A Clustering-Based Approach to Efficient Resource Allocation in Fog Computing

Abstract

Fog computing, which provides low-latency computing services at the network edge, is an enabler for the next generation Internet of Things (IoT) systems. In scenarios such as smart cities, multiple applications are simultaneously deployed and distributed across the Cloud and fog nodes, offering various IoT-based services. Moreover, each application has its own quality of service (QoS) and resource requirements that must be met. Appropriate resource allocation mechanisms are needed to determine which fog node or group of nodes can host the services of a given application. A critical challenge is how to select fog nodes for resource allocation in order to maximize fog resources utilization and minimize service latency, while satisfying QoS requirements of the application. This paper is aimed to address this challenge through a two-phase QoS-aware resource allocation scheme. Firstly, in the layering phase, we assume a hierarchical architecture for fog nodes—organizing heterogeneous nodes into a multi-layered hierarchy based on node resources capacity and network characteristics. Layering facilitates finding fog node(s) based on application requirements, and improves resource management. In the second phase, the fog nodes are grouped to facilitate resource pooling and reducing delay in service provisioning. We use the Agglomerative Hierarchical Clustering algorithm for classifying fog nodes. This helps selecting those fog node(s) in a fog layer with which the latency will be minimized. We evaluate the proposed approach through simulation. The evaluation results show that the proposed approach promises high application acceptance rate (80% on average), and reduces considerably the application placement time.