By Heikki Almay, Co-Founder, Poutanet
Current mobile networks are awfully centralised. This is a big handicap for many non-public network use cases – and potentially also a blocker for many exciting public network scenarios. Especially when 5G or 6G is to be used for replacing a high-speed Ethernet connection, the current networks pack your traffic into a GTP tunnel and carry it to a core element – which might be located nearby but more often is not.
I do not want the sensor data from the factory line to go to the company IT cloud but to the control unit at the other end of the hall. Nor do I want the 6G VR LAN party, where you have friends with subscriptions from different service providers, to have all the game data sent back and forth between different distant clouds. I want the data to be switched locally as quickly as possible. That is the whole point to achieving low latency. There simply is no way to build a radio interface that would compensate for the delay caused by 100km of fibre as well as a few active elements along the way.
Reliability is another topic. There are many more opportunities for things to go wrong if you carry traffic across networks when compared to offering the service locally. Adding redundancy and fast failovers is easier said than done. 5G was supposed to provide ultra-reliable low latency communication, but if you were waiting for URLLC to fix your networking problems when 5G was hyped five years ago you may have given up by now.
There are a number of additional arguments for allowing user traffic to leave the base station directly and why the core should be made optional. Some of these are discussed in my earlier Market Needs of the 2030s article. Here the focus is on rearchitecting the traditional mobile network architecture to support decentralised operation and to point out a few approaches that seem to work.
Symmetric authentication is legacy
In 2022 the Roam Networks Web3 project showed a demo where a previously unknown user connected to an LTE network. The attach succeeded and the user was able to purchase a data subscription using tokens in his crypto wallet. The latter part is on YouTube. This was just to demonstrate that in mobile networks there is a genuine option for the user to interact with a smart contract for buying a service and for the user identity to be bound to an app instead of the SIM card. This includes the authentication key, the nasty secret shared between the SIM and the central database of the mobile service provider.
While the Roam Networks demo used tweaks to short-circuit SIM authentication it would be more than desirable to have an asymmetric authentication as part of the 6G design. From an engineering point of view it is a minor addition to add a private key to the dataset of the SIM and to allow the user (or UE) to decide whether to authenticate with the network using the traditional methods, or to a nearby base station using public key cryptography. Naturally some specification work is needed, for example on deriving session keys from the asymmetric authentication, but it is a relatively small effort.
Changing authentication could make roaming and BSS history
Public key cryptography would also provide a shortcut to the fairly complex roaming procedures and excessive charges. Instead of sending the authentication request to the home network of the user you could just decide if you trust the party that has issued the key/SIM trying to attach to your network.
If we look very closely, eSIMs already contain a private key – but that is for the over-the-air provisioning of SIM profiles. That domain is controlled by GSMA and looks like a minefield from a private network perspective, as you are expected to use someone in the GSMA camp to do your eSIM provisioning. As an enterprise I would like my authentication data to remain in my organisation. If there needs to be an authority the government would perhaps be a suitable actor. You would get a digital identity at the same time. This would of course raise the debate about privacy.
Obviously the private key could also be stored outside of the SIM – but it is hard to imagine that hardcore telco traditionalists would approve a spec that allows keys that do not come from a tamperproof media. We need a compromise that does not force user equipment manufacturers to decide whether to support the old (huge) operator ecosystem or the new (initially tiny) liberal networking camp.
If you are working for a business support system vendor (or own some shares) you may become worried at this point. Today the subscriber data and all information about the available quota sits in some large central databases. In a decentralised world the data (e.g. what I am willing to pay for my connection per gigabyte or per hour) could be carried in an app on the phone. The app would be happy to share that information with any suitable network that I trust and to strike a deal with a smart contract. Payments would be in tokens or using traditional credit cards. No need for a massive BSS – and no extra robotic process automation for fitting my 6G charges on the same bill with the legacy fixed internet, 2G and 4G … and perhaps 5G if my service provider does not decide to stick to NSA forever.
Another benefit of moving away from the charging model is transparency. The mobile operators collect vast amounts of call records from their zoo of old and new network elements and feed these into a massive billing system for calculating how much they can charge you. As a consumer you now have to trust that your service provider does the counting right as you typically do not have the means to challenge them. If on the other hand software on your device makes short contracts the network you connect to, you have the option to walk away and look for alternatives when you or your software come to the conclusion that the connection is not what you are paying for.
Decentralised 6G is the best we can afford
If 2030 is expected to be 6G prime time it might be that very few operators show up for it, if 6G is yet another next generation where huge investments for wide population coverage are needed. 80% of 5G deployments are still NSA. 5G millimetre waves are still waiting for users. At the same time traffic growth is slowing down and telcos and vendors are reducing staff. You can expect money to be tight and the people needed for trying new stuff long gone. A decentralised approach that would allow 6G to be deployed one base station at a time seems economically much less risky than the all-in campaigns seen at the early days of 4G and 5G.
Long time reader, first time contributor. Love technology and the great outdoors. Looking forward to discussing everything beyond 5G and the future of wireless technology!