Multi-Tier Fog Architecture: A New Delay-Tolerant Network for IoT Data Processing

In order to gather the full profits of Internet of Things (IoT) technologies, it will be necessary to provide efficient networking and computing infrastructure to support low latency and fast response for IoT applications. In this paper, we introduce a new fog architecture for IoT applications. We propose to deploy servers at the network fog level as a tree hierarchy, to efficiently use the cloud resources to serve the peak loads from devices. To schedule different devices demands, we develop an optimal workload placement method by solving a mixed nonlinear integer programming (MNIP). Then, the optimal solution is aggregated over different tiers using the Simulated Annealing Algorithm (SAA) to find out the optimal allocation using numerical iterations. The advantage of the proposed architecture is proved over different performance metrics and trough a probabilistic model and an analytic comparison.