AI-Driven Network Slicing in 5G and Beyond: Unlocking Intelligent Connectivity

The introduction of 5G network slicing allows operators to create multiple virtual networks within a single physical infrastructure. However, manual configuration and management of these slices can be complex and inefficient. To tackle these issues, artificial intelligence (AI) is now being integrated into network slicing orchestration, enabling the creation of autonomous, self-optimising and dynamic network slices that adapt to demands in real time.

With 5G connectivity being rolled out globally and 6G research accelerating, AI-driven network slicing will become a cornerstone of future connectivity. This article explores how AI is redefining network slicing, the benefits for next-generation networks, and the challenges that still need to be  overcome.

Network slicing allows operators to partition a single physical network into multiple independent virtual networks, each customised for specific use cases. AI enhances this process by:

• Predicting network demand and adjusting resource allocation;
• Optimising quality of service (QoS) for latency-sensitive applications;
• Enforcing security policies based on AI-driven threat detection.

1. AI-Powered Network Slice Orchestration

Traditional network slicing relies on predefined policies and rule-based algorithms, which may not respond effectively to real-time network fluctuations. AI-driven network slice orchestration, brings several benefits, including:

• Self-adaptive network slices that automatically allocate bandwidth based on AI-predicted demand;
• Intent-based networking, where AI interprets user or application requirements and configures slices accordingly;
• AI-enhanced service level agreement (SLA) compliance, ensuring optimal resource utilisation while meeting contractual performance guarantees.

2. AI for Predictive and Proactive Slice Management

AI enables predictive analytics for network slicing, enabling  operators to accurately forecast network conditions, including:

• Traffic congestion and bandwidth needs, ensuring seamless service continuity;
• Potential network failures or degradation, which will then prompt preemptive corrective action to be taken;
• Energy consumption levels, adjusting power use based on demand.

These capabilities reduce the need for manual intervention, increase network efficiency and improve the overall user experience.

3. AI and Multi-Domain Slicing for Cross-Network Services

As 5G rollout continues and 6G connectivity develops, network slicing will no longer be limited to a single operator’s infrastructure. AI plays a crucial role in multi-domain slicing, where slices span several layers, including:

• Public and private 5G networks used in enterprises, factories and smart cities;
• Terrestrial and satellite communications to provide global connectivity;
• Cloud and edge computing environments for distributed applications.

Autonomous Vehicles and Smart Transportation

• AI-powered network slicing enables ultra-low latency connectivity for vehicle-to-everything (V2X) communications;
• Predictive AI models allocate resources for traffic control systems, minimising congestion and accidents.

Industrial Automation and Smart Factories

• Robotic control and machine learning-driven automation in Industry 4.0 environments;
• AI-based fault prediction and self-healing mechanisms for industrial networks.

Healthcare and Remote Surgeries

• AI-optimised network slices enable remote surgeries to be performed in real time and with high precision;
• Secure and ultra-reliable low-latency communication (URLLC) for telemedicine and connected healthcare devices.

Cloud Gaming and Augmented Reality (AR)/Virtual Reality (VR)

• AI adjusts latency and bandwidth based on user location and network conditions;
• Optimised resource management for real-time rendering and immersive experiences.

1. Interoperability and Standardisation Issues

• Multiple vendors and operators use different architectures, emphasising the need for  AI-driven interoperability frameworks;
• Efforts by ETSI, 3GPP and the GSMA to standardise AI-powered network slicing are still in development.

2. Security and Privacy Concerns

• AI-driven automation introduces new attack vectors into network environments, which requires continuous threat monitoring;
• Analysis of  real-time network and user data can result in privacy risks.

3. Computational Demands and Energy Consumption

• AI models require significant computational power, increasing hardware and energy costs;
• Optimising AI efficiency through edge AI and energy-aware algorithms is an ongoing challenge.

As research into 6G develops, AI-driven network slicing will evolve in various ways, including:

• Integration with quantum AI, enhancing network performance and security;
• Holographic communication support, which will require dynamic AI-powered bandwidth allocation;
• AI-native and intent-based network slicing, where AI autonomously configures network slices without the need for human intervention.