logo_header
  • Topics
  • Research & Analysis
  • Features & Opinion
  • Webinars & Podcasts
  • Videos
  • Dtw

A new GenAI performance toolkit for cost-efficient network management

The GenAI powered toolkit for network & service management Catalyst combines finely-tuned genAI with a digital twin to reconcile data between systems for smart infrastructure management

Ryan Andrew
25 Jun 2024
A new GenAI performance toolkit for cost-efficient network management

A new GenAI performance toolkit for cost-efficient network management

Commercial context

The advent of FTTX, 5G, IoT, and edge computing technologies, coupled with growing customer demand, has created an urgent need for CSPs to transform how they upgrade, manage, and operate network infrastructure. Current OSS processes often involve costly re-runs, ultimately impairing customer experience, which is unsustainable in an enterprise market increasingly characterized by sophisticated, data-intensive use cases. While the goal of achieving accurate, real-time understanding of operations has long been prized, it must be achieved against the backdrop of rising capital expenditures (CAPEX) for new technologies, necessitating cost reductions and efficiency improvements through operational transformation.

The solution

The GenAI powered toolkit for network & service management Catalyst seeks a means to undertake this transformation by introducing to the market a ‘Generative AI-based Toolkit.’ The toolkit, which encompasses an adaptor agent and large language model (LLM), will simplify and assist with decision-making across operational domains. To do this, it accommodates diverse sources of data (both structured and unstructured) and the various IT systems and network and service elements that produce it.

The data is processed through a finely-tuned generative AI (genAI) model and undergoes a reconciliation process between network elements and the underlying BOSS and OSS systems. It’s then represented as a digital twin using a knowledge graph which simulates network topology and provides CSPs with actionable insights, contextually relevant information, and commands that are directly applicable to solving the problems at hand using an LLM query.

The solution is underpinned by TMF921 Intent Management APIs, and TMF640/TMF641 APIs and draws on the TM Forum’s business process framework and information reference model principles (such as ETOM/SID) and IG1251 AN Reference Architecture. The Catalyst demonstrates, in the form of a proof of concent, the genAI toolkit’s ability to enhance customer experience via network insights, discovery and closed loop service fulfilment. Moreover, by achieving Autonomy Level 4, the Catalyst achieves an average IP/VPN provisioning time of 2 minutes, down from the industry average of 45 minutes.

Application and wider value

The advantages of this approach are substantial, particularly in improving time-to-delivery and time-to-resolution. According to Mohamed Irshad Deen, Deputy Chief Innovation Officer for Sri Lanka Telecom PLC, “we estimate that this generative AI model has the possibility of reducing network operational costs by 20% to 25%, and will be instrumental in reducing operational costs in a market where CSPs are expected to be both more efficient and innovative.”

To facilitate its wider market availability, the project team is developing the project roadmap in a phased manner, validating the model and architectural capabilities for service assurance use cases, and enabling conformance of the model and capabilities as a TM Forum Open Digital Architecture drop-in, with a view to creating a shared genAI platform-as-a-service co-creating environment. This environment will serve as an accelerator tool for onboarding CSPs and their own nuances and complexities in data management and ingest, compute, and development and testing. The environment is designed to enable rapid testing, validation, and onboarding of the toolkit, ensuring that CSPs can move to implementation as quickly as possible.

The eventual goal is to enhance customer experience by lowering the time needed to provision a service, and eventually enable autonomous network and service orchestration. The project marks a significant step forward for the industry as a whole, setting the benchmark for how genAI can be integrated into network and service management. In the long term, the breakthrough in combining a genAI-based toolkit and a digital twin could also assist with the integration of new technologies, helping CSPs maintain their competitive edge in a rapidly evolving tech landscape.

To learn more about how this Catalyst is combining genAI and digital twin technologies to reconcile data between systems for smart infrastructure management, please click here to view the project space on the TM Forum website.

GenAI Powered toolkit for network & service management