In the previous two parts on the topic I’ve been looking at why there is a need to evolve the initial 5G network architecture in which 5G base stations are merely a side-kick to the LTE access and LTE core to a pure 5G access and a 5G core network. There are several ways how to evolve from the initial state, and the options that seem most likely to me are ‘Option 4’ and ‘Option 7’. Both options have in common that they use a 5G core network and that 4G and 5G radio networks are combined to increase the throughput for high speed Internet users. Over time, spectrum is then piece by piece transitioned from 4G to 5G as the number of 5G devices that can talk to a 5G core network increases. While these options look nice they have a weak spot for network operators with little low band spectrum.
Imagine a scenario in which a network operator has only little low band spectrum, lets say 10 MHz in 700 or 800 MHz and wants to transition from 4G to 5G. This spectrum is used today for LTE and acts as an anchor for LTE to which additional spectrum is added with Carrier Aggregation if devices can receive a signal on higher frequencies as well. This is done because the reach at higher frequency ranges is much shorter. When a network operator introduces 5G with Option 3, the 10 MHz continue to be used for LTE. 5G resources in other frequency bands can be added. So far so good.
The next step in the 5G evolution is to introduce a 5G core network (5GC) and mobile devices that implement the new 5G Non-Access Stratum protocol so that they can talk with the new core. At this point the LTE base station then has to support both Option 3 and Option 7, so legacy 5G devices can talk to the 4G core while new devices can talk to the new 5GC. Moving to a 5G only network architecture would then mean switching from LTE to 5G on the low band channel and thus dropping support for LTE-only and 5G Option 3 devices. This is obviously not possible for a very long time to come without severely disgruntling a sizeable part of your subscriber base.
A potential solution to this undesirable situation might be ‘in-carrier LTE+NR coexistence’. Little has been written about this transition technology so far, but at MWC 2019, Ericsson publicly announced that they will add support for this on their radio platform. The idea behind co-existence is that LTE and NR will run on the same channel, with a part of the time domain being assigned to LTE and another part assigned to NR. To make this work, LTE-only devices must never know that this is going on, which means that certain periodic downlink LTE signals must be sent even during times when the NR part occupies the air interface. This obviously means that there must be a tight integration between the eNB and gNB.
This way, LTE-only and 5G Option 3 devices could use the LTE part of the channel while 5GC capable UEs could use the NR part. In practice such an approach would make sense if the partition between LTE and NR is either very flexible or if the percentage of 5GC capable UEs is already high enough to justify removing a certain percentage of the spectrum for them in a semi-static fashion. Concerning flexibility, the Ericsson press announcement linked to above has the following to say:
“For every millisecond, the split of simultaneous 4G and 5G capacity is adjusted to secure an optimal performance for any mix of 4G and 5G active devices in the network. This minimizes spectrum wastage and results in the best end-user performance.”
Another reason to go in this direction could be to make better use of network slicing, e.g. to have different kinds of 5G air interfaces which could be useful once things like NB-IoT have been ported in the standards from 4G to 5G and devices are available to leverage it. Don’t count on this happening in the next few years.
I’ve found some further information about this here, and there are a number of 3GPP input documents on the topic. However, there’s not a lot to work with yet from a higher level point of view. So the only way at the moment to find out more seems to go directly to the 3GPP specs and dig through the relevant parts. If you can find them.
So what do you think, will we see LTE/NR co-existence on the same channel in practice or will operators just go for Option 7 in practice and ride it out till the very end?