What happens then as a user device UE, a handset makes a phone call through the comp network is that information flows through the radio, whether you train her hands node B into, while the MME is going to interact with it a little bit from a control standpoint. But the signaling gateway interacting then with the Packet Gateway into the control function or the IMS and finally, if necessary, destined out to the PSTN as opposed to routing those types of devices out through the P-GW into the Internet. So, we touched a little bit on the EPC. There's going to be some other sections we dig down a little bit deeper into it. But again, not unlike the IMS, the EPC itself is actually a collection of functions and those functions are identified based on the need of their operational aspects in the network and they may be replicated in many places. So, again, it's not unusual to find the EPC in today's network consolidated into 10s or 20s of millions of subscribers per instantiation of the EPC. Going forward though we've recognized the fact that we can take functionality and network slicing and move functionality for example, the MME functionality or an S-GW and move it closer to the edge to a particular bits of function on that network slice. So, similar to the IMS, we can talk about the EPC into two different areas; the control plane entities and then the data plane entities. The control plane entities, the MME are responsible for knowing about the device itself, knowing about the usability of the device; tracking its location, assigning IP addresses if you alter the channels, they're going to be used for in recovering those IP addresses as that device either is no longer using a channel as it moves from base station to the base station. Depending on those diagrams you look at you're going to find the HSS and the PCRF supporting the EPC. Some authors may claim that the total proper parts of the EPC themselves, but they are certainly critical in the operation of those devices, the signaling plane control devices as well. Again the PCRF comes in from a charging and ability standpoint, and in the future we might even see it come into a quality of service type of operations. Then down below, the signaling gateways, both the S-GW and the P-GW, and in some cases you may find that functionality is actually unified and from a platform standpoint, depending on the functionality of the load or it can be distributed if it's a very high load type of environment, you may find them that those are actually realized on a multiple systems, multiple solutions and those may be a little more difficult initially to virtualize. Then where the control plane functions but there are a lot of vendors out there that have made significant progress in virtualizing the EPC. One of the other areas that we find virtualization is a key driver for is inside the Customer Premises Equipment or the CPE, particularly, for the enterprise. Now, you're going to find some authors call this a hybrid CPE, H-CPE, universal CPE, or virtualized CPE and it goes by lots of different names. But, what this really then is that collection devices are to provide their communication infrastructure, termination capability and the translation that into a local area network at the enterprise level. Most of us are going to be familiar with some type of a router device in our home a home gateway or something like that. At the enterprise level, similar functionality only at a different scale and possibly with additional requirements that are in there. Inside this CPE, we find not just the translation from interface A to interface B that is comps interface to a local interface and in the natural NATing that takes place from a router standpoint. It's in there, but VLAN capabilities, encryption capabilities for VPN and maybe deep packet inspection and a variety of services or functions that can be deployed onto those platforms and either managed by the enterprise or by the CPE provider, the communication service provider itself. In addition, we're going to talk a little bit in some future sections about the SD-WAN capability and that's sort of an overlay or a control aspect that allows us to apply policy shaping or traffic management, if we've got multiple interfaces or multiple capable tagging inside a similar face that we see there. So, some of the drivers for that are the enterprise or the automation of provisioning a service. Moving more to a standard high-volume server cuts server on that and then allowing the service provider to offer multiple services whether that's keeping your firewall up to date or offering multiple firewalls to different users based on their requirements. You may have one financial institution whose CFO has decided that they like vendor A's firewall and another one who likes vendor B and the comps service provider can provide both of those services and ensure that the security requirements are met and that those software elements are delivered, updates are delivered in a timely and managed way into that type of an environment. This applies most to the enterprises will have a branch type of an operation, multiple locations.
