The Voice of 5G & LTE for the Americas

Select a Language: Spanish | Portuguese

Close this search box.

AI and ML Evolve Alongside Wireless Cellular Networks

Chris Pearson, President, 5G Americas (March 2024) – We are living in the age of AI and machine learning, which is touching every industry and platform. It seems like everywhere you turn, AI/ML is popping up in every conversation and news article. It was a major storyline at Mobile World Congress 2024 in Barcelona and I expect it will continue to make additional headlines throughout this year and beyond.

As this is a huge and emerging topic, I’ve broken up this blog series into several articles, which we will publish over the course of the next several weeks. Part I, the article you’re reading, focuses on where we’ve been with AI in 5G networks and how Generative AI is fundamentally different than other AI/ML processes. In Part II, we look at where we’re headed next with Generative AI in network planning and deployment – and spectrum management. Part III will deal with AI in content optimization and delivery, virtualization and Network Slicing, and enhanced user experiences. I’ll look at AI/ML in security enhancements and predictive maintenance in Part IV. And finally, in Part V, will examine AI/ML in training and simulation, as well as innovative services and applications.

It seems that part of every successful technological wave involves a crest of excitement, followed by a trough of disillusionment, and then settling in on a steady, productive gain of success. AI/ML is no different, so it is important for us to separate the hype from the reality and help peel back the onion.

First, it’s important to realize that the use of AI/ML has not been a new thing for wireless cellular networks. This has been a work in progress for several years and 5G Americas has documented its evolution in many of our white papers. Most recently, we covered AI/ML in a few key white papers, including: 3GPP Technology Trends, Energy Efficiency and Sustainability in Mobile Communications Networks, and the State of Mobile Network Evolution.

Indeed, the Third Generation Partnership Project (3GPP) has been working hard to establish to the specifications to further integrate AI/ML into 5G (and soon, 5G-Advanced) networks. Excellent work continues to be completed, as wireless cellular networks are poised to take advantage of some of the latest advances in artificial intelligence.

When it comes traditional uses of AI/ML in wireless cellular networks, efforts have tended to focus on a few key areas where intelligent classification and regression are useful, including:

  • Network Optimization with Self-Organizing Networks (SONs), AI algorithms can dynamically adjust network parameters in real-time, improving performance and efficiency. Additionally, AI has been used in predictive maintenance, where it can predict equipment failures or identify when maintenance is needed before issues arise, minimizing downtime and improving service reliability. In Management Orchestration & Automation, 5G Americas explored the use of AI/ML in the automation and management of wireless cellular networks.
  • Enhanced Performance through traffic prediction and management, where ML models analyze traffic patterns to predict surges in demand and adjust network resources accordingly. It then automates resource allocation bydynamically allocating bandwidth and other network resources to where they’re needed most, optimizing the performance for high-demand applications such as streaming, gaming, and virtual reality. In 5G Edge Automation and Intelligence, we looked at how ML models could help improve the performance 5G edge networks.
  • Improved Security through anomaly detection, AI can monitor network traffic in real-time to detect and respond to unusual patterns that may indicate a security threat, such as DDoS attacks or unauthorized access attempts. In addition, AI and ML can enhance security protocols, including the development of more secure biometric authentication methods and the detection of vulnerabilities in network infrastructure.
  • Network Slicing can be helped by AI, as this network capability allows operators to create multiple virtual networks with different characteristics over a single physical infrastructure. This is essential for supporting a wide range of applications, from IoT devices with low data needs to high-bandwidth applications like 4K video streaming, with their specific requirements for latency, speed, and reliability. In Commercializing 5G Network Slicing, 5G Americas detailed how AI/ML could be used to improving network slicing techniques and apply them for commercial use cases.
  • Enhanced User Experiences. Even without network slicing, AI can potentially be used to analyze network conditions and user behavior to dynamically adjust Quality of Service (QoS) settings, ensuring optimal service levels for various applications and services. AI is used alongside predictive analytics to assess data on user behavior and device performance, to help anticipate user needs and adapt services accordingly, enhancing the user experience.

But with the advent of Generative AI, the application of machine intelligence has been turned on its head. Generative AI is taking the industry by storm. But first, what is Generative AI? And how is it different than “regular” AI and machine learning?

Generative AI distinguishes itself from traditional AI/ML by its focus on creating new, original content that mimics human-generated data, rather than just interpreting or making predictions based on existing data. Utilizing models like Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and transformers, Generative AI learns the underlying distribution of data to produce new instances that could believably be part of the original dataset. This contrasts with traditional AI/ML, which primarily employs discriminative models for tasks such as classification and regression, focusing on understanding patterns and making decisions based on input data. Consequently, Generative AI enables innovative applications in content creation, data augmentation, and simulation, expanding the creative and practical possibilities of AI technology.

Introduced by Ian Goodfellow and his colleagues in 2014, Generative Adversarial Networks (GANs) represent a significant breakthrough in the ability of AI systems to generate realistic images. GANs consist of two neural networks, a generator and a discriminator, that are trained simultaneously through adversarial processes to produce highly realistic images or other types of data.

Around the same time as GANs, Variational Autoencoders (VAEs) were developed as another method for generating data. VAEs are based on the principles of encoding and decoding data, providing a framework for learning deep latent-variable models and generating new data samples.

The introduction of the transformer model by Vaswani et al. in 2017 paved the way for significant advancements in Generative AI for natural language processing. Transformer models, such as GPT (Generative Pretrained Transformer) by OpenAI, introduced in 2018, and its subsequent iterations, have shown remarkable ability to generate coherent and contextually relevant text, making them a cornerstone of modern Generative AI in the domain of natural language.

Generative AI, particularly through large language models (LLMs), offers transformative potential for wireless cellular network operators by enabling sophisticated natural language processing capabilities. For instance, wireless cellular network operators can leverage LLMs to create advanced chatbots for customer support, significantly enhancing the user experience. These chatbots can understand and process customer queries in natural language, providing instant, accurate responses and solutions. One scenario might look something like this:

Scenario: A customer is experiencing connectivity issues with their mobile device.

  • Chatbot Interaction:
    • Customer: “I’m having trouble connecting to the internet on my phone.”
    • Chatbot: “I’m here to help. Let’s try a few steps. Could you first check if your mobile data is turned on? You can find this in your settings under ‘Connections’.”
    • Customer: “It’s on, but it’s still not working.”
    • Chatbot: “Got it. Please try toggling the ‘Airplane Mode’ on and then off. This can sometimes refresh your connection.”
    • Customer: “That worked, thanks!”

Another, less obvious, but perhaps even more powerful example involves large language models (LLMs) significantly transforming the way wireless cellular network operators manage and re-program network operations. By integrating natural language processing capabilities, these models can interpret human commands or queries and translate them into actionable technical instructions or configurations. Here’s how it could work in the context of re-programming network operations:

Scenario: A network operator needs to reconfigure parts of the network to enhance performance, address congestion, or deploy new services.


  • Command Input: The network operator provides a command in natural language, such as “Increase the bandwidth allocation for Area 51 by 20% during peak hours to accommodate increased usage.”
  • Interpretation and Translation: The LLM interprets the command’s intent and translates it into a specific set of network configuration commands or scripts that can be understood by the network management system.
  • Automated Execution: These commands are automatically executed on the relevant network elements, adjusting configurations as instructed without manual intervention.
  • Confirmation and Feedback: The system, through the LLM, generates a confirmation message in natural language, such as “Bandwidth allocation for Area 51 has been successfully increased by 20% during peak hours.”

These two examples are just scratching the surface of what Generative AI can do for wireless cellular networks today. In the next part of the blog series, I will explore the potential impact of Generative AI on 5G and future wireless cellular networks as they related to network planning and deployment, radio access network (RAN) configuration, and spectrum management.


Share this post

Sign up to receive our announcements