Fog Computing for Big Data Analytics in IoT Aided Smart Grid Networks

The recent integration of Internet of Things and Cloud Computing (CC) technologies into a Smart Grid (SG) revolutionizes its operation. The scalable and unlimited Store Compute and Networking (SCN) resources offered by CC enables efficient Big Data Analytics of SG data. However, due to remote location of Cloud Data Centers and congested network traffic, the cloud often gives poor performance for latency and energy critical SG applications. Fog Computing (FC) is thus proposed as a model that distributes the SCN resources at the intermediary devices, termed as Fog Computing Nodes (FCN), viz. network gateways, battery powered servers, access points, etc. By executing application specific logic at those nodes, the FC astonishingly reduces the response time as well as energy consumption of network elements. In this paper, we propose a mathematical framework that explains the P lanning and P lacement of F og computing in smart G rid (PPFG). Basically, the PPFG model is formulated as an Integer Linear Programming problem that determines the optimal location, the capacity and the number of FCNs, towards minimizing the average response delay and energy consumption of network elements. Since this optimization problem is trivially NP-Hard, we solve it using an evolutionary Non-dominated Sorting Genetic Algorithm. By running the model on an exemplary SG network, we demonstrate the operation of proposed PPFG model. In fact, we perform a complete analysis of the obtained Pareto Fronts (PF), in order to better understand the working of design constraints in the PPFG model. The PFs will enable the SG utilities and architectural designers to evaluate the pros and cons of each of the trade-off solutions, leading to intelligent planning, designing and deployment of FC based SG applications.