What is a network function? How is the core 5G network different from previous generations? That's what we will be looking at in this video. When engineers define a mobile network, they identify the functions and distribute them across different nodes. From 2G to 4G, they were trying to minimize the number of functions and each function carried out a wide range of tasks with the different elements in these tasks nested within one another. The functions were monolithic. In general, each function was managed by a specific piece of equipment, by a specific node. For each pair of nodes that interacted, they specified the reference point. For example, between the MME and the HSS, we find reference points S6a. For each reference point, they define a protocol stack, for example, S6a is based on Diameter over the SCTP transport protocol over IP. But for reference point S11, between the MME and the Serving Gateway, the protocol stack is different. We use GTP-C, GPRS Tunneling Protocol in the control plane over UDP, over IP. For specifications, engineers defined message formats in such a way that you would minimize the size of the message. When a piece of information had already been sent during an exchange, they tried not to send it again. And in addition, when we consider the exchange rules, they tried to limit the number of messages sent. That was because, particularly for 2G, the nodes were limited in terms of processing capacity and the links between nodes were also limited. This is no longer true with 5G and we have a different approach. We specify network functions, but rather than trying to minimize the number of functions, we try to specialize them, they will be more functions, but they will be specialized. For example, the MME was replaced by the AMF and the SMF. Each network function, or NF, offers a kind of common interface, which means the same protocol stack is used everywhere. And we will use HTTP/2 application protocol. In terms of message format, the organization is the same for every exchange. We use the term service-based interface because each NF here, the UDM, offers an interface to access a service. The interface is referenced by a sequence of letters. It's pretty logical. It starts with N for new generation and then the name of the NF, so NUDM for the UDM. The NF acts as a server and we call this the service producer. Any other NF can use the service and, that's important, we no longer think in terms of pairs of nodes. However, an NF can use the service only if it is authorized to do so. The authorization criteria can be defined by the operators and are not given in the specifications. When an NF send request to another NF, we refer to the requesting NF as the service consumer. For example, we can say that the services provided by the UDM are used by the AMF and the SMF. The AMF and the SMF are service consumers. However, when we analyze the situation, although the principle is that any NF can be a service consumer of an NF service producer, in practice, only very few NFs really use the services of another NF. For example, in the case of UDM, only the AUSF, the AMF, and the SMF use the UDM's services. That is to say that we can find the definition of reference point for a pair of NFs. The reference point is also identified by a code, here N13. It should be noted that at the heart of a procedure such as setting up a PDU session, we use the services of the SMF via the NSMF interface. It's the AMF that will use these services. But throughout the procedure, the SMF send requests to the AMF too, which means that the AMF will also behave as a service producer. So, within a given procedure, each NF can alternate between a service producer and consumer. When we consider all the NFs in a network, this gives us the concept of Service-Based Architecture, or SBA. So what does it mean? A service-based architecture is a system architecture in which the system functionality is carried out by a set of NFs providing services to all the NFs that are authorized to access their services. The definition is a little complex. What it means is that when we consider the SBA, we don't represent the reference points, but instead, we illustrate the fact that each NF provides a service via an interface and that's all the NFs can access to this service. It should be noted that this representation might make you think that all the NFs are on the same bus. But of course, this is more of a mind map and not a physical representation of the connections in reality. Frankly speaking, it's a way of masking the complexity of interactions. In the next video, we will find out how the services are structured.
