The FTI’s report dedicates one full section to the evolution of network infrastructures, noticing how much the pandemic has emphasised the importance of having a pervasive, scalable network.
As you read the report it becomes clear they are recognising that for the coming years 5G will remain the focus. Big Telcos are still claiming that 5G is the solution to all known needs, as well as to those not yet known (this is important since most of today’s known needs would be satisfied by LTE and this “old” system is expected to dominate the market for the first half of this decade, 5G might take the upper hand in 2027 – the actual date will vary depending on the region, 2027 is for the European region).
The current wireless network, a melange of 2-3-4 Gs (5G is marginal), has been able to face the challenges posed by the pandemic in most Countries. In Italy traffic has doubled on the fixed network, with a sharp increase in conference calls that are particularly demanding in terms of network quality, and has increased by 50% on the wireless network with no significant glitches. The Telco Operators have been able to double in a matter of hours (48-72 hours) the capacity of their backbones to face the surge of traffic showing that current networks can scale … graciously. This emergency also showed that the “current” networks are ok and this is somehow embarrassing since it has been claimed in the past two years that only the deployment of 5G would meet the increasing demand ….
On the other hand researchers have started to look into what might be coming next (the name is already clear, 6G, what it should deliver is also clear: more capacity, lower latency… exactly the same stuff that has been claimed for 4G, 5G and that will be claimed for 7G!) and this is double embarrassing since it underscores that 5G is not the solution! Don’t worry, it is all a deja vu.
The report is trying to surf on these two embarrassments, the focus on 5G and the trend towards the 6G.
5G, particularly used as an alternative to fixed networks, FWA – Fixed Wireless Access-, is a great solution particularly for those (many) areas that are not covered by an efficient wireline network (fibre) and even more for those that are not covered at all. This again varies from Country to Country and from region to region. In Italy we still have several areas that are not well covered and that would be too expensive to cover with a fixed network. A broadband wireless solution (still requiring a fixed network but a much more smaller and cheaper one) is a good alternative. Likewise for several sparsely populated areas of the world where satellite networks could meet the demand.
Now, meeting the demand is really the crucial point and in several cases the problem is not the network coverage but the lack of (sufficient) demand. In some cases it boils down to the chicken and egg problem, there is no demand hence Operators are not investing in the network, but if there is no network there’s not going to be a demand.
If you look at the image proposed by the ITU as justification for research on 6G, you’ll see that the demand side has names like holographic media, holographic services… That is nice and true: holography requires plenty of bandwidth, in the Gbps, and very low latency. However, devices supporting holography are what is needed most. And they are nowhere to be seen. May be ten years from now we will have holographic projectors in our homes, may be in twenty years time we will have holographic contact lenses…
Hence the question I placed on this post: When fast is fast enough?
The answer is … it depends on what your devices and your needs are. With present devices (smartphones) 4G is plenty in terms of speed. Not necessarily in terms of capacity, since the throughput of the cell is shared among the users of that cell at that particular time. LTE Advanced (that is the latest version of 4G) supports 300 Mbps of download speed and 150 Mbps of upload speed. If you can show me a service that needs more let me know. I would love to discover it.
Of course there is more than smartphones. I keep hearing of robots in assembly lines needing very low latency, future cars that need to talk with one another again with very low latency and why not, drones that are also (supposedly) requiring very low latency.
Now, with 4G you have a minimum (average) latency in the order of 50 ms, with 5G you can have a minimum (average) low latency of 5 ms. The point here is the “minimum”. The fact is that what you get in terms of latency depends on many many things: is the server busy? Latency can go up to seconds. Is the network congested? Latency can go up to seconds again, for sure to hundreds of ms. Notice that these are not dependent on the G version, they apply equally to 4, 5 and whatever G.
Also, the distance from the source of data/services introduces a delay that of course is more significant as latency decreases: if you have a great 5G connection introducing 5ms delay and you are in Rome whilst the server you are accessing is in the UK you add, courtesy of the electromagnetic field propagation, some extra 5 ms, hence you double the latency. If the server is in New York you add some 21 ms and if it were in Sydney you’ll add 55 ms, that is over 1,000% increase. Again this has nothing to do with the “G” being used so if you are starting from a 4G 50ms latency you have a 10% increase when connecting to London and a 100% increase when connecting to Sydney.
As you see the point of latency becomes relevant when we are looking at short distances, basically at the network edges (and this is why there is so much interest in edge cloud, edge computing…) but this does not require a mammoth network deployment to serve very specific edge requirements. Industry is looking at local solution for their factories, with local private networks tailored to their specific need. Engineering (and paying for) a Country wide network to meet requirements for just a few users in very specific areas does not make any (economic) sense.
The conclusion? Unless we are going to have in the mass market devices that require speeds/capacity in the 10 of Gbps range we really don’t need a 6G network. On the other hand there are still two triangles to explore, the ones pointing to new architectures and new IP.
That’s for tomorrow’s post.