ABSTRACT
In the present scenario, Internet of Things (IoT) driven devices are generating and transmitting huge amounts of data across connected devices or networks. For adequate management of this data, these devices use a wide range of applications and services such as cloud. However, during management, transmission, computation, and other data processing operations via cloud services, many challenges including latency issues, slower response time to time-sensitive applications, and security issues are introduced. Integration of the fog computing model acts like a catalyst providing more efficiency, optimized performance, and bandwidth in processing of massive data being generated. Further reducing network congestion and providing faster response time. The implementation of a fog-based decentralized system along with the merge of the cloud platform introduces resource management and security issues. To overcome these challenges, machine learning techniques and models are deployed in fog-based applications, which bring along secure models designed for fog systems to detect and respond to suspicious behavior or attacks, providing efficient models for resource management and decision making, as well as mitigation of latency issues.
The era of fog systems adopting machine learning techniques extends functionalities and services to a wide range of fog-based IoT applications that are based on time-sensitive decision-making such as Smart Healthcare systems, location-based services and privacy of data such as Vehicular network, energy consumption and accurate reading such as Smart Grid, Smart Meter. Machine learning provides models based on neural networks, clustering, support vector machine (SVM), linear regression, Markov model and many other techniques to enhance accuracy, analysis and security of data being generated, processed by IoT applications are discussed in detail in this chapter.
