In this third of our “5G is coming!” series of articles, we look at the radio access network and its transformation projected in the “5G” era. As outlined in our first article, 5G is about supporting the Internet of everything and has therefore the most diverse set of requirements in any generation. Transposing these requirements into design requirements for the RAN immediately hits the challenge of the “one size fits all” approach. Instead, an approach that builds on the principles of flexibility and fine integration appears more practical and viable.
What is in the flexibility for 5G RAN? The 5G RAN (both architecture and functions) must be able to change or be changed easily and dynamically according to the service and infrastructure contexts. Take for example the architecture, neither a distributed RAN (D-RAN as in 4G) or a centralized (cloud) RAN will be always optimal anytime anywhere. While D-RAN is best to support the below one millisecond latency goal, the C-RAN is clearly more suitable for the other requirements of high spectral efficiency, high energy efficiency and reduced cost. Only a flexible D/C-RAN hybrid architecture will therefore be able to respond effectively to the service and network needs at any given time and place. Such a flexible custom-made architecture cannot be envisioned without the RAN embracing the concepts of virtualization and software-defined networking most popularly associated today with the core network.
This flexibility requirement in the RAN architecture will transform the transport network that we know today of separate and incompatible fronthaul and backhaul into one virtual flexible x-haul. Such is the vision of the European H2020 5G Public Private Partnership project Xhaul (www.xhaul.eu). The Xhaul concept is a versatile end-to-end packet-based transport under common software-defined networking-based control for flexible and dynamic (re)configuration of the nodes and distribution of the functions in a multitenant and service-oriented framework.
While this tighter integration between the RAN and the transport network is exciting at the base station (access node) level, it will get even more interesting when virtualization is embraced as we predict down at the level of the end-user device (yes, a smartphone can and will be virtualized, too). The cloning of a device in the cloud (central or local) bears attractive benefits to both the device and network operations alike (e.g., power saving at the device by offloading computational-heaving functions to the cloud, optimized networking – both device-to-device and device-to-infrastructure) for service delivery through tighter integration of the device (via its clone in the networking infrastructure). This is the focus of another European H2020 5G collaborative project called ICIRRUS (www.icirrus-5gnet.eu).
The flexibility and “softwarization” of the RAN needs to be considered not only at the architectural and functional level, but also within the design of each component (building block) of the RAN interfaces. Take for example the waveform. In 4G, the design of the waveform, known by its four-Letter acronym OFDM (orthogonal frequency division multiplex) has been relatively simple and geared toward meeting the high throughput requirement of video (the 4G killer application). In 5G, however, the waveform design is no longer revolving around meeting the needs of one killer application, but rather the effective support of the varying needs of a diverse set of applications. This calls for a new design that must be flexible and software-defined in order to tune the waveform’s performance trade-offs (synchronicity, latency, energy, mobility, reliability, availability, etc.) to the varying applications requirements at a given time. This does not necessarily mean that 5G will be abandoning OFDM and looking for a clean-slate four-letter acronym that can magically optimize for all dimensions and in all scenarios. Rather, 5G will need to look for the waveform or waveform framework (including OFDM) that can flexibly and dynamically strike different performance trade-offs in different scenarios (e.g., through the use of flexible and adaptable numerology for sub-carrier spacing, symbol timing, resource block dimensioning, cyclic-prefix duration, filtering, etc.).
While flexibility is certainly a key principle that is underpinning the 5G RAN design, it only represents one part of the story. Fine integration of multiple existing and new radio access technologies is the other part. It is certainly shortsighted to see the delta from 4G to 5G only through the new RATs 5G might be defining. The real delta rather will be in how effective 5G will be in flexibly integrating together existing and new RATs, both cellular and noncellular, spanning existing and new (licensed and unlicensed) spectrum below and above 6 gigahertz.
As of today, in addition to a half-dozen legacy RATs that are anticipated to be part of the 5G system, as many as another half-dozen new RATs are foreseen, too. The latter include, in addition to the next generation unlicensed Wi-Fi initiatives that are already underway (802.11ax, ay, ah), three new RATs anticipated from the 3GPP cellular world. A first one is expected to be a backwards compatible evolution of LTE, a second one nonbackwards compatible evolution in new but traditional spectrum below 6 gigahertz, and a third, clean-slate technology in a new nontraditional spectrum above 6 gigahertz.
Against this multiplication of the number of RATs overlapping in 5G, and the challenge this raises on the orchestration of the RAN resources, the only viable design appears to be one that pushes the bar of integration down to a granular (fine) level compared to today’s coarse offloading from cellular to Wi-Fi. This has already been recognized within 3GPP with work to support for example dual band connectivity, control and data plane decoupling, and link level integration with Wi-Fi. This trend will continue to grow in 5G, with the integration pushed even further down to the level of the radio resource, so that intelligent (software-defined) mechanisms can be used in the RAN for dynamic resource composition (slicing and aggregation) across the multiple RATs, and in a transparent way to the end user.
With what we’ve seen above, a lot of 5G tends to be an evolution from 4G, the real difference though is that 5G from the outset shall aim at an inherently flexible design that can federate or integrate the multitude of RATs effectively. This is in contrast to 4G where the design was first fixed by the needs of the one killer application (video) identified at the time, and adapted gradually for the support of other applications that have emerged since then. This 4G approach is not future-proof and the breaking point will define a very clear divide between what 4G is and what emerges as 5G.
This last article completes our “5G is coming!” series, by focusing on the RAN. Not only the key performance indicators are set to multiply by a factor of “x” in 5G, but also the degrees of freedom, number of options and range of technologies. In order to deliver on the simplified (and reduced cost) network vision of 5G, similar to the core network, the 5G RAN will need to build on the design principles of flexibility, programmability and fine integration.
Editor’s Note: In an attempt to broaden our interaction with our readers we have created this Reader Forum for those with something meaningful to say to the wireless industry. We want to keep this as open as possible, but we maintain some editorial control to keep it free of commercials or attacks. Please send along submissions for this section to our editors at: [email protected].
