Let's look at some of the key enablers for network slicing. Starting from the top, we have software defined networks, which is critical to enable programmable, un-simplified and automated operations of network management. Network function visualization abstracts the hardware and provides scalability and flexibility to execute network functions where they are needed. On the radio access network, you can use different types of spectrum, different type of radio access technology, to deliver the specific service that is required. In terms of devices, some new type of devices will come with 5G, whereas in 4G and previously it was mostly the smartphone and similar type of devices. Now, there will be a new type of devices that will come to market that will be able to receive these different type of services being delivered by 5G. From the point of view of end-to-end security, different type of services may have different type of requirements. So, having end-to-end security service level agreements will be important. With just a little bit of management already, and it is really critical to be able to manage a 5G network that is very dynamic and very adaptable to ultimates of the typical processes required in network management, such as performance management, fault management, et cetera. Another key aspect of network slicing will be energy. Energy consumption is an important aspect for every service provider, there is managing of 5G network, and for the user, is also very important that 5G delivers battery life at least equal or better than previous generations. And finally, resiliency and robustness, that some use cases require, such as, autonomous driving or remote surgery. Let's look at how 3GPP has architected the 5G system to enable the implementation of different types of networks slices. On one side, they have clearly desegregated the user plane functions from the control plane functions, which enables each type of function to be scaled independently and to be located in the network. Let's have a look at how 3GPP has architected the 5G system to enable the implementation of different types of network slices in the 5G network. On one side, they have clearly desegregated the user plane functions from the control plane functions. These enable each type of function to be scaled independently and to be located in the part of the network where it is require. This type of modular design enables a flexible and efficient network slicing, by taking advantage of the SDN and NFV principles. At the same time, it was also critical to clearly separate the access parts of the network and the core part of the network, and ensure that network slices could be implemented on both sides. With the advent of network transformation, the 5G networks of the future will be comprised of distributed data centers, where the network functions defined by 3GPP can be placed as software functions. There will be different layers of these data centers. On one side, there will be the typical core network, big data centers, where some of the more centralized functions will be placed. But there will be further layers closer to the edge, such as regional data centers and local central offices, where compute platforms will be placed and some of the network functions that require more distribution can be placed. This will be specifically very relevant for the management of end-to-end latency for different types of services. We will typically see that the base station is located less than five milliseconds from the user, while the local central office is within ten milliseconds usually. So, this type of services that require extremely low latency will need to be placed in a more distributed manner in this type of local central office compute platforms. Regional data centers are located a little bit further out, and therefore, the latency will be a little bit wider typically within 40 milliseconds. For those network functions that need to go all the way to the Cloud, latency is to be expected to be around 100 milliseconds.
