I’ve been hearing a lot about telco 6G, and Nokia has published a blog on the recent advances in 6G formulation and the schedule of activity. The first market rollouts are scheduled in 2030, and in any world but telecom standards that would be more than enough time to suggest major changes be made. Obviously, we’re in the standards world with 6G, though, and so I think it’s likely too late to see any significant shifts. The question is what that lack of reconsideration might mean to the market overall. Nokia says “The 6G train has left the station.” OK, but is that good or bad? It depends on where it’s heading and who’s willing to ride.
The background of 6G is as important as its goals and plans. 5G was considered by over three-quarters of operators I’ve chatted with as having failed to make a business case. A bit under 15% thought that the whole 5G process was a vendor manipulation of standards to raise their own revenues, since almost all said a primary motive for 5G deployment was competition. They also thought that market hype generated by vendors are what made it competitively impossible for them not to deploy 5G.
This attitude is why operators have been pushing for a 6G model that was software-driven rather than one that displaced 5G gear, though of course the radio elements of 5G would likely need to be updated whatever else was protected. The problem is that this is clearly not in the interest of vendors, and vendors are countering, say operators, by pushing operations benefits to justify hardware impact. Cost reduction, in both capex and opex, is one of the top motivations the 3GPP has presented for 6G.
The top stated motivation, as always is the case, is new services and use cases, since this is what would first drive revenue gains and second engage the media to create public buzz over the migration to 6G. The problem is that the goals are things like augmented/virtual reality, AI services, and more slants on IoT. All of these were raised as motivations for 5G except AI, but the real problem with them all is that they are applications whose networking dimension and investment requirements are far below the requirements in the IT and business process spaces. 6G not only can’t drive these, but has to wait for others to do so—perhaps many others. Even some of the less futuristic and more network-contained goals, like convergence of wireless, wireline, and satellite communications, has no currently validated incremental revenue opportunity for mobile operators.
The 3GPP has, in at least an indirect way, acknowledged that 6G has to fix some thing that were done wrong with 5G. One such thing was migration from the earlier generation. 5G migration from LTE, many operators feel, pulled too much “legacy” into 5G, which limited its early utility. 5G architecture actually added to LTE complexity, partly because of the migration issues. Unfortunately, there doesn’t seem to be clear recognition of some of the real issues. 5G’s vaunted advances were features anticipating a justification; network slicing is perhaps the clearest example. Too much emphasis developed on user-visible improvements to service, and none of these ended up justifying any higher pricing for 5G. The biggest problem, IMHO, was that trying to facilitate migration kept too much LTE-and-IMS-think embodied in the architecture. All of these things created “genetic load” in 5G that will weaken 6G no matter how they’re dealt with.
Truth be told, almost everything from the past generations should be scrapped insofar as implementation is concerned. A handset designed for 5G has to work with 6G, and there’s even a lot of LTE IoT stuff out there that has to somehow be supported. But the goal of 6G, first and foremost, has to be to do everything from deployment and management to mobility management in the right way for modern software/cloud technology to support.
Device support and registration, mobility management, and access-method independence demand, in my view, a separation of the user relationship to content and other users to be independent of where the user is and how the user is connected. Location-independent routing is an element of this, but at another level it’s really about understanding that a “session” or logical user/application relationship, is really not the network’s business at all. Only where the user attaches, the “network service access point” or NSAP, should be involved in any kind of “mobility”, whether it’s geographic/cell-site or service type. The majority of the mobile-network RAN infrastructure, from the tower to the PDN/Internet connection, should simply be IP networking. To make this happen would likely require some additional router features that fall into the purview of the IETF.
To me, this separation is the most critical piece of the network of the future, and thus should be explicitly mandated in 6G, which so far it is not. It’s not that the idea hasn’t been kicked around; I remember a “Project Icarus” back in the early 2000s that had much the same set of goals for session independence of location or network, but I can’t find any references on it today. The question is how this important goal can be met, especially if operators don’t want to fork-lift infrastructure for a 6G upgrade.
One thing we may have to accept is that things like cell-site switching for mobile users might not be as undetectable as current systems and goals describe. Most users don’t care if they have a second or two lag in a voice call as a phone moves to another tower, but they don’t want the call dropped. Same with data connections; a switchover to another cell or even another service is just like an error—you recover from it and move on.
Service migration in wireless networks, to be effective, has to migrate the user and what the user sees as a service to the future service set in a non-disruptive way. Trying to preserve infrastructure or standards can only be addressed if that higher goal is assured. We didn’t do that in 5G, and if we don’t do it in 6G we face a major problem not only with its adoption but with the way network services will support our businesses and lives in the future.
The key to getting the 6G train on the right track, I think, is a better conception of openness. Open RAN was constrained by 5G’s model of infrastructure, which still has too many specialized elements. We know how to host stuff efficiently now, and how to develop software to optimize this knowledge in application to real operator concerns. We need to use it in 6G, to step away from a mobile network with boxes of functionality to a mobile network with hosted functions. That can then be truly open, and even be enough of a benefit to justify a bit more investment in taking the 6G step.
