D-LBAH: Dynamic Load Balancing Algorithm for HEC-SDN systems

Abstract

With the emergence of new applications that have very low latency (i.e. augmented reality, virtual reality, tactile Internet, etc.), it is very important to propose new solutions to meet these requirements. In addition, with the amount of data generated by these applications in edge computing systems, where resources are limited, the question of automated resource allocation is an important issue for the management of these services. Besides, Edge Computing (EC) systems such as Home Edge Computing (HEC) is one of the architectures which makes it possible to meet these expectations thanks to the proximity of its servers to end-users. HEC is a new architecture of edge computing with three levels of server organization (HEC-MEC-Cloud), that allows us to reduce the communication delay. Unfortunately, these HEC servers that are close to the devices are very limited in terms of resources. In addition, with the amount of data generated by users on these home servers and the current resource management system in this HEC architecture, compliance with the requirements, such as the ultra-low latency (between 1 to 10 ms) and the throughput, for these applications become very difficult especially in certain situations where the network is overload by the traffic from users. In this paper, we propose a dynamically automated resource allocation method in the HEC-SDN architecture. This resource allocation proposal allows the HEC architecture to be integrated with Software Defined Networking (SDN) technology to centralize management and control. In addition, we propose a load balancing technique called Dynamic Load Balancing Algorithm for HEC-SDN systems (D-LBAH) which has a positive effect on reducing latency on the communication between end devices. Thus, the simulation results show that D-LBAH effectively reduces latency to 88.91 %, and improves throughput to 28.48% on the HEC-SDN architecture.