Scholarlink

<p>With the advent of 5G, the healthcare industry is experiencing a transformative shift, enabling unprecedented improvements in clinical outcomes through next-generation connectivity solutions. The integration of diverse technologies, such as AR/VR, 3D cameras, IoT devices, and wearable, requires seamless and reliable communication. In this dynamic environment, 5G-driven Network Slicing (NS) plays a pivotal role by enabling efficient management of network resources. NS allows for the division of a single physical network into multiple virtual and isolated slices, each tailored to meet specific traffic types and Quality of Service (QoS) requirements. Efficient Traffic Classification (TC) is crucial to optimize resource allocation across slices. This study evaluates the efficacy of Machine Learning (ML) based TC methods within three primary 5G service categories: Enhanced Mobile Broadband (eMBB), Massive Machine-Type Communications (mMTC), and Ultra-Reliable Low-Latency Communications (uRLLC). We focus on classifying traffic for services such as tele-monitoring, tele-surgery, tele-consultation, and connected ambulances. Due to the absence of a dedicated dataset, we created a novel synthetic healthcare dataset (HealthNetSynthDataset). We employ a two-step approach: (1) ML models using Supervised Learning (SVM, DT, RF, Naive Bayes, and KNN) and Unsupervised Learning (K-means, GMM, and heuristic clustering); (2) we evaluate and benchmark the performance, focusing on the accuracy metrics, and conclude that SVM outperformed other classification approaches. To the best of our knowledge, and evaluate both supervised and unsupervised approaches in a single effort. This work serves as the foundation for the selection of the most adoptable approach to the efficient deployment of healthcare-specific network slicing.</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

Effective Healthcare Traffic Management Through Machine Learning-Enabled Network Slicing

  • 期刊论文

  • 发表日期：

    2025年10月27日

查看全文
  • Monika Dubey,
  • Ashutosh Kumar Singh,
  • Aditya Bhushan,
  • Richa Mishra

摘要

With the advent of 5G, the healthcare industry is experiencing a transformative shift, enabling unprecedented improvements in clinical outcomes through next-generation connectivity solutions. The integration of diverse technologies, such as AR/VR, 3D cameras, IoT devices, and wearable, requires seamless and reliable communication. In this dynamic environment, 5G-driven Network Slicing (NS) plays a pivotal role by enabling efficient management of network resources. NS allows for the division of a single physical network into multiple virtual and isolated slices, each tailored to meet specific traffic types and Quality of Service (QoS) requirements. Efficient Traffic Classification (TC) is crucial to optimize resource allocation across slices. This study evaluates the efficacy of Machine Learning (ML) based TC methods within three primary 5G service categories: Enhanced Mobile Broadband (eMBB), Massive Machine-Type Communications (mMTC), and Ultra-Reliable Low-Latency Communications (uRLLC). We focus on classifying traffic for services such as tele-monitoring, tele-surgery, tele-consultation, and connected ambulances. Due to the absence of a dedicated dataset, we created a novel synthetic healthcare dataset (HealthNetSynthDataset). We employ a two-step approach: (1) ML models using Supervised Learning (SVM, DT, RF, Naive Bayes, and KNN) and Unsupervised Learning (K-means, GMM, and heuristic clustering); (2) we evaluate and benchmark the performance, focusing on the accuracy metrics, and conclude that SVM outperformed other classification approaches. To the best of our knowledge, and evaluate both supervised and unsupervised approaches in a single effort. This work serves as the foundation for the selection of the most adoptable approach to the efficient deployment of healthcare-specific network slicing.