Fog Based Architectures for IoT: Survey on Motivations, Challenges and Solution Perspectives

Abstract

Advent of New age applications encompassing IoT, 5G wireless systems, Artificial Intelligence, real time analytics and stream mining has added new challenges to how cloud computing paradigm handles computing, storage and bandwidth. Many of the applications from the aforesaid domains are time critical and hence have stringent latency requirements and Network bandwidth constraints. It is impractical to depend on the cloud for processing the voluminous amount of data produced by these applications. Hence change in the computing- storage and Networking architectures of these cloud based systems is inevitable. The emerging “Fog Computing” paradigm promises to address the challenges specified above. Fog moves significant amount of computing, storage, communication and Networking away from cloud and performs the same in proximity to end user, thereby addressing the latency and bandwidth requirements related issues. Hence, Fog architecture enhances the distributed nature of an application and adds agility and efficiency to the applications by reducing the dependency on Cloud. However, this emerging Fog computing paradigm also poses its own set of challenges which must be understood and addressed in order to provide a viable framework of computing. This paper presents a thorough study on motivations behind the fog computing paradigm and various challenges in proposing a viable fog computing architecture for the resource intensive and time critical applications. Number of existing solution perspectives to address these challenges is studied and their strengths, benefits and drawbacks are analyzed. Also, scope for future research in this domain are highlighted.