Mohamed A. Zohdy, Ph.D., Chair, Hua Ming, Ph.D., Co-Chair, Atiqul Mollah, Ph.D., Richard Olawoyin, Ph.D., William Solomonson, Ph.D..IBRAHAIM RASHED ALRASHEADI2022-06-052020-01-062022-06-0580517https://drepo.sdl.edu.sa/handle/20.500.14154/67208In recent years, with the widespread use of the Internet of Things (IoT) technologies for building smart cities based on cloud computing, the number of zero-day attacks has been exponentially increasing. This is due to the highly dynamic and heterogeneous IoT protocols for wireless data transmission through edge devices. Among smart city applications, the concept of smart healthcare is needed for emergencies, which requires availability and preserving sensitive healthcare information in real-time for remote monitoring patients. Remote monitoring provides very significant sensitive data, collected by wearable sensors. These sensors are vulnerable to attack. Many centralized-based attack detection techniques have been introduced to detect malicious activities in IoT environments. However, those techniques have suffered from many challenges, including high bandwidth consumption and latency to satisfy IoT requirements. This dissertation proposes an efficient framework called a fog-based attack detection (FBAD) that utilizes an ensemble of online sequential extreme learning machine (EOS-ELM) methods for efficiently detecting malicious activities. This research indicates a high-level overview of the designed efficient framework for deploying the distributed attack detection technique in fog nodes for remote patient monitoring through future smart cities. The objectives of this framework are more sufficient and effective due to high accuracy, scalability, and low latency, as it is closer to the IoT devices at the network edge. Furthermore, the FBAD is evaluated using two benchmark datasets and compares the performance of this framework with other existing approaches, including ELM and OS-ELM. The results of the experiments demonstrate that distributed architecture outperforms centralized architecture in terms of the classification accuracy and detection time.121enThesis