In the previous post in our Techsplanation series, we talked about some of the ways that mobile network operators maximize their use of spectrum bands. In this post, we talk about core networks as part of larger mobile communications technologies. As before, please refer to this glossary for quick reference to some of the key terms and concepts (in bold).
This post is the third in a series all about mobile communications networks. In the first post, we learned a bit about the electromagnetic spectrum and how we use it. In part two, we talked about some of the ways that network operators get the most out of the spectrum they use. We highly recommend reading those posts before you read this one, but it’s okay if you don’t.
You really should, though. We’ll wait.
Ok, so now you understand the basics of how your mobile devices communicate wirelessly and how carriers make that work. Shall we proceed?
Sure. So my phone sends signals to and receives signals from my carrier’s antennas, and those signals are then passed along to their final destination. But you told me that the frequencies that mobile devices use can carry information only so far, so what happens when I call my cousin on the other coast? Or when I access a web server located…wherever?
For most modern networks (4G, LTE), these streams of information, whether carrying the sound of your voice or the pieces of a website, travel over the internet. (See the first Techsplanation for more on how that works.) We have already talked about the RAN (radio access network), which is the wireless part of the network between mobile devices and the carriers’ cellular antennas. But what carries information between those towers? That part of the network is called the core.
Today, most communications networks employ a packet-switching architecture. For example, most of the data traffic on the internet is sent using something called the Internet Protocol (IP), which works by breaking up data into small packets. Increasingly, voice communications are also sent using IP, hence VoIP (voice over Internet Protocol) and the wireless equivalent, VoLTE (voice over long-term evolution networks).
So basically, once wireless signals have been converted into electrical current or pulses of light, they travel over wires or fibers in these IP packets to get closer to their final destination. If that destination is another wireless device, the packets are converted back to radio waves and transmitted that way.
Regardless, the first stop for communication data traveling through the core network is a switching station. It is this station that decides what to do with your call, text, or web query. Before we move on, there is one other network that we haven’t talked about yet: the public switched telephone network (PSTN). To the extent that it exists, the boundary between the PSTN and the internet is increasingly fuzzy. For this post, you can think of the PSTN as the network connecting everything with a traditional phone number. So if you are calling a landline, or an international number, the switching station helps route your call accordingly. We won’t cover how this works, exactly. For now, the important idea is that different nodes in the network serve various functions, like switching, routing, and interconnecting multiple networks.
I think I understand. The wireless part of mobile networks also connect to wired networks, some or all of which are actually part of the internet, and within those wired networks there are boxes that do different things to help complete my call or retrieve my email. Let me guess…it’s complicated?
How did you know? You are correct, it is rather complicated and there are lots of functions that the core network performs that we will not cover here, such as controlling which carrier’s RAN your phone connects to, determining whether your phone is actually your phone or an imposter, tracking your phone from one location to the next, handling voicemail, and so on. What we will talk about are some of the ways that core networks are changing to support more functions and meet standards for 5G. (Remember, 5G is industry nickname for the next-generation, super-fast mobile network architecture. The next post will focus exclusively on 5G capabilities.) And to do that, we’ll have to introduce even more acronyms.
From a core network architecture standpoint, the three biggest transitions toward 5G are software-defined networking (SDN), network function virtualization (NFV), and mobile edge computing (MEC). All three of these technologies already exist and have been deployed (to some extent) in active networks. In the next post, we’ll talk about why combining them in the 5G context will be a big deal, but for now we will try to explain what they are and what they do.
For a long time, networking equipment has been proprietary and static—dedicated to a single function or limited set of functions. In engineer-speak, the “control plane” (the part that defines how to route information) and the “data plane” (the part that sends bits along the correct route) were merged together. This worked fine, except that this equipment could only do the one thing it was built for. You could not reprogram them to do something else, or to do what they already do slightly differently. This meant that changing how your network worked was difficult, time-consuming, and expensive because you would have to replace or reconfigure each piece of equipment one by one.
But with SDN, the control plane is separated from the data plane. This means that you can just reprogram the control plane and leave the data plane to take appropriate actions based on the new controls. Even better, the control planes for each of the many, many boxes (pieces of networking equipment) distributed throughout the network can be reprogrammed from a central location. This enables network operators to change the way their networks handle and route traffic more quickly and efficiently, which means that they can respond to changes in network usage as they happen, rather than changes taking days or weeks.
In some ways, the transition to SDN is like our transition from a bevy of dedicated purpose gadgets (camera, phone, GPS unit, calculator, etc) to a programmable device capable of performing many functions (a smartphone or tablet with different apps). The apps are like the new control plane instructions (you can create new or different ones depending on your needs), while the operating system and processor are like the data plane (they will carry out the app instructions).
On its own, SDN has real potential to make networks more agile and dynamic by making it cheaper, easier, and faster to reconfigure network nodes. But when combined with the next two technologies, SDN could help networks become something more than communication conduits.
Okay, I’m with you so far. SDN lets carriers make their network nodes into whatever they need, and lets them make those changes from a central office, so it’s much faster, cheaper, and more flexible. Right?
Right. Sophisticated SDN even allows the rapid deployment of multiple software component packages, a little like the apps on your phone, so that operators can more easily “plug and play” their SDN modules. Hang in there, dear reader, only two more highly technical aspects of networking to cover in this post.
Network Function Virtualization
Before we talk about NFV, let’s discuss the concept of virtualization, generally. Creating a virtual “thing” involves modeling and running that thing in software instead of hardware (so in bits instead of atoms). This abstract software-based “virtualized” thing runs on top of the physical hardware in which it resides—for example, a computer within a computer, a network within a network, etc. This is how a cloud computing instance works—a program sets aside a certain amount of processing power and other resources (usually on computers located somewhere else) and that chunk works as a standalone machine, but “virtualized.” There can be many virtual machines operating on a single hardware unit, or a single virtual machine utilizing the hardware resources of many physical machines.
You could also make a virtual machine on your home computer; sometimes people do this when they want to run a program without exposing the entire computer to risky software, or to run software that only works on a different operating system. In some ways, virtual computing is like a fantasy sports team—those players don’t physically play on the same team together (the pieces don’t all reside in the same computer), but function as a team in the abstract.
If you are a regular reader of Techsplanations, then you have already learned about virtual private networks (VPNs). In that context, the “thing” being virtualized is a network, only it is not a physical network of wires, boxes, and radio transmitters, but rather a network that is logically separated—not physically separate wires, just separate connection rules—from other network traffic traveling on the same physical networks.
Oh yeah, I have a VPN! I use it virtually every day… Get it?
Yes. Ha ha. Classic.
So, as I was saying, a virtual “thing” is a non-physical abstract creation, residing or operating in/on/over an existing physical structure. A virtualized network function, then, is a thing or set of things that a network can do, except instead of using a dedicated-purpose physical device, the function is enabled by software and uses whatever mix of network resources it needs. Now, it is possible to implement NFV without using software-defined networking, but SDN gives operators the ability to adjust, move, remove, reconfigure, and change the scale of virtualized network functions from a centralized orchestration platform. So if a network operator wants to create a firewall between two segments of a network (to block harmful traffic from one side or the other), it can send instructions to the general-purpose unit at the desired node to describe what the firewall should do, how to operate it, and for how long.
These virtualized functions can be customized or even strung together to create a custom service that the network (or part of the network) provides. For traditional network functions (moving information from place to place, blocking bad packets, creating secure paths, etc), SDN and NFV help operators to fine tune their networks and adjust them dynamically to meet different demands. But network operators can add other gear to their networks, too, such as data servers and general-purpose computers. In fact, many networks already have these things built into them. For example, some access networks and core mobile networks incorporate racks of servers to support the local caching and delivery of popular content. Look back at Techsplanation Part 1 and this post for more discussion of content delivery networks (CDNs).
The capabilities of these hardware components can also be incorporated into a virtual machine to provide even more functionality.
Like what? I mean, I get the idea that SDN and NFV let carriers change the way their networks work more easily, but what else could a network do, other than move information between endpoints?
That is one of the biggest questions posed by the capabilities of 5G networks, and we’ll dig into at least one possibility in the next post. For now, we will talk about one more ingredient that will be part of the 5G mix, mobile edge computing (MEC), which we’ll describe below. A quick note about predicting the future: exactly how and for what uses carriers might deploy these various capabilities (SDN, NFV, MEC) remains to be seen—they, like us, are likely trying to figure out the answers to these questions.
Mobile Edge Computing
Mobile edge computing is exactly what it sounds like. Small computers are distributed all around the “edge” of mobile networks. (In this case, the “edge” means the part of the network sending traffic directly to users, at or near cell antennas or base stations—not between cell towers or between other elements of the “core” of the network.) These computing stations are much, much smaller (and have far less computing power) than modern cloud data centers, but they offer a different advantage: they are much, much closer to you or whatever device might connect to the mobile network. This means that information sent for processing doesn’t need to travel as far and overall trip time (latency) can be far lower, making the network very snappy. Snappy communications (low-latency communications) are a crucial part of some existing and emerging applications, such as augmented and virtual reality (AR, VR) and some kinds of autonomous vehicle controls, such as driverless vehicles communicating with city infrastructure like traffic lights.
MEC also means that the network could perform some kinds of computation instead of the mobile devices doing them, potentially reducing the need for powerful on-device computing and increasing battery life. It is difficult to predict all potential impacts of this capability, but it could lead to smaller, lighter, and cheaper devices. For example, imagine a phone or other device that only needs enough battery and computing power to send/receive wireless signals and display the results, but that offers the processing power of one or more edge compute nodes (likely far more powerful than current on-device processors). This could allow devices to smaller, lighter, and more simply designed, while also employing the computing power of the network.
So, what your saying is that there might be a computer attached to every 5G cell and that it could do computations on behalf of connected devices? That sounds cool, but doesn’t that leave carriers in control of deciding what applications to support with their MEC capabilities?
Yes, or probably yes. There are still many open questions about how the new capabilities of 5G networks will be deployed, used, offered, shared, and sold, but adding new capabilities to access networks definitely increases concerns about the gatekeeping role of access providers. We will discuss this more in future posts, but be sure to read the next installment of Techsplanations in which we will combine all the parts of 5G that we have covered so far and talk about why 5G really could be a big deal, even if it doesn’t yet live up to the hype.