5G Concept & Architecture – An Introduction

What is 5G?

Hema Kadia  –  What are the key 5G technologies, related use cases & spectrum frequencies? 

Ben Cheung  –  The three big ideas of 5G, you know what makes 5G unique and different from 4G and what the future hopes to bring to the wireless industry. If you walk away with nothing but these three big ideas, I think I’ll have done my job. So, in 5G, the three big ideas are enhanced mobile broadband eMBB, massive machine-type communications mMTC, and ultra-reliable low latency communications URLLC. So, I know each of those is a mouthful and that it’s a lot of big long acronyms, but let me see if I can try to break it down. And again, those are really the big key ideas, and if you take away some basic core concepts behind those ideas, I think that that is really the main principle concept that makes 5g unique and different from 4G. 

Enhanced mobile broadband (eMBB)

So enhanced mobile broadband – the idea is that it will have a new and increased spectrum, improve spectral efficiency, and the ability to really have much higher data rates than we did in 4g. So it has 100 fold the amount of data transfer rates, the peak data rate of 5G for the download link of 20 gigabits/sec and 10 gigabits per second for uplink rate versus a hundred megabits per second. It will support high mobility up to 500 kilometers per hour or 310 miles per hour. Which is, you know, 50% more than 4g was able to deliver and a much higher area traffic capacity. So 100 megabits per second per square foot.  Now 5g, it has quite a wide variety of spectrum that will be available to it from the standards described 0.45 gigabits up to 52 gigabits per gigahertz frequencies.

Europe is currently looking at opening up the spectrum in the ranges of 3.6 to 26 gigahertz. The USA’s FCC Federal Communication Commission is looking at opening up spectrums into the 14 gigahertz ranges. It’s also split into high band, mid-band, and low band, with the low band in the 700 ish megahertz, which is comparable to 4g performance, but where 5g really shines and what’s very different and unique is when it gets into the 25 and above gigahertz, the millimeter-wave range. And that allows for these really high data rates. So what does that mean to you as a user? That means media anywhere, a broadband experience everywhere, anytime, and augmented reality and virtual reality experiences. So a lot more of the same but with much greater capability.  With enhanced mobile broadband, the ability has much higher access rates and transferred rates. In 4G, where you have 100 megabits per second, right around 2010 or so, you know it would have taken about 6 minutes, 5 to 6 minutes, to download a movie. With 5g with 10 gigabits per second, you’ll download a movie in under four seconds, so it’s about three and a half seconds. You know that that gives you a sense of the power of 5g with enhanced mobile broadband.

Massive Machine Type Communications (mMTC)

Next is massive machine-type communications. In massive machine-type communications, one of the key things presented there is the Internet of Things. So one of the things that characterize 5g technology that makes it very different from 4g is really the ability of machines to talk to machines. So 1g and 2g were all about people talking to people. And then, in 3g and 4g, we start having data and people interacting more with machines, streaming services, and video and audio. But we never really focused on this new unexplored space of machines talking to machines, and 5g caters to that with the mMTC concept. So you have connectionless radio access battery saving modes and specific standards to have machines talk to machines. One of the interesting things is that 5g technology promises to support connection densities of 1 million connections per square kilometer or two and a half million connections per square mile. So that works out to I did a back-of-envelope thing about one connection per square foot, right?

So you’ll have that 100-fold the amount of Connection density than 4g has. Now the concept of IoT isn’t new. It goes back to 1982, actually, when CMU Carnegie Mellon University first connected a Coke machine to the internet. That was the first connected internet appliance. In 2008, that was a tipping point where basically, there were more things connected to the internet than people. So, those things are not necessarily talking to each other in any meaningful way. And 5g is one of the gateway technologies that will open up a whole new world of smart devices and devices that can be aware and talk to each other. So IoT is actually split into kind of a sort of four big categories. One is Consumer IoT, where vehicles, smart homes, wearable technology, connected health, and appliances talk to each other.

And then there are things called Commercial IoT. You have medical devices, transportation-type systems, vehicles, and buildings talking to each other. Then there is Industrial IoT, also known as IIoT, So; there you have like sensing, processing information actuators, and things and now farming things and industry talking to each other. Finally, Infrastructure IoT, where you have devices on bridges, devices, and rail trains, monitor and talk to each other. So, massive machine-type communications involve lots of connected devices, meaning machines talking to machines and communicating through 5g technology.  And again, you know, the massive machine-type communications with a million devices per square kilometer, it’s estimated you’ll have 10 to 100 devices per person. And though they’re all talking to each other, so not just your appliances like radio, and car and refrigerator, but also in a factory where you have, we could have lots of devices and sensors all over a city, I can tell you, you know, where there’s danger or where there’s to help you find your way and things like that.

Ultra-Reliable Low Latency Communications (URLLC)

Finally, URLLC – Ultra-reliable low latency communications, so one of the key things you should take away from this talk is that the latency of 5g is one millisecond round trip time.

That’s the promise of 5g, right? So there are lots of things that there are lots of technologies and lots of that you’ll hear about in a future fireside chat, in which we’ll talk about how we can achieve that super-low latency. And that super-low latency, though, is, you know, what does it mean to you, right? Well, in 1g to 2g to 3g, or we have about 150-millisecond latency, your ear can detect a difference of about, you know, about that range, right about, so at 150 milliseconds, if I were to skip audio, you could tell that difference, or you could hear that difference below 20 five milliseconds, or you wouldn’t be able to tell. But then we go into, say, 4g, where we have 90-millisecond differences. And there, your sense of sight can tell, right? So, you know, why is the refresh rate of screens at 60 hertz, right? So, each frame on a screen is on the screen for about 17 milliseconds or so. And, there, you know, your vision, your sense of vision, can sense that difference.

When we come down to 5g with the one millisecond round trip time,  that’s haptic, right your ability, your sense of touch, can detect that difference. So we’ve really come to that super-fast round trip time opens up all sorts of interesting applications, the ability to remotely control devices, and there’ll be lots of things that contribute to what’s called industry 4.0. You’ll see in many white papers, and Things talk about industry 4.0. We’re really talking about smart factories and devices that can also make use of the massive broadband, slicing, and Internet of Things to have all sorts of smart things in factories. You’ll also see lots of there’s there are hundreds of applications if you look online for 5g, some simple ones are like remote surgery and examination where a doctor can, you know, look at a patient in a battlefield situation halfway around the world, or smart infrastructures in smart cities where buses can know where to go or driverless vehicles can navigate a city smartly, or where some something can tell you where there’s no parking space open dynamically. And smart, automated vehicle control, so the ability to be able to use him those things that you see in movies where like James Bond is driving a car through cellphone, you could actually do that. Not that you would do that. But think about this, what about a first responder who wants to drive a robot into a fire without risking himself to save somebody? That kind of thing is possible with 5g technology. URLLC will open up a huge avenue of new applications, which will allow people to control things at that haptic level. 

High-level 5G standards architecture

Hema Kadia  –  Can you elaborate on key elements of the high-level 5G architecture?

5GPPP – 5G Architecture

Source: 5GPPP – 5G Architecture white paper, Feb 2020

Ben Cheung  – Yeah, so this diagram shows you sort of three levels, the service level, the network level, and then the resources and functional level of a 5g network. Basically, a service provider will be creating end-to-end service and end-to-end service operations at the service level. And they’re trying to determine the optimal way to set up a network. And at the network level, one of the interesting things that 5g brings is this concept of slicing. Now slicing existed in 4g, but it was manual and fairly limited. In 5g, it promises to be much more dynamic. A slice basically allows you to set up a communication link with a group of users for a specific data flow, differentiate service quality, and prioritize these types of flows.

For example, you could have a public safety slice, which would give higher priority to first responders and fire, you know, firefighters, dedicated network functions, and segmentation. And then, basically, it takes in input things like throughput and latency, reliability, and security. And it can look at those requirements on a particular kind of slice and define or tailor things to use that slice. Here are some examples where vehicles are talking to each other about the concept of platooning; we’re pods of vehicles traveling together. Smart utilities are a connected city where basically, you might have an IoT slice. In fact, if you want to read more, it’s in 3GPP, TS 23.501. The 3GPP standards actually define three basic slices. And you can imagine those three basic slices are, no surprise, eMMB, mMTC, and URLLC. 

So I could have a slice just for people talking, or you know, doing their things with it that they normally would do with their smartphones, a massive machine type slice where basically I have a slice dedicated to the Internet of Things devoted to things that are communicating, not so often on the network to preserve battery life and their devices talking to each other who have special needs. And then, as you can imagine, a slice relay related to ultra-low liability, low latency communication, somebody controlling, you know, a robot, let’s say, far away. So now those licenses basically are controlled through the SMF, the session management function, the UPF, the user plain function, the AMF, and the access management function, and the UPF is basically can select a policy it has that works with the unified data management and also interfaces with the online charging systems.

The user plane function basically defines connection points, routes packets, and forwards information and can detect service data flow and handles per annum per flow, quality of service, and traffic usage reporting. And the access management function is all about basically keeping track of your mobile. You can register it to ensure that the UE is always reachable and establishes and releases, you know, control plane signaling and handles your mobility. So in 4g, we had the mme and the serving gateway. Those are essentially replaced by these AMF, UPF, and SMF functions that dynamically allow you to control slices. And then finally, at the resource and functional level, we see on the left here; basically, the mobile devices, and it could be lots of different kinds of things, you know, cars talking to the network or phones talking to our smartphones or things in a factory, so the little factory arm there, and so they’re basically talking to mobile and the showing towers there. Basically, in 5g, depending on the low bands, you might have towers. You might reach a mile or miles or more. But when you get to millimeter-wave communications, you know, 24.25 to 52.6 gigahertz, those towers are going to look like little radio heads, basically, and they only have a reach of 50 to 100 meters.

So they’re going to look a lot more like the router in your house, right? That’s maybe on an antenna or a telephone pole. So, those connect to an edge cloud. And the concept of an edge cloud is basically because, to achieve that sub-millisecond latency, you’re going to have to bring some of the computing closer to the end-user, right? And that’s what the edge cloud tries to do. It tries to reduce the latency. One way to reduce the latency is to have all that computing power to do the modulation demodulation, all those baseband processing type stuff closer to the end-user. And that’s what the concept of the edge cloud is. And then, of course, you go through the backhaul, the wider transport network that allows it to connect to the central core, which has all those switching things that get the data packets and where it needs to go. And then, on the far right, you see the management domain of resources and functions. So and the assurance fulfillment orchestration. Basically, there will be certain kinds of quality assurance that have to be fulfilled based on what you’re trying to do. So certain slices will have different needs.

The orchestration layers are what they do because they’re responsible for allocating a certain amount of resources once you figure out what you need. It says I need this much computing power; I need this many resources to accomplish a slice that I’m trying to set up. So there you’ll see RAN orchestration, Core orchestration, Transport, NFV, and MEC and infrastructure orchestration; those are all geared to basically trying to set up the right resources dynamically for the kind of thing you’re trying to do. So there’s lots of that you can look up a service base architectures that describe how to orchestrate those resources to achieve what you, you know, to come together to achieve what you need to do. So that’s the basic. And you see, it’s also, you see, at the top, upper left, there is called a recursive model. And that basically means that you can set up something, and it can dynamically scale; you can use it again, in a smaller, localized thing or at a much larger scale, and it scales easily. That’s the basic concept there. 

ONAP – 5G Architecture

Hema Kadia  – Can you provide a perspective from ONAP RAN architecture?

Ben Cheung  – Yeah, sure. So ONAP stands for an open network, automated platform. And it’s really a collaboration of dozens of companies coming together. It’s one of the primary things that I’m working on now.

It first started with AT&T open-sourcing their ECOMP platform. The basic idea behind ONAP is to provide a network management layer, or a (Service Management and Orchestration) SMO in RAN terms, to allow for management at that service level to manage a whole network from disparate vendors to a single common platform. It’s truly a platform, just like your laptop is a computing platform. What do you what could you do on your laptop? You can do almost anything. It’s just whatever you want to load on your computer. Thus ONAP is similar to that concept. It’s a platform where you can also upload analytics. You can add new microservices to perform analytics, machine learning, AI, and so forth. But it also has basic network management functions to perform life cycle management operations, orchestrate, and get a real-time dynamic view of your network through the active and available inventory. We are working together with the O-RAN standards and 3GPP to have common FCAPS-type functionality to allow devices to register to ONAP automatically. And it really is an open-source project, so lots of vendors are coming together to create this common platform that, in the end, will make it much easier for service providers to manage their network without having to have, you know, different EMS/NMS box for that particular vendor’s devices.