At the moment, most network operators scrambling to deploy 5G are doing so based on EN-DC (EUTRAN-New Radio – Dual Connectivity) which is also referred to as ‘Option 3’. As the longer abbreviation suggests, this way of deploying 5G uses the ‘legacy’ LTE radio network as a base and connects a 5G radio network alongside it. In fact, with Option 3, the ‘legacy’ LTE radio is in charge and the 5G radio network is merely used as a sort of side-kick to increase data rates. In the core network the LTE EPC, the LTE Evolved Packet Core, continues to be used with only minor changes that focus on controlling the higher data rates and which subscribers are allowed to use the 5G radio network side-kick. This is a nice setup to start with as it is backwards compatible and does not require any handovers between 4G and 5G when running out of 5G coverage. But it’s far from an ideal and pure 5G network that is envisaged in the future with a service oriented core network and a 5G radio network that stands on its own.
So how do we get to a pure 5G network in an orderly fashion that doesn’t break backwards compatibility for LTE-only devices and without unduly limiting the air interface capacity for LTE devices in the process? There is unfortunately not a single answer and not a single solution for all circumstances. In this and the following post I’ll have a look at what I think are the most likely migration scenarios over the coming decade and why.
The 5G Core Network as the Basis for Everything
Let’s start with one thing that will be common in all migration strategies, a 5G core network, also referred to as the 5GC. And here’s the first catch already. Mobile devices do not only communicate with the radio network, or the Access Stratum as it is also called, but also with the core network, or the Non-Access Stratum (NAS). As a consequence, 5GC capable devices need to implement a new NAS protocol. Without that, a mobile device will be limited to Option 3, i.e. it can only communicate over an LTE eNB to an LTE EPC and hope that the network is kind enough to add a 5G cell to the connection. For mobile broadband connections, the 5GC works quite similar as the EPC but its structure is significantly different from its 4G predecessor. Have a look at my multi-part 5GC introduction post for details.
Options, Options, Options…. Option 2
A few years ago, the 3GPP representatives of Deutsche Telekom put together all potential 5G options in a slide set and assigned a number to each of them. Have a look here for the slides, I’ll stick with a textual description here. In theory, the ideal end state for 5G is Option 2 in which a 5G access network is connected to a 5G core network. When running out of 5G coverage, a mobile device would have to make a handover to a 4G network with a 4G core network in a similar way as there is a handover from LTE to UMTS today.
So far so good but there is one major disadvantage: Most spectrum below 6 GHz that is available today and also in the future is already used by LTE networks for LTE devices. Going directly from an Option 3 deployment to an Option 2 deployment would mean to also transition a large part of the spectrum from LTE to 5G NR at the same time to keep data rates for 5G devices at the same level as in an Option 3 dual connectivity scenario. This would not be a good idea for many years to come as the majority of devices will be LTE-only(*) devices or EN-DC Option3 capable devices(*) at most which would be cut-off from spectrum in pure 5G Option 2 areas.
So let’s have option 2 and option 3 run in parallel at the same time, would this fix the problem? This would work from a technical point of view but would have a similar disadvantage as a pure Option 2 network in an area. While devices going for Option 3 connectivity would have the combined spectrum of 4G and 5G of a location available, devices attaching to the 5GC with Option 2 would only be able to access the 5G spectrum. As an evolution that hardly works as it would be a step back. So until the day that the majority of spectrum has been re-farmed from 4G to 5G, Option 2 is not much of an option at all.
Going directly from Option 3 to Option 2 also wouldn’t work from other angels either: From a core network point of view it would not be possible to deploy a 5G core network for many years until a critical mass of 5GC capable mobile devices are available in practice which would have never connected to a 5GC before. And from a mobile device point of view, nobody would bother to implement the protocol stacks required to communicate with the 5G core. And even if somebody would bother, protocol stacks would be largely untested, inviting massive interoperability issues once Option 2 is switched on. In other words, it would never happen.
So there must be other ways to slowly transition from Option 3 in the short term to Option 2 as the long term ideal target architecture. There are three ways to do this, Option 4, Option 7 and spectrum sharing. More about those options in the next parts on this topic to follow soon.
(*) Note for completeness’ sake: Obviously devices still support 2G and 3G as well but that’s not relevant in this context