AI, Open RAN and sustainability—Fujitsu on key themes of MWC

AI-driven community operations will cut back opex, streamline efficiency monitoring and improve operator agility

Fujitsu expects to see an acceleration within the tempo of Open RAN (O-RAN) deployments this yr and a consolidation when it comes to industrial deployments in 2025, in response to Patrick Eriksson, vp and world head of radio unit enterprise at Fujitsu.

“I believe it [O-RAN] began fairly effectively with a few greenfield operators that did pretty massive deployments. After that, O-RAN deployments halted just a little bit after which this step to come back into brownfield deployments has been just a little bit difficult to take. I believe, from our perspective, what we’re seeing…proper now’s we consider issues will occur this yr, and are available to fruition…perhaps in 2025 and onwards,” he defined throughout a current interview with RCR Wi-fi Information at Cell World Congress in Barcelona.

Given the parallel pushes for opex discount and progress towards net-zero objectives, Eriksson additionally highlighted how Fujitsu is approaching power effectivity (and attendant prices) on the hardware-level, and in addition primarily based on an understanding of actual world community demand.

“One is the uncooked {hardware} effectivity that we’re working with. And right here we’re using, for example, gallium nitride energy amplifiers (PAs) all through our entire portfolio. Along with that, we’re additionally deploying proprietary PA architectures and power effectivity algorithms which are serving to to enhance our total energy effectivity on the {hardware} facet… we even have the options that optimize the facility effectivity in accordance with the precise site visitors load of the radio,” he stated.

Utilizing AI for community root-cause evaluation and predictive upkeep

Fujitsu can also be utilizing synthetic intelligence (AI) to develop new instruments to assist operators enhance community efficiency. Throughout Cell World Congress, Greg Manganello, senior vp at Fujitsu, defined the primary options of the Alarm Storm Detector software developed by Fujitsu utilizing AI.

Alarm Storm Detector is an AI/ML-powered utility engineered to streamline community efficiency monitoring. “We have now mixed our community operations data with AI to essentially assist community efficiency. And when there’s a community bother, what we actually realized is that it’s not a low degree of bother, however large spikes, which could be a little bit overwhelming to community operations groups. And we realized they’ve been spending days and weeks searching for the basis trigger. So utilizing our AI instruments, we will discover the precise root reason behind a community bother, throwing away lots of and hundreds of by-product alarms and attending to the basis trigger,” stated Manganello.

“The subsequent step that we’re engaged on proper now to launch is predictive upkeep. So after we detect degradation of community parameters, we already current that to the operator,” he added.

Manganello additionally famous that Fujitsu is presently engaged on massive language fashions—a form of a chat GPT interface for community operators—, taking a look at log information and large petabytes versus of information, after which serving to the community operator question that database.

“However what we understand is LLM can’t function on all that uncooked knowledge. So we invented a brand new type of AI known as computerized data extraction, which actually is sort of a bridge between all that multi-vendor multi-domain knowledge, after which community operators to allow them to get the solutions again to their questions on the best way to actually function a community extra effectively,” he stated.