The Quest for the Perfect VRAN

Facebook’s Telecom Infra Project (TIP) recently announced that two of its members, Telefónica and Vodafone, each issued a Request for Information (RFI) for OpenRAN solutions for 3G and 4G mobile networks. The companies plan to identify vendors of Radio Access Network (RAN) equipment compliant with the OpenRAN requirements published by TIP’s OpenRAN project group. The OpenRAN project group aims to develop fully Virtualized RAN (VRAN) solutions capable of being implemented on Commercial Off-the-Shelf (COTS) hardware. Only equipment that demonstrates open interoperability between baseband processing, radio hardware, and software will be evaluated. The companies expect to present the results of the RFI and name the vendors that meet the RFI requirements at the TIP Summit 2018 in London in October, and to identify those vendors that meet RFI requirements.

TIP is not the only association investigating RAN virtualization. The xRAN Forum has also been investing the topic and has merged with the C-RAN Alliance to form the ORAN Alliance. (These groups were outlined in ABI Insight “vRAN a hot topic at MWC18”). The ORAN Alliance recently expanded its operator membership to include Bharti Airtel, China Telecom, KT, Singtel, SK Telecom, Telefonica, and Telstra, which now join AT&T, China Mobile, Deutsche Telekom, NTT DOCOMO, and Orange. So far, the ORAN Alliance does not include original xRAN members, such as T-Mobile, Verizon, KDDI, Sprint, and Jio.

These are important signposts from some of the world’s largest Mobile Network Operators (MNOs) wielding considerable purchasing power over the RAN hardware vendors, including the Big Four (i.e., Ericsson, Nokia, Huawei, and ZTE).

RAN virtualization is attractive to MNOs because it promises to lower costs and help to quickly add new capabilities. A VRAN promises MNO reductions in Capital Expenditure (CAPEX) and Operational Expenditure (OPEX), along with scalability in the RAN, allowing the MNO to scale network resources up or down according to traffic demand. Having a VRAN means that an MNO no longer must over-provision the RAN by designing for peak demand. The RAN is run in software by using Network Functions Virtualization (NFV) to replicate the baseband functionality of the basestation in a telco data center running COTS server hardware.

The move to VRAN is also motivated by MNOs’ desire to reduce costs by commoditizing the various RAN hardware and software elements so that multiple vendors can be “mixed and matched” in the RAN. This “white box” RAN would ideally permit each MNO to build for the lowest cost depending on the individual use case or deployment scenario.

The conventional approach to VRAN is to disaggregate the baseband and radio in the RAN, distributing the radio and antenna to the Radiation (RAD) center on the tower, and centralizing the baseband in the data center. This is an approach that defines the Remote Radio Head (RRH) as the physical network function connected to the baseband over fiber and using the Common Public Radio Interface (CPRI) protocol. However, in a VRAN, the CPRI protocol can run into problems if the baseband partitioning is not carefully implemented. This is particularly true for networks that implement advanced network functions, such as Coordinated Multi Point (CoMP), Carrier Aggregation (CA), multiband support, Multiple Input, Multiple Output (MIMO), and massive MIMO, which will all depend on the fronthaul bandwidth and latency, and that will drive trade-offs in network performance.

Although the OpenRAN RFI is for 3G and 4G, 5G will exacerbate these challenges due to the promise of high peak data rates, high reliability, and high-capacity use cases with throughput planned to be 10X to 1,000X that which is supported by 4G.

While the business case remains to be completely proven, some MNOs urge caution. Disaggregating the baseband from the radio and connecting them with fiber fronthaul may entail new upfront costs, such as laying down new fiber and constructing new telco data center facilities. This would be an expensive consideration for MNOs without access to sufficient fiber.

Other VRAN challenges involve an Operations Support System (OSS) and network management, or NFV management and organization (MANO) across all the elements in the network. If different vendors inhabit the RAN, MANO may well be a problem. Also, warranty issues could also arise, given that multiple vendors are present and if a failure occurs it may not be clear to the MNO where the responsibility lies.

The ORAN Alliance and TIP, with backing from some of the largest MNOs in the world, is an attempt to regain control of network architectures from the Big Four, raising the prospect of these vendors competing with smaller software-based startups, as MNOs look to mix and match select components from different vendors in their RANs to achieve the best performance and energy efficiency. However, this vision of a truly open vendor-agnostic RAN is an ideal world aspiration. In the real world, hyper-competitive equipment vendors will likely strive to retain some form of vendor-specific hardware or software in the RAN to maintain their competitive differentiation.

This situation stems from how the CPRI protocol was implemented 15 years ago. Although it is a standard, vendors could retain vendor-specific functionality, such as telemetry, housekeeping, and alarms, in the protocol so that only a baseband and RRH from a single vendor would work. This vendor “lock in” made sure that RAN vendor revenue was maximized.

As the industry moves to 5G and complex techniques like massive MIMO and beamforming are adopted, proprietary techniques and algorithms will still exist that are unique to each vendor, which may prevent a truly vendor-agnostic VRAN from becoming a standard. It is much more likely that each MNO will demand that its chosen RAN vendors work together to implement the VRAN.

These are serious initiatives from the worldwide MNO community that the vendor community will be unable to resist. This forced reappraisal of the RAN will force RAN vendors to react and ABI Research believes vendors that choose to open source their products will succeed. This is the way to differentiate from the competition. As a result, that value will transition from hardware to software in the RAN in such a way that, although hardware can be mix and matched, it will be vendor-specific optional and open Application Programming Interfaces (APIs) allowing access to advanced functionality, such as network optimization and automation using machine learning and Artificial Intelligence (AI) that will count.

It will be interesting to see how this power play develops and if new “white box” infrastructure vendors move into the sector.