Cost-optimal cloudlet placement frameworks over fiber-wireless access networks for low-latency applications

• Published in: Journal of network and computer applications Vol. 138; pp. 27 - 38
• Main Authors: Mondal, Sourav, Das, Goutam, Wong, Elaine
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.07.2019
• Subjects: Cloud computing; Cloudlet; Cost-minimization; Fiber-wireless access networks; Low-latency; Non-linear mixed-integer programming; Cloud computing; Non-linear mixed-integer programming; Fiber-wireless access networks; Cost-minimization; Cloudlet; Low-latency
• ISSN: 1084-8045; 1095-8592
• DOI: 10.1016/j.jnca.2019.04.014

In the recent past, we have observed a trend that edge devices are increasingly accessing computational resources through the Internet, preferably via cloud computing technologies. However, cloud servers fail to meet the desired latency requirements of low-latency applications like augmented reality, cognitive assistance, and Tactile Internet applications, primarily due to the high-transmission latency from edge devices. This issue is resolved by augmenting cloud networks with a cloudlet enabled edge computing solution that creates a three-tier network topology. In this paper, we primarily focus on static cloudlet network planning problem and propose a hybrid cost-optimization framework for optimal cloudlet placement over existing passive optical access networks, subject to capacity and latency constraints. We formulate a mixed integer non-linear program to identify ideal cloudlet placement locations over urban, suburban and rural deployment scenarios. We show that the target latency requirement and the type of deployment scenario plays a major role in determining the amount of computational resources to be installed in the cloudlets. We also show that the percentage of the incremental energy budget of the access networks due to active cloudlet installation is under 18%. Overall, we show that the newly proposed hybrid cloudlet placement framework achieves the most cost optimal solution as compared to the field cloudlet placement framework. •We formulate a mixed-integer non-linear programming model to identify optimal cloudlet placement locations.•We show that the formulated MINLP model for hybrid cloudlet placement architecture aims to achieve a very low end-to-end latency.•We use our framework against urban, suburban and rural scenarios and present a comparative study between field and hybrid cloudlet placement frameworks.