The 5G Core – Part 8 – Session Management and Network Slicing

And yet another 5G core network post, this time on session management and network slicing. I already had a post on network slicing from a radio network point of view a while ago but at the time it wasn’t quite clear how the 5G RAN will interact with a 5G core for network slicing. Recently, I found an interesting resource that clears this up from a core network point of view.

Have a look at this slide pack over on Slideshare by Sridhar Bhaskaran on 5G core network slicing and its LTE Release 13 and 14 predecessors DECOR and eDECOR. The basic idea in all cases is that different applications use different core networks. A practical example from today’s networks is LTE and NB-IoT. Both radios use the same base station and the radio same channel. LTE is used by smartphones, tablets, etc., while NB-IoT is used by Internet of Things devices that only have little data to transfer and must only consume very little power. Hence, the LTE and NB-IoT air interfaces are very different even though they share the same channel and the same base station. On the network side a single core could handle LTE and NB-IoT, or two different LTE core networks (EPCs) could be put in place. One EPC is optimized for small data transfers with a huge number of devices while the other EPC is optimized for large data transfers from a much smaller number of devices. The base station, the eNB, connects to both core networks. This is what is referred to as DECOR (Dedicated Core Networks, which was the name of a 3GPP work item in Release 13).

In practice an LTE device can only do NB-IoT or LTE and only be connected to one core network at one time. The 5G network slicing concept on the core network side adds further flexibility by allowing a device to connect to several core networks, i.e. to several network slices, simultaneously. This plays together with the way the 5G air interface is built with “localized transmissions”, which allows a 5G device to not only use one air interface at a time but to use several air interfaces simultaneously, i.e. one for fast Internet access and one with other properties. Now what exactly will be done in practice with this additional flexibility is another matter but in essence, network slicing in the core can be described in one sentence:

Allow a UE to connect to more than one core network at the same time

And this is also the reason why the 4G Mobility Management Entity (MME) was split into the Access Management Function (AMF) and the Session Management Function (SMF). A UE is always only connected to one AMF but can be connected to several Session Management Functions (SMF), one per slice, i.e. one for each core network that it communicates with!

Things are slowly getting clearer!