Scholarlink

<p>Network congestion remains a critical issue in Small and Medium Enterprise (SME) networks, often leading to degraded service quality and operational inefficiencies. This study introduces a novel system that integrates machine learning-based classification with Quality of Service (QoS) policy enforcement to optimize bandwidth allocation and application prioritization. By employing a custom-labeled dataset derived from real network traffic and evaluating multiple classifiers, the Decision Tree algorithm was found most effective for classifying network priorities. Based on the classification report, the model demonstrated high performance across four traffic priority categories, achieving excellent precision, recall, and F1-scores particularly in the very low and high categories, where F1-scores exceeded 92%. The low and mid categories also performed strongly, with F1-scores of 89% and 91%, respectively. The system is implemented via a web-based interface that automates QoS configuration for MikroTik routers. Results show significant improvements in network performance, particularly in reducing latency, jitter, and packet loss. This work contributes a practical and automated solution for enhancing network reliability in SMEs.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimizing internet bandwidth allocation and service prioritization in SME networks using machine learning algorithm

  • 期刊论文

  • 发表日期：

    2025年8月24日

查看全文
  • Jimvy P. Salise,
  • Jay Noel N. Rojo

摘要

Network congestion remains a critical issue in Small and Medium Enterprise (SME) networks, often leading to degraded service quality and operational inefficiencies. This study introduces a novel system that integrates machine learning-based classification with Quality of Service (QoS) policy enforcement to optimize bandwidth allocation and application prioritization. By employing a custom-labeled dataset derived from real network traffic and evaluating multiple classifiers, the Decision Tree algorithm was found most effective for classifying network priorities. Based on the classification report, the model demonstrated high performance across four traffic priority categories, achieving excellent precision, recall, and F1-scores particularly in the very low and high categories, where F1-scores exceeded 92%. The low and mid categories also performed strongly, with F1-scores of 89% and 91%, respectively. The system is implemented via a web-based interface that automates QoS configuration for MikroTik routers. Results show significant improvements in network performance, particularly in reducing latency, jitter, and packet loss. This work contributes a practical and automated solution for enhancing network reliability in SMEs.