Predicting the build out of RANs by carriers is becoming more complicated as carriers become more selective on where they build out 5G to reduce capex. Another twist in the prediction process has been added by artificial intelligence, which will bring a 25 percent increase in the capacity of the existing RAN equipment in the field, Joe Madden, Principal Analyst and Founder of Mobile Experts, told Inside Towers.
“This is a threat to tower companies, because it could delay the build out of wireless infrastructure,” Madden said. “It’s an opportunity for software vendors to sell upgrades to the operators, but it doesn’t involve new hardware.”
Mobile Experts uses modeling to predict how many towers, base stations, and radios that the major carriers will deploy, as well as which bands they plan to use. The firm’s report, “AI Optimization of the RAN” found that AI will have a definite impact on RAN build outs, because of the increased performance it will provide to cellular networks. That improvement is the result of the ability to optimize around 200 different settings at the site, including more accurate beamforming, modulation, and antenna tilt settings.
“When humans optimize a network, they can only grasp maybe eight to 10 different parameters and optimize those, but AI can grasp all 200 parameters and come up with the best possible solution for capacity,” Madden said.
Mobile Experts came to its conclusions about AI from analyzing field trials of the top five network vendors: Ericsson (NASDAQ: ERIC), Nokia (NYSE: NOK), Samsung (OTC: SSNLF), Huawei, and ZTE. Fifteen carriers that were engaged in testing AI told Mobile Experts that they will deploy AI when they run out of capacity, possibly delaying the build out of more infrastructure. Some are already using AI to optimize their networks.
“All of the top vendors are conducting field trials using AI technology, and the resulting capacity rose from 15 percent to 40 percent,” Madden said. “I settled on 25 percent for a network, which is already somewhat optimized but needs further optimization. But there will be cases where capacity could be increased by 40 percent or more.”
On the flipside, AI is increasing the demand for data on phones by 10-20 percent, potentially gobbling up that extra capacity as soon as it is generated, he added. “We may find that the benefit of AI is not as big as we think when we take everything into account,” Madden said.
Madden also explained how AI modeling can help site acquisition professionals to better identify areas needing new towers to increase capacity, rather than just seeing AI as a threat.
“A lot of the coverage models that have been used in the past will show where the coverage holes are based on a weak RF signal,” he said. “But now we have to look at how much traffic is happening at each spot on the map and make heat maps to see where additional capacity is required, not just where the signal is weak. That’s a completely different model.”
By J. Sharpe Smith, Inside Towers Technology Editor
Reader Interactions