GRAPH-BASED FOG COMPUTING NETWORK MODEL

Abstract

IoT networks generate numerous amounts of data that is then transferred to the cloud for processing. Transferringdata cleansing and parts ofcalculations towards these edge-level networks improves system’s, latency, energy consumption, network bandwidth and computational resources utilization, fault tolerance and thus operational costs. On the other hand, these fog nodes are resource-constrained, have extremely distributed andheterogeneous nature, lack horizontal scalability, and, thus, the vanilla SOAapproach is not applicable to them. Utilization of Software Defined Network (SDN) with task distribution capabilities advocated in this paper addresses these issues. Suggested framework may utilize various routing and data distribution algorithms allowing to build flexible system most relevant for particular use-case. Advocated architecture was evaluated in agent-basedsimulation environment and proved its’ feasibility and performance gains compared to conventional event-stream approach.

Keywords

• software-defined networks
• fog computing
• smart sensors
• iot