V2X Network Slicing Prediction with Deep Learning.
Published:
Abstract
The rapid advancement of connected and autonomous vehicles has significantly increased the demand for efficient and reliable Vehicle-to-Everything (V2X) communication systems, which are essential for ensuring real-time data exchange between vehicles, infrastructure, and other entities. Network slicing is a technique that creates multiple virtual networks optimized for specific services, and is important in addressing the diverse Quality of Service (QoS) requirements of V2X communication, such as ultra-low latency, high bandwidth, and reliability. However, traditional cellular networks often struggle to meet these demands due to their static and inflexible architectures. The introduction of the Open Radio Access Network (O-RAN) architecture addresses these challenges by incorporating intelligent controllers, specifically the Near-Real-Time RAN Intelligent Controller (Near- RT RIC), which enhances network slicing through dynamic and adaptive management of network resources. These controllers leverage real-time data and machine learning algorithms to optimize resource allocation, ensuring that each network slice meets its specific QoS requirements even in highly dynamic environments. In this paper, we present a method of network slicing through a deep learning approach within an O-RAN-based xApp to predict and manage network slices in a V2X environment. Our method is able to achieve a 92% accuracy in slice type prediction, demonstrating significant improvements in network performance and resource allocation. This work showcases the potential of combining O-RAN’s intelligent control capabilities with advanced machine learning techniques to meet the stringent demands of dynamic vehicular networks.