Multi-layer convergence and new approach to network architecture in support of 5G
5G is all about more–more capacity provided in a more efficient manner that creates more value for consumers and more attractive business cases for communications service providers. Next-generation connectivity will be realized through a convergence of wired and wireless sites, and ultra-dense networks comprising not only more macro towers, but also more small cells that support more frequencies including licensed, unlicensed and millimeter wave bands.
Fiber deployments can be leveraged to build increasingly dense wireless networks at the macro level as well as in support of massive small cell roll outs. Operators have shown an early interest in pre-standard 5G fixed wireless access. These fixed sites, being tested now, are connected to fiber and could easily be upgraded to support 5G mobility services once the technical specification is finalized. This level of convergence tracks closely with the centralization of base station equipment. The animating principle is leverage that access to fiber and power to the greatest extent possible. That’s one aspect of network convergence
“Practically,” Ray Butler of CommScope said, convergence “refers to delivery of services–residential, business and wireless services–over a common fiber network. On the mobility end of it, we’re seeing a lot of interest, and people starting to explore how you converge a site to support fixed wireless and mobility. How do you leverage that investment for both mobility and fixed? Those are kind of the ripe areas we see a lot of work being done on and a lot of potential opportunity. As the industry evolves to centralized RAN and cloud RAN with network function virtualization, we see an opportunity for both capex and opex savings.”
Derek Peterson, chief technology officer of Boingo Wireless, explained the longer-view on convergence of infrastructure from the perspective of a neutral host provider. “When we talk today about neutral host systems, I’m building a neutral network that allows many different people to connect onto. That’s what we try to do. If we can share fronthaul, backhaul, cabling, switches, that’s a win. We need an infrastructure for everyone.”
He gave the example of a shared roadway. “We want to be able to have all that share the same so that way the truck is not impeding the speed race car or whatever it might be. Now we’re moving to things like digital RAN or cloud RAN where you’re now able to share even more. The only parts that’s different in your road is your driveway, your gateway to identification. It reduces cost and it reduces maintenance. That’s where we’re going.”
Bill Cune of Corning, reiterating the importance of deep fiber, elaborated on the trend toward C-RAN where “The C could be cloud or centralized or coordinated. All of those things make the network work a lot better. To do that, fiber has got to go to more places, but it also has to have an architecture back to a centralized location that connects or hundreds or dozens of these sites in a given radius. That is some new. We’re in the very early stages of building out a fiber network that will really densify and enable the advanced features of 4G and 5G. All of this applies to indoor and out.”
Monica Paolini, principal of Senza Fili, described the trend as a move from atomic networks to pervasive networks. “We can’t just do more of the same to get more out of our networks. We all know that there is an increasing usage, everybody wants to use mobile networks, wireless networks. It’s clear that networks are asked to do more with basically the same level of funding.”
She continued: “There are different trends that are driving us to pervasive networks” that will increase flexibility and dynamic functionality. Today, “Different elements to different things…one cell, one area. Everything is pretty much very stable, but doesn’t really meet the demand that we have out there. Instead, pervasive networks are going to take us to converged networks at different levels. That will allow us to address demand in a much different way. Eventually we might be getting to a situation where there are no more cells–cells in the sense of separate units. There will be antennas placed in the environment providing ubiquitous coverage, but we don’t have the same distinction. We will move from a single-layer architecture to a multi-layer architecture.” And capacity will be made available based on the service requirements of a given application–this gets into the 5G idea of network slicing wherein a bespoke data pipe is automatically provisioned to perfectly meet the requirements of an application across domains from the device to the cloud.
Another factor informing 5G network architecture is the physical location of hardware and virtualized functionality. “Location becomes much more important because, depending on where you put a function, you have a difference in performance. Everything becomes more fluid and more dynamic. As a result, also the separation between core and RAN disappears. You have on the one end, The RAN end becomes virtualized, and so it moves towards a more centralized type of architecture. And then the core, with things like mobile edge computing, moves to the edge and becomes more distributed. This is actually taking us to KPIs like latency that are becoming much more important. QoE becomes more important than KPIs themselves so we need to understand what is the perceived performance rather than the performance itself.”