Algorithm design for resource management of time critical applications in an edge-cloud architecture

• Main Author: Shaan, Aryaman
• Other Authors: Arvind Easwaran
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/163588
• Institution: Nanyang Technological University

The advent of edge computing and 5G technology has opened up new possibilities in the cloud computing space. Due to lower wireless latency, the cloud can now cater to applications that are time critical and require quick response times. As such, it can now cater to applications that are safety-critical as well, like autonomous driving cars, robotic surgery, manufacturing etc. This introduces a new challenge to algorithm designers: more specifically, the challenge of resource allocation of deadline-constrained tasks on an edge-cloud architecture. The problem studied in this report was challenging due to the fact that each task needed two different types of resources allocated to it on two different servers. A task needs to first be offloaded to an access point after being allocated communication resources (bandwidth). Next, it needs to be transmitted to a server where it is allocated computational resources in order to be processed. Hence, the problem of allocating a variety of finite resources on different servers to deadline-constrained tasks, while trying to maximise profit, was challenging. This study proposes a well-performing heuristic algorithm to solve this resource allocation problem. It also introduces an approximation algorithm that has polynomial time complexity when the total number of access points and servers is a constant. An additional assumption the approximation algorithm needs is that resources can only be allocated in discrete units. Experiments devised to test the heuristic algorithm under various scenarios have also been shown.