How is Fujitsu using accelerated Root Cause Analysis for intelligent apps?
What’s the difference between AI, ML, and Generative AI?
Artificial Intelligence (AI) is the overarching technology space. And there are multiple dimensions within AI that use Machine Learning (ML) and Gen (Generative) AI. AI leverages technology to learn how humans learn and applies that to application spaces in a multitude of industries. ML is a subset within AI, and it’s focused specifically on models that evolve with learning and can apply that learning in different situations and application spaces. ML is very focused on use cases with modeling techniques. Commonly, you’ll hear about unsupervised and supervised learning. Within this, you’re talking about a model, an unsupervised space that can learn on its own. It doesn’t need any human interaction or a feedback loop. In supervised learning for ML, we’re talking about models that need the human feedback loop to validate what it’s learning. Generative AI is typically focused on reproducing content, such as text, video, and audio. And with the tools we see out there, it leverages a vast data set, which is where it will go to learn connectivity between text, for example, and reproduce that text, based on what it’s learned. Generative AI can also leverage techniques of ML models to learn in both a supervised and unsupervised fashion to produce and replicate data. We often see solutions that are built with Generative AI and ML models. When combined, they can be very powerful solutions in application spaces. At Fujitsu, we have combined AI, ML and Generative AI to create powerful network software like accelerated Root Cause Analysis.
What is accelerated Root Cause Analysis?
Accelerated Root Cause Analysis (sometimes known as aRCA) application is a tool that leverages AI and ML technology to produce faster mean time to repair. It leverages ML modeling and Generative AI to understand the network, as well as events and faults that take place in real time to produce an output to the user. Traditionally, operators have to use manual policies and correlation techniques, but there’s still an investigation that has to take place. The operator has to go into Element Management System (EMS) platforms and perform an inspection, which takes more time to resolve the issue. With accelerated Root Cause Analysis, we’re able to understand the network right when it happens, run it through models, and have an output ready for the user so the user can understand where to look and what’s going on immediately. Not only that, but Generative AI contextualizes an event as we would understand it. It interprets text that we’ve put in, things like field tickets or other content, so it can internalize that and provide it back to the user to understand the event right when it happens. The accelerated Root Cause Analysis technology is a very powerful technology and we’re seeing great results in the field.
Early insights from the accelerated Root Cause Analysis trials
Fujitsu has been running extensive trials with the accelerated Root Cause Analysis app and many Proofs of Values (POVs) with our customers. What we’re seeing is the ability to look at millions of alarms and then process them very quickly to provide a result for the user. With alarm suppression, we can take in, for example, 700,000 alarms and suppress that to below 1,000 events. This is already a great achievement with respect to alarm suppression and what that means is it can focus on the right alarms and the locations that are important to the operator while reducing all the alarm noise that’s on the network.
How real are these use cases? We leverage our Technical Assistance Center (TAC) tickets that we have with operators. As an example, we had a case where we’ve seen hardware replacements on our TAC tickets. We ran this against the accelerated Root Cause Analysis app and the app actually would have told us a month before that the hardware needed to be replaced. This was right down to the card and module that needed to be replaced, and it had ten notifications prior to a TAC ticket being created. This is a great example of AI coming together to provide insights into the network and do it more efficiently than we traditionally do today.
At Fujitsu, we are learning to use new technologies like AI, ML, and Generative AI to help network operators. Learn more about our accelerated Root Cause Analysis app and how it can be used in your network.
