Optimizing Edge Computing Performance using Machine Learning for Low-Latency Applications
M. Karthiga, Jayanthi Paramasivam, P. Vishnu Raja, Sanjay Kumar Suman, L. Bhagyalakshmi, Hitha Poddar
Abstract
Due to increased demands in applications that require low-latency and high performance in edge computing systems, new optimization methods have to be investigated. This paper introduces a new way of the system optimization of edge computing workload associating machine learning (ML) algorithms specific to low-latency applications. Edge computing is also a potentially vital technology, as it will help decrease the latency and enhance the realization of a decision in real-time. Nevertheless, the shortage of resources, variable workloads, and fluctuation of the network can impede its efficiency. Our ML applications include reinforcement learning, deep learning and supervised learning models to predict and optimize task allocation, resource management and data-routing in edge networks. The proposed remedy is capable of optimizing the behavior of the system according to the real-time data, and thus transform the best usage of resources and reduce the response time. The effectiveness of the suggested method in decreasing the latency and providing high throughput in a variety of use cases has been demonstrated through experimentation. Our strategy offers a flexible, dynamic architecture of developing low-latency applications in an edge computing architecture with the high-quality user experience and the satisfaction of performance demands.