Distributed Approach to Fog Computing with Auction Method

Abstract

Fog Computing models are discussed to efficiently realize the IoT (Internet of Thing). In our previous studies, types of the TBFC (Tree-Based Fog Computing) models are proposed to reduce the electric energy consumption and execution time of fog nodes and servers to make it fault-tolerant. Here, fog nodes support application processes to calculate output data on sensor data and are hierarchically structured in a tree. Here, a root node shows a cloud of servers and leaf nodes indicate edge nodes which communicate with sensors and actuators. Each fog node calculates output data on input data received from child nodes and sends the output data to a parent node. A node $f_2$ which can process the output data of a fog node $f_1$ is a target fog node of the node $f_1$. Since the tree structure is static, if some fog node is faulty or heavily loaded, the tree structure has to be changed so that every output data from child nodes of the faulty node can be processed by another target node. In this paper, we propose a dynamic network-based fog computing (DNFC) model. Here, one or more than one target fog node for each fog node is dynamically selected in a set of possible target nodes in the auction among the nodes. We present an auction method among fog nodes.