Network Slice type Prediction

The proliferation of 5G and LTE networks has introduced the concept of network slicing to meet diverse service requirements across different application domains. This study addresses the classification of network slice types by analyzing features such as application use case, network technology category, supported communication standards, temporal usage patterns, and quality of service metrics including guaranteed bit rate, packet loss rate, and latency. These attributes collectively reflect user demand profiles and network conditions that guide slice allocation decisions. To effectively predict slice types, this work employs a combination of machine learning and deep learning models, including Light Gradient Boosting Machine (LightGBM), Artificial Neural Networks (ANN), and Convolutional Neural Networks (CNN). The models are trained on structured network data to distinguish between slice categories based on service-level characteristics. Experimental results demonstrate that hybrid and deep models can accurately classify slice types, supporting intelligent network management and enhanced resource allocation.

Keywords: 5G, network slicing, slice type prediction, LightGBM, ANN, CNN, QoS classification, AI.