Crowd-sensing for smart city applications: towards solving crowd-sensing data challenges by introducing edge and cloud services

Thumbnail Image
Alkhelaiwi, Aseel T.
Journal Title
Journal ISSN
Volume Title
University College Cork
Published Version
Research Projects
Organizational Units
Journal Issue
Crowd-sensing is the ability of a crowd to utilize sensors embedded in mobile devices to sense the surroundings and then send data to a centralized server or the cloud. With crowd-sensing, a wide range of applications have been empowered, such as smart city, healthcare and marketing, of which the smart city is the domain of interest in this research. However, sending a large amount of data to the cloud has introduced several challenges, such as data truthfulness, redundancy, transfer cost, bandwidth consumption and the way data are stored and managed in the cloud. This thesis presents a crowd-sensing architecture for smart city applications. This architecture contains several services that play a key role in solving a number of the challenges listed earlier. Services are distributed between the cloud and public local servers. The local servers are distributed around a city to improve citizens’ quality of life. Services located on public local servers are called edge services and are concerned with trust, the scheduler and compression. Services located in the cloud are known as cloud services and contain a partitioning method along with two reduction techniques: optimization and context extraction. The trust service calculates trust using different factors. Then, if the trust value is above a predefined threshold, data are trusted; otherwise, they are discarded. The scheduler removes redundant data and schedules sending data to the cloud depending on their priority. The compression service compresses single precision floating-point data using two lossless compression algorithms. The partitioning method in the cloud highlights the importance of data entries using time, access rate and singularity factors. Then, based on the output of this method, users can apply optimization and context extraction to optimize data entries and extract important information, respectively. The order in which these services are performed and how they work and communicate are presented. Evaluations and use cases are performed on the mobile, local server and the cloud using Android-based mobile devices and the Amazon EC2 cloud. The results show the effectiveness of the proposed work by meeting predefined requirements, such as reducing the amount of the data transferred.
Crowd-sensing , Cloud computing , Smart city
Alkhelaiwi, A. T. 2019. Crowd-sensing for smart city applications:towards solving crowd-sensing data challenges by introducing edge and cloud services. PhD Thesis, University College Cork.