SmartCIC says the accuracy of its algorithm was confirmed during recent driving tests in the Washington, DC area.
Monetizing 5G is one of the most pressing challenges facing the telecom industry. In Toby Forman's view, this leads to two questions about 5G networks: How can operators monetize their investments in 5G, especially 5G standalone? And how can businesses leverage these networks for enterprise applications to become more agile and disruptive?
Forman, CEO and co-founder of SmartCIC, said his company is trying to answer these questions, which, in his view, are fundamentally tied to the specifics of 5G coverage and network performance in specific locations. After all, it's impossible to convince businesses to use 5G for their applications if the coverage and performance don't meet their needs.
SmartCIC's view is that if AI can predict metrics such as network speeds and latency at a granular level, it can not only help enterprises leverage 5G and operators monetize it, but it can also streamline related processes by reducing the amount of testing needed on the road and “achieve the data density needed for predictive analytics”, and in the process, optimize the drive test routes that need to be conducted, ultimately reducing resource usage and carbon emissions across an organization.
But first, the AI needs to prove it can accurately predict real-world network conditions, so SmartCIC ran the numbers in the Washington, DC area and then conducted drive tests earlier this spring to gather actual numbers on network performance to compare with the AI predictions.
The results? A spreadsheet comparison of predicted and actual network data revealed that SmartCIC's AI was more than 95% accurate in predicting metrics such as upload and download speeds for the three domestic carriers, reaching 98% in some cases, according to the company's data.
The accuracy “far exceeded our expectations, and we're pleased to see this level of accuracy achieved across a major U.S. city,” Forman said. “This result reflects the skill level of our data and analytics teams and the progress they make every day. They're optimizing their approach to AI modeling and refining their algorithms, producing impressive results.”
SmartCIC is spending this year establishing “independent and enterprise-grade cellular intelligence” in 30 National Football League (NFL) cities in the U.S., with a mapping tool that will provide heatmaps of performance. The company also plans to collect data that will make its modeling even more robust. SmartCIC's AI engine will combine measured network performance down to the specific address level with predicted performance to “provide visual clusters of areas detected as having consistent behavior.”
“We visualize the data and make it immediately available to engineers, network planners and senior decision makers,” says Forman. “Operators gain insight into where they can deploy enterprise-grade solutions, accelerating network monetization. They can sell more enterprise services if they can instantly know the quality of the connectivity available at a site.”
Asked about the bigger picture of how the company's data relates to monetization, Forman said, “I think people started investing billions and hundreds of billions of dollars around the world in these next-generation networks without a clear understanding of how they were going to derive value from them.” [and] Other companies can extract value from high-speed network connectivity, but how do you extract that value if you're an infrastructure provider? … It's very hard to monetize that other than adding a few dollars more to the value of the subscription every year. So you have to think of other ways. And I genuinely think that's a challenge that MNOs struggle with, and what we want to do is help them.”
He continued: “We believe enterprise solutions are the way to go. They build stronger relationships with customers, and we see that through our professional services business. And if you can provide a solution that works, customers will pay for it. And customers want that opportunity.”
Monetization could have a variety of meanings, Forman thought. Perhaps it would make it quicker and easier to predict where FWA services can be launched. Better data on roadside coverage across specific areas or regions could help drive automotive-related use cases. The company is also looking at signal height and building penetration to get a holistic view of potential support for things like smart building use cases. Ultimately, Forman said, “There are a lot of different use cases, and what we're aiming for is to be able to give MNOs and enterprises an independent view. So if you're an MNO, you can look at this data and say, 'This is how my network is performing, and this is how I can develop a compelling enterprise solution to sell to my customers.'” Rather than simply providing a platform for someone to generate revenue, [they can] … You actually start developing your own solutions.
“Enterprises have the information to make investment decisions about how to use these networks. There is certainly a desire to use these types of networks. People just don't understand how they perform,” he continued. And they don't necessarily trust what mobile network operators say about their network performance. That's why Forman sees a role for companies like his and the value of AI that can accurately predict network performance down to the individual address. “One thing we're pretty sure of is that enterprise customers don't trust as much information directly from the MNOs. There's a role for independent organizations to be able to test and validate the performance of these networks,” he said. As CTOs and CIOs want to better understand how wireless access can be used to power enterprise applications, “granular” testing like SmartCIC's SD-WAN is “starting to create qualitative information that shows how these types of applications work in these environments, and we can provide that to organizations and sell it to them,” Forman said.
SmartCIC is based in France and holds an operator license there. Earlier this year, it opened a new innovation hub in Barcelona. The company has a technical division and a professional telecom network services division, with a network of 25,000 field technicians around the world. Forman says the company already works with Tier 1 operators and large system integrators, and developed its AI in-house. SmartCIC's approach to testing differs from the typical crowd-sourcing model or simple RF testing, Forman explains. With crowd-sourcing data, you don't have control over what Forman calls “where, why, and what you test.” “If you can control those factors, you can really start using that data and use AI to provide predictive analytics when you can't control a lot of variables.” Meanwhile, testing that focuses on the RF environment “only tells part of the picture,” he says. … You can have great RF, but still have very poor network performance because of issues with the architecture or something else.
“I don’t think anyone is at the halfway point where we are right now, so it’s very exciting,” he concluded. SmartCIC is currently focused on developing AI to predict terrestrial cellphone performance, but is also looking at low-power wide-area network technologies and eventually low-Earth-orbit radio coverage.
“We're very interested in providing information,” Forman said. “We're working on it in the wireless space first, and that's our focus right now. But there's a lot of really interesting things out there that we'd like to provide answers for.”
