Virtualized radio access networks (vRAN) promise many benefits, but they have some limitations that are rarely addressed. Most of these drawbacks hinge on spectrum and how mobile network infrastructure uses those airwaves to propagate coverage and transmit data, but there are also constraints that are unique to vRAN. 

Millimeter-wave (mmWave) spectrum, the uppermost spectrum used by a small number of cellular networks today, has many inherent disadvantages including extremely-low propagation characteristics and complex computational requirements. And this too is where vRAN runs into a mosaic of technical limitations, according to analysts and some RAN vendors. 

Alok Shah, VP of networks strategy at Samsung Electronics America, a RAN vendor that recently introduced a fully virtualized 5G RAN portfolio that was first deployed by Verizon, posits that vRAN is unlikely to deliver the benefits that operators will require in mmWave bands. 

“One thing we do feel is that as you move further up into millimeter wave, architecturally it may not be the right decision to attempt to virtualize that baseband because you’re talking about a massive amount of data, a massive amount of computation, and the economics of trying to fully virtualize that are not to the operators’ benefit,” he told SDxCentral in a recent phone interview. 

The industry has largely coalesced around virtualizing 5G networks operating in low-band spectrum because of these myriad and compounding challenges, he explained. 

“Higher frequency simply requires faster baseband processing,” Francis Haysom, principal analyst at Appledore Research, told SDxCentral.  

The complexities of high frequency also makes processing on generic equipment more difficult because fade, or the attenuation of signal processing, is more demanding on the hardware and software in mmWave spectrum, Haysom explained. 

Moreover, the “point-to-point nature of mmWave means that massive MIMO (multiple-input, multiple-output) has to be used to create lots of narrow beams," and that requires the RAN architecture to manage more processing, Haysom said.

Finally, the distributed workload environment that vRAN employs through a complete separation of hardware and software could lead to unknown latency between workloads, Haysom added. 

All of this makes vRAN more challenging in mmWave networks, he said. 

These problems aren’t exclusive to mmWave so much as they are heightened and increasingly complex in networks that push into higher frequencies. 

Samsung, for example, is confident vRAN can solve those problems on mid-band 5G networks but virtualization faces technical challenges there as well, according to Shah. 

“As the capacity increases and the bandwidth increases, the computational requirements increase,” he said. 

“There’s been a lot of great work done by folks like Intel to bring a tremendous amount of computation into the vRAN architecture. We’ve worked closely with them on that,” Shah said. “There’s also the opportunity to add acceleration, whether that’s FPGA or GPU. There are different types of acceleration that can be brought to the table to further increase the computational capacity.”

The technical hurdles facing vRAN are complicated and being addressed by many vendors that hope to overcome these problems, but some of these limitations are also built into the foundation of how networks are operated and managed. 

Mid- and high-band spectrum present problems that aren’t exclusive to vRAN, but these difficulties are compounded in a virtualized environment. 

“Mid-band spectrum is both a blessing and a curse,” Dan Hays, partner at Strategy&, PwC’s consultancy, told SDxCentral in a recent interview. “It’s a blessing in the capacity that it offers for mobile operators and for end users, but the challenge of deploying mid-band spectrum is the significantly larger site count that’s required to get meaningful coverage.”

The cellular radio sites required in 2.5 GHz, compared to 3.5 GHz and 3.9 GHz, are “vastly different even though they’re pretty close together,” he said, adding that “it’s an order of magnitude kind of difference in coverage” between 2.5 GHz and 3.9 GHz.

“The challenge for the industry is going to be funding, locating, and constructing those extremely large numbers of sites that will use the mid band,” Hays said.