الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 200 |  |  |
| | | from | 200 | | |
Abstract
A smart city exploits the use of Internet-of-Things (IoT) technology to share and process data among smart devices. The core element of IoT is the complex data generated from the massive and diverse interconnected sources at real-time. These data are characterized by the big data characteristics and some additional features related to the IoT nature. Hence, these IoT data features have come with several data processing and analysis challenges.
In this thesis, we first investigated many IoT-based systems at different IoT domains to deduce the main IoT data features and studied their impact on both the big data analytics and software engineering concepts to explore the current challenges research gaps. We then proposed the domain-independent Spatiotemporal Data Reduction (STDR) approach as an IoT-based data reduction approach that considers both the spatial and temporal IoT data features while performing the data reduction process. Next, we expanded the STDR approach by proposing the IoT-based Spatiotemporal Data Fusion (STDF) approach as the first low-level data-in–data-out fusion approach that maintains all IoT data features before data analysis to ensure efficient and effective IoT data processing. This includes maintaining the data expiry, quality, trustworthiness and spatial and temporal IoT data features, in addition to the data volume and velocity. Furthermore, we proposed the Triple Phases Resource Utilized Data Fusion (TPRUDF) as the first IoT-based cost-aware resource utilization approach that optimizes the usage of the processing resources for IoT-based systems. It utilizes three phases of data fusion: (1) data-in – data-out, (2) data-in – feature-out, and (3) feature-in – decision-out. TPRUDF fuses the raw IoT data by maintaining the complex IoT data features using the STDF approach. Then, it fuses the uncorrelated data features using a features extraction technique. Finally, it employs two different resource utilization techniques and decides the optimum utilization results.
from the data analysis point of view, data prediction in specific, we proposed the domain-independent Data Fusion for Data Prediction (DFDP) approach that considers IoT data massive size, faults, spatiotemporality, and freshness, which ensures data prediction accuracy and reliability
All the proposed approaches are validated using different real-world publicly available smart cities datasets. In addition, they are independent from any specific IoT domain, in which they can be integrated to any computing model.