A Framework of Fog Computing: Architecture, Challenges, and Optimization

TLDR

Fog computing is an emerging distributed platform that brings computation close to data sources, reducing latency and cost, and is especially valuable for latency‑sensitive and mission‑intensive IoT services. The paper defines and compares fog computing architecture to other technologies, then investigates an application scenario by formulating and solving a resource‑optimization problem with a genetic algorithm. It presents a resource‑allocation framework that balances latency, reliability, fault tolerance, and privacy, and formulates the optimization problem for the application scenario, solving it with a genetic algorithm. The analysis yields insights into the scalability of fog computing systems.

Abstract

Fog computing (FC) is an emerging distributed computing platform aimed at bringing computation close to its data sources, which can reduce the latency and cost of delivering data to a remote cloud. This feature and related advantages are desirable for many Internet-of-Things applications, especially latency sensitive and mission intensive services. With comparisons to other computing technologies, the definition and architecture of FC are presented in this paper. The framework of resource allocation for latency reduction combined with reliability, fault tolerance, privacy, and underlying optimization problems are also discussed. We then investigate an application scenario and conduct resource optimization by formulating the optimization problem and solving it via a genetic algorithm. The resulting analysis generates some important insights on the scalability of the FC systems.