Panelists at Open RAN Forum discussed how AI and automation will ‘uncover’ a great deal of power savings for telco operators
Energy efficiency has become a major priority for operators as they look to contribute to environmental sustainability efforts, as well as lower the operational cost of their mobile networks. At the Open RAN Forum, panelists discussed how artificial intelligence (AI) and automation can help operators get more efficiency out of their radio access network (RAN).
According to Bijan Nowroozi, chief technology officer at OpenRAN Compute Project Foundation, AI has the potential to “uncover” a great deal of power savings for telco operators. “This dynamic AI replacement of different layers… [in the RAN network] is going to unleash a lot of power savings,” he said.
The increasing trend of implementing intelligence at the edge means more algorithms are popping up in various network components, such as transceivers, where John Oates, systems manager at Analog Devices, said that features like channelization are giving operators the flexibility to adjust capacity for fluctuating user traffic flows, allowing them to use sleep mode features more intelligently and dynamically.
Further, thanks to the RIC, or RAN Intelligent Controller, algorithms and automation tools targeting energy savings can be developed faster than ever. The RIC is considered the key to Open RAN network programmability and automation, and it is a critical element in the management of 5G network functions like network slicing and low-latency applications.
“What RIC solves for operator is really tremendous in terms of automation, not only for energy saving but for many other features,” commented Ari Uskudar, senior product manager at VMware. She followed up with an example in which traffic appears to be “bursty.” Automation features can help operators detect if this is true “burstiness” or if traffic is actually picking up. Then, the correct actions, such as putting certain cells to sleep or waking up certain cells to accommodate the influx of traffic, can be taken. “Because putting cells too much into wake up and sleep mode could also depreciate the radio [and] the user experience,” she said.
Beyond helping operators decide which cell sites to put to sleep and when, Rob Hughes, the head of wireless marketing at Fujitsu, shared that AI can also allow them to predict the outcome of potential cell site shutdowns prior to enacting sleep mode features, providing more confidence that the user experience won’t be negatively impacted.
“And then the real benefit is that the system learns over time, so if there’s a new residential building that gets built… the system adapts to the changing traffic patterns and can alter based on the new patterns of traffic… And that way you’re always getting the maximum amount of energy efficiency you can get with[out]… impacting the user experience,” continued Hughes.
While AI has potential to help operators more easily assess energy saving opportunities, Irfaan Salahuddeen, the product planner for testing company Keysight Technologies’ edge to core business, highlighted the need for a standardized measurement tool for energy consumption and energy efficiency. “You can’t optimize something that you cannot measure,” he stated. “There’s not… enough test cases and test equipment [that] can do this on a standard model.” Keysight, then, has focused on taking the European Telecommunications Standards Institute (ETSI) test models, which Salahuddeen said are both static and dynamic models, to implement solutions that can be used to test the energy consumption of network equipment. Ultimately, though, he argued that the industry needs to foster partnerships to facilitate the creations of more standardized models.
For Oates, achieving energy efficiency requires the entire ecosystem to work together beyond the development of appropriate test solutions. Vendors, from DU and CU vendors to chip makers to stack vendors must “come together… intentionality” through “an explicit definition of energy saving modes” and “benchmark KPIs” in order to produce a “truly energy efficient” RAN, he said.
