5G is expected to open up whole new application areas that will transform everyday life all around the globe. No longer will mobile communication be confined to providing the mobile broadband that 4G enabled. We will see massive machine to machine communication and ultra reliable, high availability, low latency communication as well as enhanced mobile broadband. These new capabilities will enable 4K video streaming of immersive content, advanced automotive and traffic management, sophisticated energy management, advanced manufacturing and remote healthcare. 5G has to be able to deliver the high bandwidth, low power, masses of connected devices and low latency to meet all these requirements. It has to be able to cope with a vast array of differing demands and a diverse set of devices from the very smallest to most powerful.
How will all of this be possible?strong>Networks Well, if we could create dedicated networks tailored to each individual user’s need, based on their current demand, that would solve the problem but this has not been possible until now. One of the big advances in 5G is end to end Network Slicing. Network slicing is the ability to divide up the network to create smaller, virtual networks between the source and the mobile device based on the current demand and requirements. In this way each network slice can be tailored for bandwidth, number of devices, latency and so on as required at any given time by the application.
But it also needs to be adaptive, to tear down virtual networks and rebuild them as demand and requirements change. For example if I’m streaming immersive content, a virtual network will be set up to provide high bandwidth and probably low latency as well but, if I then stop streaming and look at my email instead then that network configuration is wasted. I don’t need that performance for email. 5G networks need to be able to detect these changes in demand and adapt accordingly to achieve this.
5G also promises the perception of infinite capacity; that I get exactly what I need, when I need it. No lag, no spinning daisy wheels, instant access. Using the network efficiently is crucial to supporting this and the massive increases in traffic that are anticipated in the coming years. A way to think about this is it’s like electricity. When you plug something in the wall you don’t think, “I wonder if there’s enough electricity to power this?” you take it for granted that there is. You have a perception of infinite capacity but there isn’t infinite electricity generation capacity, it’s finite. So how does it work? There are control rooms where the load across the country is constantly managed to available generation capacity. That’s what 5G will be like except it’s software that is constantly building, tearing down and rebuilding the network based on user demand.
This approach is though essentially reactive. What if we could anticipate changes in demand? If we could do that there is potential to be even more efficient as we could predict changes in the network before they happen. This is the promise of Artificial Intelligence (AI) and Machine Learning (ML). If we can use these powerful techniques to learn from past behaviour how to predict future demand there is huge potential to optimise network efficiency even further.