Featured Story
VMO2 judges Ericsson vs. Nokia on AI-RAN #
Virgin Media O2 has installed an Nvidia processor at a busy Nokia site in central London to see if it offers genuine AI benefits.
By using Federated Wireless' Adaptive Network Planner (ANP), fixed wireless access provider Element8 saw its installation success rate jump from 79% to over 90% during a five-week deployment.
Element8, a Texas-based fiber and fixed wireless access (FWA) provider, cut its RF-related failure rate by more than half when using Federated Wireless' physical AI for network planning.
The deployment occurred over five weeks in Texas and Oklahoma, with Element8 using Federated's Adaptive Network Planner (ANP) to ensure locations were serviceable and installers had appropriate equipment.
Federated Wireless has invested $5 million in R&D for ANP and anticipates further spending as it shifts from software-as-a-service (SaaS) and toward AI for network planning.
As broadband operators experiment with the role AI can play in improving network deployment and performance, Federated Wireless is out with a new case study demonstrating how its Adaptive Network Planner tool can boost the installation success rate for FWA providers. That comes after Texas-based operator Element8 was able to cut its FWA installation failures in half when using ANP for a five-week deployment in Texas and Oklahoma.
Element8 exceeded its typical 79% installation success rate, reaching over 90%, after employing ANP's pre-qualification and installation planning models over five weeks, according to the companies. The tool enabled the operator to anticipate deployment obstacles and adjust equipment and placement to reduce signal failures.
Federated's ANP, which runs on Nvidia graphics processing units (GPUs), initially came about roughly two years ago, when Google announced it would retire its CBRS network development tool. That led Federated to "rush" the first version of its own ANP tool (now on version 4.0) to market, to fill the void left by Google, said Federated Wireless CEO Iyad Tarazi. Since then, Federated has broadened its physical AI platform to also include a "Spectrum AI" tool for improving CBRS shared-spectrum network deployments.
Tarazi sees ANP as a part of a broader shift away from using SaaS for network planning functions as AI enables operators to embrace "digital twinning" – essentially a living, virtual model of a network (or as Tarazi puts it, a "video game equivalent").
"Digital twinning is a huge cost savings because if you know what the plant looked like before you go ... the more you're able to do planning properly," he told Light Reading.
Notably, some researchers have warned of risks with digital twinning in live networks, pointing for example to how inaccuracies in a digital twin's representation of a physical network can result in degraded performance. Moreover, Tarazi acknowledged that Federated Wireless was "naive" when it chose to pick up the ANP mantle Google abandoned. "We found out why Google left, because it was too complex," he said.
But he added that Federated is forging ahead. According to Tarazi, 20 operators are currently using ANP, with another 50 in discussion.
For Federated Wireless, the shift to physical AI has led the company to invest "somewhere north of $5 million" in R&D over the last two years, with more investment to come as the company explores additional benefits, said Tarazi. The company will demo its tech at the Wispapalooza conference in Las Vegas next week. "It's valuable for us beyond just a planning tool," Tarazi said. "It is now beginning to replace the SaaS technology for spectrum management because it's far more accurate."