Generative AI for Telecom, Understanding the potential beyond ChatGPT

Pouring intelligence in telecom networks is not a new concept. AI and ML are being leveraged in 4G and 5G network architectures, especially to achieve the vision of autonomous networks for self-healing and zero-touch service/operations management. AI/ML algorithms and closed-loop capabilities have been introduced at the network management layer for each network domain and management/orchestration layer (resource & service layer).

Several standardization bodies like ETSI, 3GPP, ONAP, and TM Forum have defined the architecture, use cases, and roadmap for achieving this. The focus of this automation remains on achieving the higher intelligent use of network resources, reducing operations costs, and improving the Quality of Service (QoS).

Within this context, Generative AI can be leveraged for network optimization, predictive maintenance, enhancing customer service, personalization, and network planning.

Various ISVs have introduced intent-based solutions for Self-organizing network solutions (SON) to adjust, reconfigure, and optimize the network functions and parameters according to particular network conditions and user demands.

Moving from self-organizing networks of 4G/5G to self-evolving networks for 6G is a paradigm change in which networks will perform self-provisioning, self–configuration, self-optimization, and self–healing via real-time AI on network parameters.

It requires an AI-embedded architecture to realize the total network intelligence for autonomous sensing, autonomous decision-making, and autonomous control without or with minimal human interaction, focusing on improving the quality of experience for the end users.

The complexity will increase with the seamless autonomous requirements for an integrated network evolving network integrated between land, sea, space, and air connectivity domains.

From different AI techniques and machine learning algorithms at different levels in the network architecture, LLM is a Generative AI technique. It is envisaged to be of great utility to achieve self-evolving networks. LLM is a type of natural language machine learning model in AI, used to analyze the patterns in large data sets and generate new content based on the pattern.

• LLM can enable multiple tasks within autonomous sensing by using a pre-trained model to perform beamforming, power allocation, and so on.
• With techniques like GAN, LLM can be utilized for generating 3D images from text or vice-versa. The utility of this particular feature is immense in real-time network optimization use cases.
• Beyond self-adaptation and self-optimization; LLM will also be leveraged in designing, planning, deploying, configuring, and operating the networks.
• LLMs can be used to identify the optimum configuration of the network based on the initial design requirements, and user intents. For instance, user intent like “reduce the network energy consumption by X%” can be converted into actions of manipulating the power on different sub-carriers and this requires LLM to be grounded to the real network.

To conclude, generative AI and LLM in particular have great potential to transform the way network management is carried out in 4G/5G architecture and can assist in achieving the KPIs of 6G architecture at various levels of the network architecture.

About the Author:

Dr. Jigna Srivastava, founder of DT Simplified, is a renowned thought leader in the space of telco digital transformation. She brings more than two decades of industry and consulting experience. Her expertise is in devising end-to-end digital strategies and enterprise architecture for telcos spanning across IT, Digital B/OSS, Cloud, and Network Automation. She has been instrumental in defining and driving the adoption of digital strategies, and building digital architectures, and has been instrumental in large transformation programs to drive innovative digital business models. She has also devised the DT Simplified® framework, which is an extension of her Doctoral research model outlining the “Antecedents of digital transformation and their impact on the business performance” for a successful digital transformation of telecom services providers. She also teaches subjects like Information Systems Management, Business Intelligence, Data Analytics and Digital Technologies as a Visiting Professor in leading post graduate management colleges/universities in Mumbai, India.