Millimeter Band? Beamforming? This is a 7-part text that takes you deep into 5g – the next generation of mobile networks.
How is data transported and processed in a mobile network? How does the signal go from your phone, to the radio mast and out to the internet?
What really sets a 5g network apart from 3g and 4g? What are millimeter bands, beamforming and network slicing? And where in the net does Ericsson’s investment in artificial intelligence end up?
5g is the first mobile network to use the millimeter bands
The mobile screen shows a racing game that is too powerful to run on a smartphone. The phone is firmly seated in a fastener that protrudes from an Xbox control.
Since 5g is designed for industrial customers as much as for ordinary end-users, many of the applications have been developed for that purpose. But computer games are a grateful example if you want to demonstrate how 5g differs from UMTS (3g) and LTE (4g). While we have become accustomed to streaming music, film and TV series in a short time, almost wherever we are, it is only during the 5g era that we will be able to do the same with games.
Thus, the racing game is not run locally on the smartphone, it is just the interface for picture and sound. The engine that makes the game work is on a server in the cloud.
It is a data rate that is about ten times higher than what we normally get with today’s 4g network. But it is no match for 5g, which under the right conditions can deliver 20 gigabits per second. 20 gigabit is the same as 20,000 megabits and downloading a high-resolution movie at that data rate would take a few seconds.
5g is the first mobile standard to claim what is called millimeter bands. They are called that because radio waves at frequencies between 30 and 300 gigahertz have wavelengths between 10 and 1 millimeter.
On the millimeter bands, the range is shorter and the waves are easily blocked by physical objects. Not only do things like walls and ceilings but also foliage and even rain can weaken them. On the other hand, there is more spectrum available and thus, potentially, more bandwidth.
If an operator today can handle network traffic at a bandwidth of 100 megahertz, it could reach up to 800 megahertz with millimeter waves. That’s at least the bandwidth Ericsson’s US operator customers are already taking advantage of, in what has become some of the world’s first commercial 5g networks.
With 800 megahertz maybe 20 gigabits per second can be made. Then people usually ask what the hell they are going to do with 20 gig. And that’s maybe the gut feeling you get.
But the thing is that if you get a 20-gigabyte feed right at the radio mast, it goes down as you move away. Somewhere here, far away, where you reach your 400 megabits per second.
Thus, the speed at which virtual reality can be streamed in ultra-high resolution.
Lots of small cells provide a faster cellular network
Already at the end of 2019, the early established commercial 5g networks had begun to emerge. But let’s jump to 2025, the year when Ericsson, according to its latest trend report, estimates that there will be 2.6 billion 5g subscriptions worldwide.
But the user with a 5g unit that is within the coverage radius is inside the cell. The user’s unit then has the ability to communicate with the radio unit and thus has what we commonly call coverage.
In this new 5g world, there are such small cells everywhere. On lamp posts. At bus stops. In factories. In the arena where the sporting event is going on.
They can be different sized (and can be called microcell, picocell or femtocell), have different coverage radii and have advantages and disadvantages compared to macrocells. One disadvantage is that a small cell has less coverage area than a large one. This means that it usually reaches a lower number of users compared to a macro cell. One advantage is that a small cell is easier to install (although it may take time to get a permit for installation).
But above all, it is close to the user. Telecommunications occur high up in the radio spectrum, on what is called millimeter waves around 30 gigahertz. And as we noted above, it is possible to get high data rates over short distances on the millimeter bands as long as there is free visibility.
– One reason for building with small cells is simply to get coverage where it is otherwise difficult to access. Modern buildings are very energy efficient, which of course is the top. But from a radio perspective it is sad because they are so well insulated. While we see that 80 percent of the traffic in a typical network is generated indoors, says Hannes Ekström.
Part 3: That is why massive Mimo and beamforming is so important for 5g
Two concepts that have become popular when talking about 5g on millimeter bands are massive mimo and beamforming. Let’s leg them out.
Mimo. Stands for multiple inputs, multiple outputs. You may have seen in the specifications of your mobile phone that it is equipped with 4×4 mimo? This means, if you talk about downlink, that it has four receiver antennas that can communicate with a base station that has four transmitter antennas.
One way to use this technology is that the Netflix movie you stream over 4g can be transmitted in downlink as four different signals on one radio channel. It is usually called spatial diversity or “spatial diversity” and provides redundancy. Simply put, it can be said that redundancy provides an abundance of signals if one of the channels fails.
Another way of using the technology is that four users can watch each Netflix movie by sending different data streams between the respective transmitter and receiver antenna (say four users is a simplification: the number of users is not necessarily equal to the number of antennas). Then the concept is multi-user mimo, or “spatial multiplexing” and thus provides increased capacity.
Mimo technology utilizes the wave-physical phenomenon known as multi-path propagation. By positioning the antennas some distance apart, the signals from the transmitters will – depending on the conditions prevailing in the environment – take different paths to the respective receivers.
Whether you choose redundancy or capacity, it is an effective way to make the most of the spectrum available.
Massive mimo. At higher frequencies, the length of the radio waves is shorter, which also means that the antennas can be reduced. Massive does not therefore refer to antennas that are massive in size, but rather that network builders can mount a “massive” number of antennas (these can be tens, hundreds or thousands of antennas), without having to significantly increase the size of the radio unit.
Beamforming. It is similar to the propagation of radio waves from a traditional macrocell as a large spotlight, with the difference that it is not visible light.
Beamforming is more like a laser pointer, which uses a large number of antennas to create a directed beam at each individual user. The more antennas, the narrower the beam.
Just as in the case of mimo, you gain the benefit of multipath propagation by processing the signals for the different antenna elements in an antenna set each, in order to obtain a so-called constructive interference. Then it is also possible to dynamically direct signals to be able to accurately follow a user, something called beamsteering.
Massive mimo and beamforming take into account the fact that more and more devices will be connected during the 5g era, and that the amount of data consumed will increase more or less exponentially.
With more antennas and directional beams, you can create higher signal strength and faster data transmission to individual users, but also increase the capacity of the entire network. Massive mimo, beamforming and small cells are, in short, a smarter way to utilize available spectrum – and purely decisive if you are to make best use of the advantage of millimeter waves.
With a small antenna just outside the window that helps amplify the signal, with massive mimo and beamforming from a nearby cell, Alice can get speeds that far exceed what is required to have a good VR experience. At least that’s a scenario.
You can play virtual reality games in different ways. Either directly in your smartphone connected to a wifi device, which is then connected to the antenna unit that sits outside the window. It communicates over 5g to a micro or macrocell. The other way is that the smartphone is connected directly to a micro or macrocell. If we start with the antenna in the window, it will connect to the micro or macro cell that has the best signal strength. Then beamforming will provide antenna directivity to the data flow, which increases the signal-to-interference ratio and can thus provide higher transmission speeds, says Hannes Ekström.
The signal-to-interference ratio Ekström mentions is a way for engineers to measure the quality of a wireless signal.
The radio network gets more base stations with 5g
Not long ago it was considered a very strange event. The team that calls this arena their home is playing this particular away game, in a town 30 miles south. Still, it’s full.
Everyone in the arena wears augmented reality glasses and the match is projected on the lawn below them. It feels like the players were there for real.
Let’s go because that scenario may still be a bit like science fiction. But the point is that during the 20th century we will be consuming more data. Huge much more data.
In Ericsson’s latest trend report, global data traffic is forecast to reach 160 exabytes per month in the mid-20s. In 2015, the corresponding figure was below 10 exabytes and in 2019 just under 40 exabytes.
Whether that means augmented reality streaming on empty arenas is an open question, but that video content will account for much of the increase no one doubts.
To stream high-resolution video requires high transfer rates (think autobahn). To stream video to 50,000 people on a small area also requires high capacity (think autobahn with a lot of files).
So even though small cells each have a small coverage area, the telecom industry’s strategy is to deploy many of them, close together. This is usually called “densification”, which can be translated into densification.
At the same time, a number of people in the arena may want to use the phone for something as old-fashioned as voice calls. Then the nearest macrocell and a low band of 700 megahertz can meet that need.
The signals to and from the users’ 5g units (smartphones, glasses, VR headsets and so on) thus pass over the air to the most suitable cell in the area. The cell consists of antennas which together with a custom built radio unit go under the abbreviation RRU (remote radio unit). It also consists of a BBU (baseband unit), that is, a baseband unit.
The baseband unit can be described as the brain of the radio network and, unlike the radio unit, is positioned a bit away from the antennas. A common scenario is that the antenna and radio unit are installed close to each other on a mast on a high roof and that the baseband unit is located in the basement of the high-rise. The radio unit and the baseband unit are then connected to a fiber or copper line.
The radio unit is like a conversion station between a digital and analog interface, or between ones and zeros and electromagnetic waves.
– The radio unit takes a signal from the baseband and amplifies it. It also adds a carrier frequency, because the generic signal must be up to radiofrequency. So it takes care of this radio can. But it also listens in uplink to be able to send a signal back to the baseband unit.
While the radio unit helps the antennas to transmit and receive signals, the baseband unit is responsible for determining how and when a signal is to be transmitted or received to make the network as fast and efficient as possible. It is in many ways a scheduler.
But the baseband unit is also the first or last stop, depending on the direction of traffic, in what is called the radio access network, or robbery. A RAN network consists of base stations and base stations consist of antennas, radio unit and baseband unit (as well as own cooling and backup batteries).
Radio access network is one of three important components of a complete mobile network. The other two are transport networks and core networks (telecommunications people never say core networks, but core networks).
It is an excellent parable if you talk about how mobile networks have worked so far, but as we shall see, it is somewhat less useful to illustrate a 5g architecture.
When it comes to the transport network, the term “backhaul” is often heard. Used as a verb, the English word “haul” can mean transport, transport or convey. Backhaul is the connection between the radio access network and the core network. The connection normally consists of a fiber line, but in recent years microwaves have also become a popular method, especially in sparse areas or in places where it is expensive to install fiber.
If several baseband units have been moved to a central facility that can be located one kilometer or more from the radio unit, the connection between RRU and BBU is usually called “front hole”.
– The radio unit and antenna should be close to each other to increase network performance. But then we have newer types of architectures also where we don’t put the baseband unit in the basement of the same house, but … it can be miles away. Then we put several units in one central plant. This can be for cost reasons, for performance reasons, or because you want to allocate resources by balancing the use of the various baseband units. But it is time-sensitive and requires fiber infrastructure if you want to build in that way.
The Core Network gets smarter with the help of artificial intelligence
Data to and from their devices is sent back and forth through the core network, which directs traffic between many different base stations, the Internet or various cloud services.
Ever since the mobile phone’s breakthrough on the GSM era, there has been a circuit-switched core network. Its most important function at that time was to keep track of where subscribers were and to connect them. The Core Network was – and still is – responsible for keeping track of basic things such as whether the subscriber has a SIM card and whether it has paid their mobile bill.
In conjunction with 3G, it became possible to send data between mobile phones and the core network was developed and adapted to provide even that type of traffic. You started talking about packet core networks, and if you ever encountered the abbreviation EPC it stands for “evolved packet core” – a common 4g corenet architecture.
There are different ways to describe a packet core network, but to keep it reasonably clear you can divide it into four components.
User plane. The data required for interaction and rendering of Alice’s virtual reality games. In short, it is here that the user traffic itself is transported and processed.
What kind of capacity and data rate does it need? What kind of subscription does she have? Where is she? This type of control signaling takes place here.
– If user plane is the road or street and user data corresponds to the vehicles moving on it, then the control plane is to be thought of as the traffic lights. It is on the control plane that it is decided whether a channel through the operator’s network should be opened for a mobile device.
OSS, or “operation support system”. It can hardly sound more boring than that, but aside from the exciting millimeter technology on the radio side, OSS is one of the most interesting parts of a 5g network.
OSS is an orchestration system that ensures that the products that are part of a mobile network work best. OSS keeps track of error signals in the network and any software upgrades are done via this system.
BSS, or business support systems. Business system: thus the part of the core network that, for example, handles invoicing. Here, among other things, there is a CDR, or “charging data record” – information about, for example, how long and where a subscriber used a payment service by the operator.
The mobile network is moving to the cloud
Physically, they have been sitting as blade servers (servers you set on high) in a refrigerator-like cabinet in a number of data rooms. For cost reasons not too many data centers. For redundancy reasons not too few.
During the 10s and 4gs, the core network has largely been virtualized. Then, operators have been able to exchange expensive custom-built hardware for so-called “off-the-shelf” products. You call it “virtual machines” (or just the VM), which means that with software you emulate a computer system, regardless of the hardware on which the software is running. Often the kind of x86 processors that you can find in ordinary home computers are used.
What the telecom industry did in practice was to mimic how the IT industry began to build cloud platforms.
– When 4g was developed, nobody talked about cloud. But then they started looking into data centers on the IT site and saw that it was possible to get much better resource utilization of the hardware through virtualization.
And once 4g had matured, giants like Google, Amazon Web Services and Microsoft had stopped talking about virtualization and started talking about “cloud-native”, “containerization” and “microservices”.
The term container could be called virtualization 2.0. Instead of emulating the computer system itself, the operating system is emulated. With this type of design it is possible to divide computer programs into separate instances, called containers. Then, the use of available hardware is further streamlined, since a container can run on basically only the operating system kernel and the resources required for the contents of the container.
Parables are common to this technology. A container should convey the idea of how physical containers are transported on ships. The name of the Google-developed and now open container system Kubernetes is taken from Greek and may mean captain, or helmsman. The Kubernetes logo represents a classic steering wheel.
This separation of code means that what was previously considered a single large chunk of software (imagine how smooth it would be if all the goods on a freighter were packed in one and the same huge container) can be divided into microservices. It is to build programs, IT systems or web applications where each function is run in its own instance so that it is easy to add or remove functions, or update a function without affecting everyone else. If a microservice needs to communicate with other microservice, an api connection is established between the containers.
One of the benefits of micro-services is that the development time can be drastically reduced. Partly because containers are small in size compared to virtual machines, and partly because they are faster to start.
We are deep in the territory of the IT world now, but these are design principles that the entire telecom world with 5g also claim. Ericsson, for example, builds its cloud infrastructure with Kubernetes.
– It is a better way to handle software, it is more efficient and it is easier to upgrade individual parts.
This development, where software can spin on standard hardware, and where program code is packaged portion-wise, makes the similarity between the core network and the trunk of a tree misleading.
Instead, you are talking about the distributed cloud where more functions or instances are placed along the network edge, close to the user. A 5g core network becomes a combination of central, regional and local clouds. A corenet will, as well as spread from the trunk, to the branches, and out to the leaves themselves.
For the telecom industry, such an architecture becomes important when it comes to 5g. Why? In order for radio technology to become so fast that the core network risks – unless parts of it are not distributed to the outskirts of the network – become a bottleneck for time-critical applications.
The VR game that Alice is sunk in runs on a 5g network where some of the core functionality is placed in a node that is geographically close to her home. Operators are likely to use different types of clouds – distributed and central, private and public – depending on cargo.
– You might have a private cloud where you run your core services, but the opportunity to reload to a public cloud, if you have an agreement with, for example, Amazon. If there is a huge load, maybe you can create a new control plan instance in the public cloud. And we at Ericsson have orchestration services that extend all the way into Amazon’s cloud, says Anders Rosengren.
Network slicing ”is one of 5g’s most spectacular concepts
Sometimes, you may not get the transfer speed you need. You could of course upgrade to the premium subscription where a certain performance is guaranteed.
The guarantees (data rate and capacity) are in the arena. The blue-light staff who are thundering outside have no guaranteed data transfer speed, but good capacity: their communication is secured despite the high data traffic nearby. And the hospital from which the ambulances started is also guaranteed a maximum delay of 1 millisecond. At least in the wing of the hospital where they just started remote-controlled surgery.
All of these guarantees go under the concept of “network slicing”, one of the more spectacular concepts in the 5g architecture. Where the operators’ customers can order a network with a certain requirement.
There is basically no limit to how many network disks can be assigned. In a self-driving vehicle that goes past the arena in the opposite direction as the ambulances there are three different disks. One with ultra-low delay for the vehicle to respond quickly. One with high data rate that belongs to the car’s entertainment system. And one that has lower requirements because it only passes non-critical data from vehicle sensors.
Network slicing also enables clear isolation between different discs. Thus: if problems occur in one disc, accessibility in other discs is not affected.
– The operator is given the opportunity to define a business chain through the network, which is basically resource allocation, says Hannes Ekström.
Of course, we are not in the middle of the new decade, but are just about to enter it. 5g has been launched in a few countries, including a very modest introduction in Sweden, under the direction of operator Tre.
This is 5g on the so-called non-standalone standard (NSA, which stands for non-standalone). Then 5g New Radio is used: thus new radio technology which is combined with already existing 4g core networks.
But in five years, everything we have listed here should be in place: millimeter bands, massive mimo and beamforming, a densification of the number of small cells, a mobile network built with the cloud architecture design principles orchestrated by artificial intelligence and network boards where the customer is guaranteed special features.
Then the 5g networks will be standalone (SA, or stand-alone), with both a new RAN network and a new core network.
It’s 5g – a fifth generation mobile network that will look nothing like it.
