The concept of a 5th generation of mobile networking is gaining traction – not only among policy makers but also in technical standardisation work.
The 3rd Generation Partnership Project (3GPP) started work on the next generation architecture (SA2) late last year, and in March the work on the radio layers (RAN) is expected to move to the phase where the actual technical standards will be defined. While many 5G concepts do not immediately touch the Internet, some of the new concepts and the expected increase in network speeds, capacity and mobility will have some impact on our industry.
The work in the radio layers is set to profit from much higher frequencies in the millimetre bands and will benefit from more interference robustness, increasing the potential throughput to gigabits per second. Together with other changes, such as direct communication between devices instead of via the central base station, the new standard is also expected to decrease latency.
One of the big concepts introduced by 5G is “networking slices”. Recognising an ever more diverse base of users, devices and applications, a 5G network is expected to adapt and fine tune network parameters to closely match the application’s needs. That way, assisted by a high level of virtualisation of the network functions, a small sensor (such as will become increasingly common as part of the Internet of Things) that only sends a few bytes per hour but requires a ten-year battery life, can share the network with an Ultra-HD video device that might require speeds in excess of 10 gbit/s.
Whether such speeds are really achievable remains to be seen, of course, but regardless, it's likely there will be a bit more separation between applications based on their needs – and that separation might not stop at the device. The architectural debate already focuses on methods to separate different traffic flows from a single device and forward those streams via the different “slices” that are available.
On the conceptual level, such methods are of course not new; traffic classification, quality-of-service, priority lanes, tunnels and virtual private networks all are iterations of essentially the same idea of making forwarding decisions based on criteria other than finding the shortest path for a particular destination IP address.
How these developments will impact the work of the RIPE NCC or the RIPE community in general is not yet clear. As the standardisation work continues and some of the draft outputs become available, we will hopefully get a clearer picture. Recently, our colleagues in the IETF started a mailing list that aims to identify any gaps in the current standards in relation to network slicing; an initial zero draft is available.
In the context of RIPE Policy, as the architecture aims to have a device connected to different logical networks, the actual means of separation are still open for discussion. Some of the proposed solutions aim to have this separation at or even below the IP network layers. In such a scenario, a single device could end up with a number of different IPv6 subnets, some of which might originate from different networks or which might not all be visible in the global routing table.
We are not there yet and commercial implementations might still seem far away, but as opportunities arise to participate in and give input to the relevant standardisation processes, it is the right time to think about the potential impact this could have on you and the RIPE community.