Distributed Orchestration in Large-Scale IoT Systems

Abstract

With the growing popularity of smart things and the omnipresence of wireless communications, the Internet continues to revolutionize from the Internet of hosts, the Internet of People, to the Internet of Things (IoT). Intelligent and scalable orchestration of massive IoT objects using a multitier architecture are critical to embrace the vision of IoT. This paper presents our vision and our initial development of a distributed orchestration framework, called ISYMPHONY, with the ultimate-goal of scaling real-time and on-demand IoT services provisioning in large scale IoT systems, while guaranteeing quality of service with respect to performance, availability and security. This paper makes two original contributions. First, we present a distributed orchestration architecture that enables edge devices and edge clients running on top of edge devices to contribute to the IoT computing tasks based on their computing and communication capacities. A main idea behind our distributed IoT orchestration architecture is the intelligent partition of a real time IoT computing task into an optimal coordination of server-side processing and IoT object side processing. Second, a set of optimization techniques are introduced to limit the amount of computations handled by the edge clients and to enhance the overall performance and resource utilization of the ISYMPHONY system. Our initial experimental results show that ISYMPHONY can lead to significant savings in terms of server load and high accuracy by leveraging and coordinating edge client processing capabilities, compared to the solutions relying solely on server level processing for real time IoT services provisioning.