One of the key technology areas that needs attention in the push to develop competitive open radio access network (RAN) solutions is a more diverse supply of radio units (RUs). This, in turn, requires a better supply of the RF modules used to build power-efficient, cost-competitive RUs.
Competitive, open RU products are important because the RF chain (including CPRI interfaces, digital-to-analog converters, power amplifiers [PAs], filters, antenna subsystems and so on) typically accounts for around 70% of the bill-of-materials cost of a mobile basestation. And because the RU consumes most of the power and a major part of the physical real estate needed to deploy a basestation, it also accounts for a significant portion of the operating cost.
Established vendors with large product volumes and long-standing relationships with silicon suppliers can afford the expertise, time and financial cost to design and contract these solutions. However, for many in the wider ecosystem, from new entrants to midsized RAN players, radio hardware is an expensive and difficult proposition.
Open RU reference designs for pico cells
One way to address this barrier to market competition and bring the benefits of scale to more vendors is through open RU reference designs. Silicon suppliers are good candidates to catalyze this work because they develop many of the foundational products used to create RUs. In particular, pico cells can take advantage this type of reference design because there are many deployment scenarios and many new entrants and established vendors targeting this product segment.
So, what are the key characteristics of an open RU pico cell reference design?
Tight integration between that analog module, which comprises the RF frontend (RFFE) — the bulk of the hardware cost — and the digital part of the RU is essential. There are many co-optimizations between these subsystems that a reference design should address upfront to ease the burden on the product developer. Specifically, for the analog module, it is important to have a flexible RFFE that will support the following:
- Very wide bandwidth across diverse spectrum bands: For example, a 5G pico cell may need to support multiple carriers in the mid and low bands. Common configurations might be two carriers of 3.5GHz + 2.1GHz or two carriers of 2.6GHz + 1.8GHz. The industry also expects to use millimeter wave (mmWave) for pico cells, introducing new RF requirements.
- Scale to wide carrier bandwidths: For midband 5G pico cells, 100MHz is table stakes, with the ability to add 20MHz carriers at lower bands also important. Certain products will require 2 x 100MHz carriers, meaning 200MHz occupied bandwidth per pico RU will be needed in some cases. No question, this is a demanding requirement.
- Multiple antenna schemes: For example, a pico cell may require 4T4R for one 5G radio carrier or 2T2R for a 5G carrier and 2T2R for a 4G carrier. Product companies need platforms that have the flexibility to support end products with diverse antenna configurations.
- Ability to easily integrate PAs and filters for different product versions: Vendors need the flexibility to offer product variants with different power levels and bands, which may require alternative PA and filter modules. Some analog module vendors are providing built-in digital predistortion (DPD) algorithms that can adapt to different PA characteristics, making it simpler for the product designer to use new PA suppliers.
In the digital component of the RU — the part that interfaces with the distributed unit (DU; not the digital baseband itself) — there are similarly demanding requirements. This interface is particularly important to support the open RAN architectures now emerging. Some of the key features in this module include the following:
- Flexibility to map the data stream generated by diverse RF configurations to digital transport: This is critical to co-optimize the RU and DU to create a high performance 5G basestation. A flexible approach is to use a field-programmable gate array (FPGA) to connect to the optical module over SERDES.
- Different rate CPRI interfaces: Product designers need the flexibility to support multiple CPRI profiles. Typically, this means 6Gbit/s, 10Gbit/s and 25Gbit/s transport.
- Packetized fronthaul interfaces: eCPRI is the baseline, but there is an increasing need to support open RAN interfaces, such as split 7.2 and split 7.3. These are standardized interfaces but are still relatively immature. Vendors need programmable modules they can adapt as standards evolve and to lock down real-world interoperability with different DU vendors.
- Massive MIMO and beamforming: The fronthaul interface must enable tight interworking and co-optimization of RU and DU functions to achieve competitive massive MIMO systems. This is an area of ongoing development, with several options in play for features such as beamforming and uplink performance. Again, flexibility and programmability are critical.
A catalyst for commercial development
Reference designs do not solve the entire issue for ambitious open RAN players, but they can shorten design cycles and act as a catalyst for commercial development. Common, open reference designs can also help to scale R&D and spread the cost across multiple companies. Given the performance demands of 5G and the ongoing development of 5G features, the sharing of R&D costs has significant consequences.
The first opportunities for commercial products are likely to be in the small cell market. This is a broad category of radio basestation products that ranges from outdoor and indoor pico cells to fully fledged venue coverage solutions and private network systems. The market for these products includes the classic mobile network operators but extends into interesting and emerging sectors such as neutral host providers, venue owners and enterprise/industrial companies.
– Gabriel Brown, Principal Analyst – Mobile Networks & 5G, Heavy Reading
This blog is sponsored by Intel.