if intersection regulators are mapped. In order to verify the proposed method two sets of trajectories were used, one of which is an open dataset, from two different cities, Hannover and Chicago. Two classification methods were tested, random forest and gradient boosting, using exclusively either dynamic features (trajectories), or static (only data from OSM) or a combination of the dynamic and static features (hybrid model). The results show that the gradient boosting classification with hybrid features can predict traffic regulations with high accuracy (93% in Chicago and 94% in Hannover), outperforming the other detection models (static and dynamic). At the end directions for further research on this topic are proposed.

Privacy-Enhanced and Practical Truth Discovery in Two-Server Mobile Crowdsensing

In mobile crowdsensing, truth discovery (TD) enables a crowdsensing server to extract truthful information from possibly conflicting crowdsensing data. TD provides a more accurate truth estimation than traditional truth inference methods like majority voting and averaging. However, there still exist crucial data privacy (including sensory data, inferred truths, and intermediates) and practicability (e.g., efficiency, utility, and non-interaction) concerns in real-world crowdsensing applications. Existing researches either fail to provide adequate data privacy protection throughout the entire TD procedure or suffer from low practicability. In this paper, we propose two schemes: a basic privacy-aware TD scheme (BPTD) and a privacy-enhanced TD scheme (PETD) with two servers for mobile crowdsensing, comprehensively considering both privacy and practicability. BPTD is straightforwardly conducted on shared data with few user-side interactions, while achieving high efficiency. To further liberate mobile users and prevent disclosure of the intermediates, PETD incorporates a novel partial decryption-based Paillier Cryptosystem to work with secret sharing, offering enhanced privacy protection without relying on any user-side involvement. Additionally, we improve the efficiency of PETD via data packing. Security analysis shows the desired privacy goals. Compared to prior studies with the best security guarantees, our extensive experiments demonstrate a comparable and superior performance regarding different metrics.

Assessing neighborhood-scale traffic from crowd-sensed traffic data: Findings from an environmental justice community in New York City

BackgroundThe waterfront in the South Bronx in New York City is used industrially and harbors the Harlem River Yards (HRY). The HRY borders an environmental justice area, which includes a mixed-use area that is separated from a densely populated residential area by interstates. Recently, development of the HRY has expanded including the 2018 opening of a large online store warehouse. ObjectiveThe goal of this study was to evaluate trends in traffic congestion nearby the HRY between 2017 and 2019. MethodsWe analyzed one-hourly time series of crowd-sensed traffic congestion maps, both at the neighborhood scale and the road stretch level. Traffic radar measurements at two locations did not indicate bias in the crowd-sensed data over the study period, i.e., changed mappings between vehicle speed and the reported congestion. ResultsIn the mixed-use areas, traffic congestion increased significantly during all hours of the day, with greatest increases at night and in the morning. Congestion increased close to the entrances of the HRY and along routes used by pedestrians and bicyclists to access a nearby recreational area. In the residential area, congestion increased significantly from midnight to morning and was unchanged for the remainder of the day. On the interstates, congestion decreased during the daytime but increased at night. ConclusionsNeighborhood-scale traffic congestion increased in mixed-use and residential areas in an environmental justice community. Our methods can be applied globally as long as crowd-sensed traffic data can be acquired. The data enable communities to advocate for mitigating measures.

Influence of occupant presence patterns on energy consumption and its relation to comfort: a case study based on sensor and crowd-sensed data

BackgroundIn recent years, the monitoring of occupant presence patterns has become an imperative for building energy optimization. Very often, there is a significant discrepancy between the building energy performance predicted at the design stage and the actual performance rendered during the building operation. This stems from the difference in user occupancy. In spite of this, user interaction and feedback are rarely taken into account and evidence of the impact of occupant presence patterns on energy consumption is still scarce. Thus, the purpose of this study is to apply crowd-sensing techniques to understand how energy is consumed and how appropriate performance indicators should be defined to provide inputs for building operations regarding more efficient use of resources.MethodsMonitoring strategies were implemented in an office lab with controlled variables to collect quantitative data on occupancy patterns, ambient factors and energy consumption. In addition, crowd-sensing techniques were applied to model user activity in different ambient conditions over time and to contrast their occupancy with energy consumption patterns in combination with new inquiry tools to identify how occupants perceive their comfort level. In addition, a set of energy efficiency indicators was used to compare energy performance over different periods.ResultsIt was discovered that there is a strong relation between user occupancy patterns and energy consumption. However, more than 50% of energy was consumed when no user activity was registered. Energy performance indicators revealed that measuring energy efficiency in terms of kWh per surface area encourages a less efficient use of space and, therefore, including a coefficient of person hours is advisable. It was also discovered that users do not fully rely on feedback mechanisms and they prefer to take action to adapt the ambient conditions rather than simply expressing their opinion. Analysis of energy usage during the Covid-19 lock down revealed substantial use of energy contrary to what was expected. This was because home computers were used as terminals only, while the actual tasks were performed on the lab computers, using remote desktop connections, which were turned on 24/7. In addition, energy consumed by each employee at his/her home should be taken into account. Moreover, a set of practical recommendations was formulated.

Improving Destination Choice Modelling of Non-Routine Trips from Clustered Crowd Sensing Data

Improving Destination Choice Modelling of Non-Routine Trips from Clustered Crowd Sensing Data

Large-Scale Assessment of Mobile Crowdsensed Data: A Case Study

Mobile crowdsensing (MCS) is a well-established paradigm that leverages mobile devices’ ubiquitous nature and processing capabilities for large-scale data collection to monitor phenomena of common interest. Crowd-powered data collection is significantly faster and more cost-effective than traditional methods. However, it poses challenges in assessing the accuracy and extracting information from large volumes of user-generated data. SmartRoadSense (SRS) is an MCS technology that utilises sensors embedded in mobile phones to monitor the quality of road surfaces by computing a crowdsensed road roughness index (referred to as PPE). The present work performs statistical modelling of PPE to analyse its distribution across the road network and elucidate how it can be efficiently analysed and interpreted. Joint statistical analysis of open datasets is then carried out to investigate the effect of both internal and external road features on PPE. Several road properties affecting PPE as predicted are identified, providing evidence that SRS can be effectively applied to assess road quality conditions. Finally, the effect of road category and the speed limit on the mean and standard deviation of PPE is evaluated, incorporating previous results on the relationship between vehicle speed and PPE. These results enable more effective and confident use of the SRS platform and its data to help inform road construction and renovation decisions, especially where a lack of resources limits the use of conventional approaches. The work also exemplifies how crowdsensing technologies can benefit from open data integration and highlights the importance of making coherent, comprehensive, and well-structured open datasets available to the public.

Building a Secure Knowledge Marketplace Over Crowdsensed Data Streams

It is increasingly popular to leverage the wisdom of crowd for knowledge discovery and monetization. Among others, crowdsensing with truth discovery has emerged as a promising way for leveraging the crowd wisdom, which can mine reliable knowledge from the generally unreliable sensory data contributed collected from diverse sources. Building a knowledge marketplace based on crowdsensing with truth discovery for knowledge discovery and monetization, however, is non-trivial and has to overcome several challenges. First, the sensory data should be protected as they may carry sensitive information. Second, many real crowdsensing applications usually yield sensory data in a streaming fashion, posing the demand that truth discovery should be conducted over data streams to continuously mine reliable knowledge in each data collection epoch. Third, knowledge monetization should be well treated, fully addressing the practical needs of parties in the monetization ecosystem. In this article, we take the first research attempt and propose a new full-fledged framework for building a secure knowledge marketplace over crowdsensed data streams. Our marketplace supports secure monetization of reliable knowledge mined privately from data streams in crowdsensing applications. Our framework leverages lightweight cryptographic techniques like additive secret sharing to enable privacy-preserving streaming truth discovery, continuously producing reliable knowledge over data streams. For monetization of the learned truth, i.e., knowledge, we resort to the emerging blockchain technology and deliver a tailored and full-fledged design, which promises monetization fairness, knowledge confidentiality, and streamlined processing. Extensive experiments on Amazon cloud and Ethereum blockchain demonstrate the practically affordable performance of our design.

Performance Evaluation of Hybrid Crowdsensing and Fixed Sensor Systems for Event Detection in Urban Environments.

Crowdsensing offers a cost-effective way to collect large amounts of environmental sensor data; however, the spatial distribution of crowdsensing sensors can hardly be influenced, as the participants carry the sensors, and, additionally, the quality of the crowdsensed data can vary significantly. Hybrid systems that use mobile users in conjunction with fixed sensors might help to overcome these limitations, as such systems allow assessing the quality of the submitted crowdsensed data and provide sensor values where no crowdsensing data are typically available. In this work, we first used a simulation study to analyze a simple crowdsensing system concerning the detection performance of spatial events to highlight the potential and limitations of a pure crowdsourcing system. The results indicate that even if only a small share of inhabitants participate in crowdsensing, events that have locations correlated with the population density can be easily and quickly detected using such a system. On the contrary, events with uniformly randomly distributed locations are much harder to detect using a simple crowdsensing-based approach. A second evaluation shows that hybrid systems improve the detection probability and time. Finally, we illustrate how to compute the minimum number of fixed sensors for the given detection time thresholds in our exemplary scenario.