Fog data analytics: A taxonomy and process model

Abstract Through the exponential growth of sensors and smart gadgets (collectively referred to as smart devices or Internet of Things (IoT) devices), significant amount of heterogeneous and multi-modal data, termed as Big Data (BD), is being generated. To deal with such BD, we require efficient and effective solutions such as data mining, analytics, and reduction to be deployed at the edge of fog devices on a cloud. Existing research and development efforts generally focus on performing BD analytics overlook the difficulty of facilitating fog data analytics (FDA). In this paper, we discuss the unique nature and complexity of fog data analytics. A detailed taxonomy for FDA is abstracted into a process model. The proposed model addresses various research challenges, such as accessibility, scalability, fog nodes communication, nodal collaboration, heterogeneity, reliability, and quality of service (QoS) requirements. To demonstrate the proposed process model, we present two case studies.