Latest chip from Marvell simplifies transition to cloud-native RAN

‘The easiest way to migrate from the traditional to a cloud-native RAN is to use the same proven silicon and software for both,’ says Marvell

With Mobile World Congress Barcelona less than a week away, RCR Wireless News caught up with Marvell’s Senior Director of Solutions Marketing Peter Carson, to get a look at what to expect from the company at this year’s show. 

The biggest news is the latest addition to Marvell’s OCTEON 10 family processors: the OCTEON 10 fusion baseband platform. “[The platform] supports use cases from the RU [radio units] to the DU [distributed units] for both integrated — which some people call the traditional — RAN architecture to open, virtualized RAN,” Carson said, adding that the company is already working with big names like Nokia, Samsung, Dell and Fujitsu as well as top cloud software providers.

For O-RAN RUs, the OCTEON 10 Fusion processor has specialized accelerators optimized for processing complex 5G massive MIMO beamforming algorithms. The platform is also the world’s first 5-nanometer baseband solution, which Carson said offers improved performance and power efficiency, a critical enhancement as complexity continues to rise in the radio, increasing power demands. To facilitate time-to-market for developers, the company worked with Analog Devices to develop an RU reference design with the OCTEON 10 Fusion chip that supports 32T32R antennas with 400MHz of operational bandwidth and 300MHz of instantaneous bandwidth, enabling up to 40% savings in power consumption.

According to Carson, the new unified platform serves existing customers as they migrate to more advanced features and better performance and power. He also claimed that it provides what carriers are truly asking for. “[It] actually delivers open, scalable cloud-native RAN that meets the performance of existing and established networks,” he said, adding that this is a “critical requirement” because the transition to an open architecture is going to take several years. “Cloud-native RAN is not going to happen overnight,” he noted. “This is all about [delivering performance that] enables new, enhanced services that are the growth engine for 5G innovation; it’s about [enabling] new enterprise services, mobile edge computing, industrial AR and wireless factory automation.”

In a previous conversation with RCR, Marvell’s Senior Director of Wireless Marketing Joel Brand explained that the platform’s ability to support both the traditional and Open RAN markets makes the chip “very unique” and helps operators transition to vRAN and O-RAN without compromising a consistent user experience or high network performance.

“The problem with providing new chips for cloud RAN is that they are different from what is being used by all the operators around the world,” he continued. “They have different performances, different sets of features and different profiles in terms of power consumption. We’re not arguing good or bad, but, given the steep learning curve in wireless, the easiest way to migrate from the traditional to a cloud-native RAN is to use the same proven silicon and software for both. We’ve proven that this can be done in an open, virtualized cloud architecture.”

Brand further argued that for open architecture transition to be successful, it’s critical to tap into the expertise of the large, existing infrastructure vendors, rather than attempt to box them out. “We need their support. These vendors need to bring their expertise from the current environment to the future environment, and we are helping them with that process,” he said. “The goal needs to be to foster competition and an environment where all vendors, including incumbents, can be successful in the cloud environment. If everyone aligns themselves behind this kind of goal, O-RAN and vRAN will succeed.” 

To learn more, make sure to visit Marvell at Mobile World Congress at booth 2F34 in Hall 2 at the Fira Gran Via in Barcelona, Spain.