AI is opening new possibilities for telecom, from optimizing investments to accelerating integration, debugging and validation. Yet the value of an AI system depends heavily on the data behind it. Standards and synthetic datasets can describe how a RAN should function, but they remain theoretical. Live networks operate across many conditions, making real test data essential to model performance.
VIAVI Solutions' VALOR facility brings vendors together in a real Open RAN environment where automated testing generates practical data from multi-vendor configurations. According to Srinivas Sriram, the lab has more than 600 automated test cases and has delivered more than 2,000 test cases across more than 30 customer engagements. That scale helps produce cleaner data while reducing integration and debugging time and enabling faster test results.
VALOR is also working toward a dark-lab model that uses AI for 24/7 monitoring and, as its training data develops, self-healing operations. Combining autonomous lab capabilities with real-world data creates a feedback loop: automation accelerates testing, testing produces better data and better data improves model performance. For the Open RAN ecosystem, that can support faster innovation and stronger confidence in how systems will perform beyond the lab.
VIAVI Automated Lab-as-a-Service for Open RAN (VALOR) is a purpose-built, AI-enabled Lab-as-a-Service for Open RAN in Chandler, Arizona. To learn more about VALOR, visit: https://www.viavisolutions.com/en-us/valor
Steve Saunders:
Let's start with a big question. How is AI affecting telecom infrastructure?
Srinivas Sriram:
So one of them would be how would AI help M&Os optimizing the investments in returns, basically. As you know, telecom is already been struggling with the returns.
Steve Saunders:
Absolutely.
Srinivas Sriram:
So AI is going to bring a lot of use cases, new use cases to the telecom world as well.
Steve Saunders:
Yeah. So it's obviously having a big impact on telecom infrastructure. How does it affect operating a lab like this?
Srinivas Sriram:
It's entirely transforming how we operate our labs as we start to use AI more and more. To give an example, now we are working towards a dark lab here using AI, which means it's monitoring 24 by seven today based on the next data trainings. It will go into the self-healing mode as well. That's our belief here in VALOR. So probably-
Steve Saunders:
So you're doing basically level five autonomy, but applying it inside a test environment.
Srinivas Sriram:
Absolutely. Yes.
Steve Saunders:
That's pretty cool.
Srinivas Sriram:
Yep.
Steve Saunders:
I mean, it all goes in a huge circle, doesn't it? Because O-RAN is designed to enable interoperability and open up the RAN access market. And in order to do that, the test lab has to be one of the first implementers of all of those first principles of interoperability. And the answer to all of those things, it turns out, is AI, both as a solution but also as a test protocol as well.
Srinivas Sriram:
Absolutely. I mean, that's where we are focusing right now to bring in how would we optimize these solutions or bring faster to the markets. This is helping enormously, cutting down our integration times, debugging times, and enabling faster testing results.
Steve Saunders:
Why is having high-quality data so important for AI test systems?
Srinivas Sriram:
The model training depends on the data you provide, and that's where this lab plays a crucial role. I mean, anyone can generate synthetic data. I mean, synthetic is something which is easily available in the market. Example, you take a test specification that's in a document form that's easily available for the internet. But what these kind of labs brings in is the real, actual test data. The more data you have, the more cleaner data you have, the better your model performance.
Steve Saunders:
So when you say synthetic data, what does that mean?
Srinivas Sriram:
Synthetic datas are more of a standards written somewhere, which is just how a RAN network should function, but that's just something written on the paper. I mean, it's theoretical.
Steve Saunders:
It's abstract.
Srinivas Sriram:
Very abstract. Exactly. But in real performance, that's not something you often see. You see a hundred different conditions.
Steve Saunders:
And here you're actually creating real environmental-
Srinivas Sriram:
Real environment with different vendors, which is more valuable, like Open RAN, as again, going back to the disaggregated aspect.
Steve Saunders:
This is all being applied to speed up innovation in the telecom industry, and that's very, very important, isn't it? Do you agree with that?
Srinivas Sriram:
Oh yes. That's the reason. I mean, this lab brings in that aspect of bringing it faster as well. I mean, this is operated at an autonomous scale wherein we could generate data, not just data, but the overall integration of any network faster, accurate results. I mean, we already have 600-plus automated test cases here.
Steve Saunders:
Well, it's been built incredibly quickly, and you're way ahead of the expected KPIs for the testing which you're doing here. I guess you've arrived in the market at exactly the right time.
Srinivas Sriram:
Yeah. I mean, our numbers speak to what we have done. We have so far had 30-plus engagement, 30 customer engagements. We delivered over 2,000 plus test cases. We are forefront of bringing AI innovations to this lab. We are looking forward what comes next.