Architectures, splits and options: SCWS will make sense of a dizzying array of choices

When small cells first became part of the mobile landscape, they were relatively unified in terms of technology and form factor, as befitted a primarily residential system.

Small cell developers and deployers may feel very wistful for those days now, since the architecture choices that go with rolling out a network have become as diverse as the locations and use cases the systems must support.

There are important choices to be made related to spectrum bands; 5G Standalone options; whether or not to virtualize and disaggregate the network; if so, which functional split to adopt and whether to specify open architectures. And of course, then there are knock-on decisions about how these various choices impact on one another, and how the whole network can be aligned with other deployments such as an edge cloud.

All these will be discussed in Session 4 of the Small Cells World Summit conference, which is entitled ‘Architectures, splits and options’. This will be the most technically focused of the sessions, but with a clear focus on making technology choices that are fully aligned to commercial objectives, recognizing that these will be different for various deployers and their supply chains.

One of the overriding objectives of any platform is to reach sufficient scale that cost efficiencies and innovation are automatically enabled, that one enabler of scale is to allow organizations to make different technology decisions, but within common frameworks and platforms. Presentations such as ‘Getting real about vRAN’ by Radisys and ‘Into mainstream – making small cell deployments take off’  by EdgeQ will discuss ways in which hi-tech innovation can develop alongside commonality and deployability.

One of the key technology issues relates to emerging spectrum bands. These have proliferated with 5G, and the potential to use millimeter wave frequencies, while presenting many opportunities for hyperdense networks, also raises many engineering challenges. These have been the focus of considerable effort by chip providers and radio developers in recent years, as highlighted in Small Cell Forum work to extend its small cell definitions and specifications to these high frequency bands. In ‘Taking mmWave to the next phase’, Qualcomm France’s Philippe Poggianti will examine the implications of adding high frequencies to the small cell mix, now that supporting chipsets and products are becoming commercial.

Then there are the often confusing combinations of options and splits that any modern RAN designer has to consider. Options refer to the way that 5G Standalone can be implemented, and how it relates to any existing 4G or Non-Standalone core and RAN. Meanwhile, splits specify how RAN functions are divided between a radio unit and a baseband unit, in networks where the two elements are disaggregated, and the baseband potentially virtualized. The array of combinations that may be adopted may look obscure, but they will have a profound impact on the type of use case for which the network will be best-suited, and on the type of chips, fronthaul, servers and antennas that will be needed. Understanding the cost/performance interrelations and the different scenarios is essential so that deployers make informed decisions about whether, and when, to move to cloud-based architectures, and how the precise splits and options they select will align with their commercial priorities.

This is why the panel debate that will conclude this session should be particularly informative, and includes a wide range of participants from across the ecosystem to discuss ‘a use case driven roadmap for Open RAN’.

This will address the most important aspect of any architectural debate – how the architectures need to be planned with optimal performance for particular use cases in mind. A small cell network deployed for next-generation factory control will have very different requirements to a smart city broadband hotzone, for instance. Recent SCF surveys of small cell deployers have found that operators recognize that different functional splits are optimal for different scenarios. In a survey of 113 deployers, 44% expect to support three or more splits by 2027 across various macro and small cell networks.

A very common approach is to favor Split 6 for in-building or enterprise networks, while adopting Split 7.x in a macro or outdoor micro RAN, and some operators are calling for development of software-defined splits for even greater flexibility. SCF has focused mainly on Split 6, in which all the Layer 1 (PHY) functions of a RAN are left to run on dedicated hardware in the radio unit, while the Layer 2 (MAC) functions and above are in the cloud. This reduces the processing burden that some splits place on fronthaul media and processors, and so in some scenarios can reduce the need for expensive fiber and hardware. Paul Senior of DenseAir will argue the case for Split 6 in Open RAN deployments during the session.

Of course, Open RAN has been one of the hot topics of debate in the mobile industry for several years and is sure to be prominent in SCWS again. Decisions about adopting disaggregated vRAN architectures often trigger plans to support open platforms and perhaps introduce new suppliers. In small cells, this is often happening more rapidly than in macro networks because small cells more commonly go into greenfield environments such as enterprise private RANs, where existing architectures and suppliers are less incumbent.

And in many ways, small cells have always  been an open ecosystem – SCF has been defining open interfaces since its foundation, and recent years have seen it specify the important FAPI and NFAPI families of open interfaces to drive multivendor interoperability at component and system levels. This has encouraged a more diverse set of suppliers, integrators and service providers than in the macro networks, and that helps to drive a significant level of innovation and cooperation across the segment, which will be showcased at the conference, and particularly in this year’s SCF awards.

Number of functional splits deployers expect to support by 2027