Fog Computing Support for Internet of Things Applications

With more devices on-board the Internet every day, there is a constant drive to balance Quality of Service (QoS) with an efficient use of resources. At present, the Internet of Things (IoT) applications are entirely hosted in the cloud. With emerging ‘smart’ scenarios in verticals such as dairy farming, health, home, mobility, etc., the real-time communication delay from the cloud platform necessitates the need to use computing platforms closer to the data source. While a traditional centralized cloud approach has led the path towards a pivotal revolution in modern-day computing, the emerging IoT era gave way to its own range of applications demanding a lower response time, efficient network usage, and improved data protection, to name a few. In this age of IoT, the devices along the things-to-cloud continuum present a unique opportunity to additionally serve as computing hubs. Termed fog computing, this paradigm can be used to host applications and process data closer to the source. However, these intermediate devices are usually resource constrained in nature, and are thus limited in computational flexibility. This paradigm shift towards fog computing brings up a challenge of using these intermediary computing resources efficiently to host application(s) and serve as additional computational resources without affecting their primary functionality. The research presented in this work addresses these demands and challenges, and presents how to use the fog computational platform to support these requirements. It presents a set of tools, algorithms, approaches and methodology of developing and deploying these emerging IoT applications while leveraging the fog computing paradigm. With extracting knowledge from the generated data being one of the prime objectives of IoT deployments, this work also presents how the data analytics computing operations can be decomposed to run on these resource-constrained devices without affecting their fundamental operation.