The post below, is an oversimplification of the subject for the benefit of non-technical readers and the general public.
In today’s telecommunication industry, the internal infrastructure of mobile operators comprises of a complex architectural implementation with a large number of hardware elements used to run the network.
The picture below is a simplified version of a 5G network showing just the most important elements that make a phone call, SMS or a download happen for you, the Mobile Network user.
The phone to the left represents your device and some of the badges with the network functions are represented by one or more hardware components such as servers routers, switches, cables e.t.c.
Some of the equipment are connected with each other mostly with cables and if any equipment breaks, it must be physically repaired, or physically removed and replaced.
The minute you press the “Call” button on your phone or you fire up your browser, some of the elements in the picture above use a little bit of their processing power to do the heavy lifting and serve your request, routing the traffic you generate through the system wherever it needs to go.
If you add all the available processing power from all the CPUs that the servers, routers, switches in the network have available, we get a total number. We will call this number total available network resources.
Because there is a finite number of total CPU power the operator has in their network, the total available network resources number is also finite and fixed.
Unless the operator invests some money to buy new servers with new CPUs to add to the pool, the total network resources remain the same.
Also, every individual network element’s available resources are finite and fixed too. This means you cannot use one server’s CPU to do the work for another server’s workload, or use one router’s CPU to do the routing of another rοuter’s traffic
For years, this is how the network was traditionally setup and this poses certain limitations on both the operator and you as a user.
The main limitations for the operator are:
The user is also affected by these limitations too.
As the mobile network user population grows, the total available network resources remain the same so at one point, the network usage demand exceeds the available network resources. This results in bad network performance and poor user experience. We will call this the scalability problem.
Another problem is that if the network can serve 1000 people in a neighbourhood, and suddenly 1500 people are trying to download simultaneously, everyone will get slower download speeds, and possibly some users will fail to connect.
This will happen even the rest of the city’s network is completely unloaded, because the available network resources per device are finite and not transferrable. We will call this the resource un-elasticity limitation.
In order to solve the scalability and resource un-elasticity problems in the traditional network architecture, the operator will need to physically add/replace/upgrade all the affected network elements in order to increase the available network resources.
This is not only time consuming, but also incredibly expensive and usually the cost is passed onto the user. This is the investment cost problem.
A solution to the investment cost problem presented above, is to virtualise the hardware components themselves.
According to this implementation, all the functions of a network device are analysed, copied and then imitated in the form of a software program that does exactly the same thing as the actual device.
You have experienced a real life example of virtualisation yourself if you have ever taken a computer based exam. The computer program replaces the pen and paper so you can say that the pen and paper have been “virtualised”.
With many of the hardware components of the mobile network turned to software components the maintenance /upgrading /removal any of those devices can be done just by someone installing a new program on a computer saving cost and time.
To address both the scalability and resource elasticity problems, the telecommunication industry is utilising the current advancements of the Cloud Computing infrastructure.
In the simplest terms, cloud computing means storing and accessing data and programs over the Internet through a cloud service provider instead of owning all of the infrastructure.
If you’ve ever used Gmail, Hotmail or Yahoo mail you’ve experienced Cloud Computing in real life. In this case, you are using your browser on your home device to access a program, say Gmail which also contains all your emails, attachement, contacts on Google’s cloud platform.
The cloud service provider offers on-demand CPU usage on a huge datacentre filled with an vast amount of servers.
The network operators therefore, have started the adoption of Cloud Computing infrastructure to run their virtualised components.
This way, they can take advantage of the scalability of the CPU assignment and instead of upgrading their hardware physically, they just increase the CPU count on their existing contract solving the scalability problem.
This also addresses the resource un-elasticity problem because in our previous example, when 1500 subscriber download simultaneously in a neighbourhood, the operator can instantly add a few more CPUs from the cloud provider on the already virtualised hardware to increase the available network resources. This will enable the network to achieve network resource elasticity.
The solutions proposed above, are still in development and there is plenty more research to be done in order to ensure that they will actually address the problem.
Our research will therefore create a virtual environment where a component will be virtualised, and the properties of a cloud infrastructure will be utilised in order to investigate potential gains with relation to mobile radio network performance.