Recent research from Gartner suggests that there are three possible trends that will drive the move to 5G: new services and applications, 4G network obsolescence, and cost disruption. While overall I tend to agree with their assessment, it left a few questions unanswered for me, one of which I’d like to address here, which is: “When we get to 5G, which will be more important — the radio or the network itself?” This is a fundamental question, which in some ways is relatable to each of the three primary drivers Gartner identified.
Obviously, we need both a radio and a network. These two elements go hand-in-hand, and both are major and critical areas of research and development in 5G. But which is more important? And perhaps a deeper question is “where will the greatest evolution need to take place?”
These questions will be answered by discussing advances in radio access technology, and by examining changes to network architecture.
At the most basic level, there are certain aspects of 5G use cases where simply a faster radio at the device/handset end is not going to be sufficient to address the challenges we expect to face. What do I mean by that? Well, we will need the radio to be fairly robust as well as improved in order to support the kinds of throughput that we expect to see in the future (for instance, someone walking around watching a 4K resolution movie on a mobile device). Innovations will be required in order to make that a reality, and to ensure all kinds of interoperability.
However, when we zoom out a bit to a larger point of view, and recognize that in a densely populated urban area, or a stadium scenario where thousands or tens of thousands of people are simultaneously using that kind of bandwidth, we’re looking at a tremendous amount of network complexity added to the story. The scenario simply won’t work without the network itself being properly evolved to a 5G network. Add to this a growing range of scenarios, each of which comes with its own requirements regarding throughput, delay and reliability, and the complexity further increases.
This is different from the challenges we faced when migrating to 4G LTE networks we have today. In that case, 4G was primarily about facilitating internet access. Contrast this with 5G, where we’re talking about highly-integrated vertical use cases that span many different industries — vehicle-to-vehicle communications, transportation and industrial automation, the internet of things, and so on, all requiring high levels of interoperability — and again, radio alone isn’t going to suffice. The 5G networks are going to require new approaches to network access and management technology.
This is not to say that we won’t see significant advances in radio technology: 5G radios are going to be multi-radio access technology enabled, and they will have highly specialized radios for low bandwidth applications. There will be another set of radio access technologies that are highly specialized for high throughput (like 4K or even 9K resolution video and virtual reality applications). But all of this new radio technology will put even more strain on the needs of the network.
We will need a network that facilitates the user to utilize network resources as they move around, even one that utilizes core networks that span more than one operator. For instance, a given user might have a service contract with Verizon, but might actually be using two or three different core networks at the same time with different service breakouts for each, depending on how they’re using their device.
That’s quite different from the way 4G assembles services today. Instead, network software will grow to become a key driver of the 5G network future, as also recognized by Gartner. This means that certain aspects of the network will become “softwarized” and virtualized, in order to improve performance and capability. This will largely be done with two complementary approaches: network function virtualization (NFV) and software-defined networking (SDN).
Initial examples for NFV target, e.g., to accommodate a sudden increase in call numbers and attempts by virtualizing the core network’s call processing. In 5G, we move much more toward virtualizing entire chains of network and application functions, so that anything that can be executed or computed can be supported by NFV. And NFV will not exist only in central places like data centers, it will exist in distributed locations like base stations, network attachment points and other edge locations.
On the other hand, SDN exists in certain forms today, but is largely a data center technology. In 5G, SDN will evolve into a core access network technology that will be more uniformly used across anything that has to do with networking and switching. These two technologies will help overcome the network obsolescence issues presented by 4G and enhance overall capabilities of the network.
As radio technology evolves, and in ways that will place additional strain on existing networks, the networks will have to stay ahead of the radio advances. For instance, as we move toward 5G and see the proliferation of multi-radio access technology (which will become a standard feature of 5G), changes will be mandatory. In today’s networks, we see all kinds of a given user’s network and data traffic being sent over a single radio, but this will change in 5G. We may see low bandwidth data being sent over one type of radio, and enhanced mobile broadband traffic sent over a different radio. This is again where the network comes in: the ability to not only handle these different types of radio access technologies, but to route and manage all of the data efficiently and in a holistic way will become vitally important.
