Latency-Aware Distributed Resource Provisioning for Deploying IoT Applications at the Edge of the Network

Abstract

With the increased success of Internet of Things (IoT), the conventional centralized cloud computing is encountering severe challenges (e.g., high latency, non-adaptive machine type of communication), that proved insufficient to meet the stringent requirements of IoT applications. Besides requiring fast response time, increased security and privacy, they lack computational resources at the edge of the network. Motivated to solve these challenges, new technologies are driving a trend that distributes the computational resources and shifts the function of centralized cloud computing to the edge. Several edge computing technologies, edge and fog paradigms, originating from different backgrounds have been emerging to overweight these challenges. However, to fully utilize these limited devices, we need advanced resource management techniques. In this paper, we present a novel distributed resource allocation algorithm with the purpose of enabling seamless integration and deployment of different applications in an IoT infrastructure. The algorithm decides: (i) the mapping of an IoT application at the edge of the network; (ii) dynamic migration of parts of the application, such that Service Level Agreement (SLA) is satisfied. Furthermore, we analyze and discuss our approach and the potential to minimize the latency of different IoT applications.