Seven AI Concepts Shaping Network Intelligence

AI has become so deeply embedded in our everyday working lives that it is no longer limited to data science teams or research labs. In telecoms, AI now plays a central role in network planning, optimisation, assurance and automation. As a result, the industry is rapidly absorbing a growing set of AI-related terms and concepts, many of which are directly relevant to how networks are evolving towards higher levels of autonomy.

I recently came across the video embedded below, which provides clear explanations of seven AI terms that are becoming increasingly important in the context of network intelligence and autonomous networks. Some of these concepts are already being applied in operational networks today, while others point clearly towards the direction of travel for AI-native 5G Advanced and 6G systems.

The video begins with Agentic AI, a concept that aligns closely with the telecom industry’s vision for autonomous networks as defined in 3GPP. Unlike traditional AI models that respond to a single prompt, AI agents can perceive their environment, reason about next steps, take action and observe the outcome in a continuous loop. In practical terms, this maps well to closed-loop automation use cases such as self-healing, energy optimisation, dynamic resource allocation and intent-driven network management.

Closely related are Large Reasoning Models, which are designed to work through problems step by step rather than producing an immediate response. This capability is particularly relevant for telecom networks, where decisions often span multiple domains, layers and vendors. As AI systems take on greater responsibility for operational decisions, reasoning-based models become essential for safe and explainable automation.

The video then moves to more foundational enablers, starting with Vector Databases. Telecom networks generate vast volumes of unstructured data, including logs, alarms, performance metrics, configuration data and documentation. Vector databases allow this information to be searched and correlated based on semantic meaning rather than simple keywords, enabling more context-aware and intelligent AI systems.

This naturally leads to Retrieval-Augmented Generation (RAG), which is already gaining traction in telecom operations. By combining large language models with operator-specific data sources such as standards, network documentation or operational procedures, RAG helps ground AI outputs in trusted information. This is particularly important in network operations, where accuracy and reliability are critical.

Another important concept discussed is the Model Context Protocol (MCP), which addresses how AI models interact with external tools and systems. For telecom operators, standardised mechanisms for AI access to network management systems, data platforms and orchestration tools could significantly simplify integration and accelerate the deployment of AI-driven automation across the network lifecycle.

The video also touches on Mixture of Experts (MoE) models, which provide a more efficient way to scale AI by activating only the parts of a model needed for a specific task. This approach is especially relevant for telecom use cases where compute efficiency, latency and energy consumption are key constraints, particularly as AI capabilities move closer to the edge of the network.

Finally, the video briefly discusses Artificial Superintelligence (ASI). While ASI remains theoretical, it is often referenced in long-term discussions around AI evolution. For the telecom industry, it serves as a reminder of the rapid pace of change and the importance of governance, trust and control as networks become increasingly autonomous and software-driven.

Overall, this video offers a useful technical refresher on AI concepts that are already shaping the development of network intelligence, autonomous operations and AI-native architectures. For anyone working on 5G Advanced, autonomous networks or early 6G thinking, these are terms that are quickly becoming part of the industry’s everyday vocabulary.

Related Posts and Articles:

• Cambridge Wireless - Agentic AI Security: Navigating Trust, Autonomy and Resilience
• The 3G4G Blog - AI/ML in 3GPP: Progress, Challenges, and the Road to 6G
• Operator Watch Blog: Automation and Data Driven Network Optimization in Swisscom’s Mobile Strategy
• Operator Watch Blog: Telefonica’s Journey Towards End-to-End Autonomous Networks
• Operator Watch Blog: The Power of AI in NTT Docomo’s 5G Journey
• Operator Watch Blog: Building and Scaling AI the Cloud Native Way at Singtel
• Operator Watch Blog: How AI Is Reshaping Network Operations at Deutsche Telekom
• Free 6G Training: Agentic AI and the Path to Hyper-Personalised 6G Networks
• Free 6G Training - Pioneering 6G with AI: Reflections on NVIDIA’s Keynote at the Brooklyn 6G Summit 2025
• Free 6G Training - AI for Wireless: Evaluating Neural Receivers with Test and Measurement and Digital Twins
• The 3G4G Blog: AIoT and A-IoT 

Realizing Zero Trust Architecture for 5G Networks

Over the last couple of years, I keep on coming across Zero-Trust Architecture (ZTA). A simple way to explain is that the standard model of security is known as perimeter security model, where everything within the perimeter can be trusted. In zero-trust (ZT) model, no assumptions is made about trustworthiness and hence it is also sometimes known as perimeterless security model.

This short video from IBM clearly explains what ZT means:

This blog post from Palo Alto Networks also clearly explains ZT:

By definition, Zero Trust is a strategic approach to cybersecurity that secures an organization by eliminating implicit trust and continuously validating every stage of a digital interaction. Zero Trust for 5G removes implicit trust regardless of what the situation is, who the user is, where the user is or what application they are trying to access.

The impact of Zero Trust on network security specifically protects the security of sensitive data and critical applications by leveraging network segmentation, preventing lateral movement, providing Layer 7 threat prevention and simplifying granular user-access controls. Where traditional security models operate under the assumption that everything inside an organization’s perimeter can be trusted, the Zero Trust model recognizes that trust is a vulnerability.

In short, Zero Trust for 5G presents an opportunity for service providers, enterprises and organizations to re-think how users, applications and infrastructure are secured in a way that is scalable and sustainable for modern cloud, SDN-based environments and open-sourced 5G networks. Delivering the Zero Trust Enterprise means taking Zero Trust principles, making them actionable and effectively rebuilding security to keep pace with digital transformation. 

A research paper looking at Intelligent ZTA (i-ZTA) provides an interesting approach to security in 5G and beyond. The paper can be downloaded from here. The abstract states:

While network virtualization, software-defined networking (SDN), and service-based architectures (SBA) are key enablers of 5G networks, operating in an untrusted environment has also become a key feature of the networks. Further, seamless connectivity to a high volume of devices in multi-radio access technology (RAT) has broadened the attack surface on information infrastructure. Network assurance in a dynamic untrusted environment calls for revolutionary architectures beyond existing static security frameworks. This paper presents the architectural design of an i-ZTA upon which modern artificial intelligence (AI) algorithms can be developed to provide information security in untrusted networks. We introduce key ZT principles as real-time Monitoring of the security state of network assets, Evaluating the risk of individual access requests, and Deciding on access authorization using a dynamic trust algorithm, called MED components. The envisioned architecture adopts an SBA-based design, similar to the 3GPP specification of 5G networks, by leveraging the open radio access network (O-RAN) architecture with appropriate real-time engines and network interfaces for collecting necessary machine learning data. The i-ZTA is also expected to exploit the multi-access edge computing (MEC) technology of 5G as a key enabler of intelligent MED components for resource-constraint devices.

Ericsson Technology Review covered Zero Trust in 5G Networks in one of their issues. Quoting from the article:

The 3GPP 5G standards define relevant network security features supporting a zero trust approach in the three domains: network access security, network domain security and service-based architecture (SBA) domain security. 

The network access security features provide users with secure access to services through the device (mobile phone or connected IoT device) and protect against attacks on the air interface between the device and the radio node. Network domain security includes features that enable nodes to securely exchange signaling data and user data, for example, between radio and core network functions (NFs).

The 5G SBA is built on web technology and web protocols to enable flexible and scalable deployments using virtualization and container technologies and cloud-based processing platforms. SBA domain security specifies the mechanism for secure communication between NFs within the serving network domain and with other network domains. 

While the new requirements and functionality introduced in the 5G specifications are already aligned with many of the zero trust tenets. It is already evident, however, that further technology development, standardization and implementation are needed in areas such as policy frameworks, security monitoring and trust evaluation to support the adoption of zero trust architecture in new telecom environments that are distributed, open, multi-vendor and/or virtualized.

While various technologies can support organizations in adhering to the guiding principles of zero trust as part of their total active defense strategy, it is important to remember that technology alone will never be sufficient to realize the full potential of zero trust. Successful implementation of a network based on zero trust principles requires the concurrent implementation of information security processes, policies and best practices, as well as the presence of knowledgeable security staff. Regardless of where a CSP is in its transition toward a zero trust architecture, the three pillars of people, processes and technology will continue to be the foundation of a robust security architecture.

Related Posts:

• 3G4G: 5G Security Overview by Mpirical
• The 3G4G Blog: Key Technology Aspects of 5G Security by Rohde & Schwarz
• The 3G4G Blog: Bug hunting in 5G Networks and Devices
• The 3G4G Blog: Impact of 5G on Lawful Interception and Law Enforcement.
• The 3G4G Blog: AT&T Cybersecurity Experts Provide 5G Security Overview
• The 3G4G Blog: Everything you need to know about 5G Security
• The 3G4G Blog: Nokia Lectures in Collaboration with Bangalore University
• The 3G4G Blog: Lawful Intercept in 5G Networks
• The 3G4G Blog: 5G Roaming with SEPP (Security Edge Protection Proxy)