One radical question I'm considering for #6G is "how it can be developed to be core-optional?". In other words, how can the system be broken into separate elements - RAN, core, transport - that *can* work together, but can also be deployed independently because they are loosely coupled? I'm increasingly convinced that 3GPP #mobile networks have succeeded in spite of the core network, not because of it. The huge complexity and cost of getting to "cloud-native" #5G SA has been a continuation of a long downward trend. The issue is that the core has evolved over the G's from being a fairly simple switch and registry (MSC and HLR) in the 2G GSM era, to being a huge monster of multiple "functions" that supposedly acts as the nexus for various advanced services and capabilities. Instead, it's mostly a cost centre, not a revenue generator. It's a vendor-led excuse for all the pointless friction from failed ideas like #networkslicing and a lot of network policy complexity. It makes it difficult to create wholesale models, increases network latency by adding processing, and has put years on the deployment of later versions of the RAN. Fixed networks don't need "cores" of this type. There's some intelligence in nodes like the BNG, but it is nowhere near being the same type of immense monolith. It also allows the 95% of traffic that is going to/from the Internet or public cloud to take the shortest, fastest path. It's ironic but unsurprising that the most successful new 5G service has been FWA, which essentially mimics and extends a service that has long been delivered with 802.11 and proprietary wireless networks that are unencumbered by a core acting as a boat-anchor. The other area of growth for has been in #private5G networks, which typically use a straighforward "core in a box", or the equivalent hosted in a cloud. It's broadly equivalent to a Wi-Fi controller in that guise. A lot of the more-useful functions of a cellular core network could probably be transferred to an Open RAN RIC - which would also allow the vitally-important capability of local-breakout to the nearest fibre and IP interconnection, in the same way as a local exchange / head-end for fixed broadband. I realise that there's many in the mobile industry who continue to focus on things like NEF, NWDAF, SMF, UPF, AMF, PCF, NSSF and so on - and that's fine. If you can make orchestration and service-based architectures work commercially, then go for it. But for the 6G era, why not ensure that the latest versions of the radio can be deployed without that baggage if a CSP or enterprise wants? Create the system to be modular and with APIs, so a "naked 6G radio" can be driven by a Wi-Fi controller or a proprietary box? It would democratise 6G to new players. It could also enable much simpler federation between different network owners - at the moment, core network interconnects are complex, and represent a major security risk surface. Let's be heretical! Does 6G really need a core?
Gustavo Costa
4mo
Quite a thought provoking question. I like that (I like to think out of the box). But did you ever hear of "Conway´s law"? From wikipedia: "Organizations which design systems (in the broad sense used here) are constrained to produce designs which are copies of the communication structures of these organizations." 3GPP is organized is organized in RAN, SA, CT and working groups within. These WGs communicate mostly via Liason Statements (slowly, via small carefully crafted LSs). Thus, the prospects of a somewhat revolutionary approach coming like "core-optional" from 3GPP are slim, to say the least. Interoperability with previous generations are also both an asset and liability in this case. Thus, your question could also to a great extent be also phrased as "Will 6G be only the 3GPP 6G?" Last generation where another big system was tried it did not end up so well... . R.I.P. mobile WiMax
Roy Wright
4mo
The basic components for a decentralized environment are all available and not reliant on "3/4/5/6G". The primary components of routing between wide-area networks to regional hubs is all available through common place tech rather than the "next big thing". The software methods for mass data sharing to sustain a resilient backbone are also all present, with hybrid block-chain and P2P data partitioning. It can be deployed on a small scale to start with, being intrinsically extensible by virtue of the P2P model. The key, (as with most projects of this nature) is the "industry standards" for protocols. As long as nobody tries to impose a restrictive commercial grasp of the market, there is no good reason this cannot grow. Being a notionally "mobile" industry, I believe that agriculture and food security are primary beneficiaries of a decentralized, distributed environment resilient to disruption.
Heikki Almay
4mo
Originally the core was valuable as a mobility anchor - but today all apps can handle a change of the UE IP address. The key topic is enabling another authentication mechanism in addition to the legacy SIM that requires a central subscriber database. Asymmetric authentication would essentially move the user identity to the UE. Making SIM authentication optional comes with some strings attached. At least session keys need to be rethought.
Arno Koch
4mo
Many "core" functions relate to authentication, policy control, routing, subscriber profiles & charging - plus network exposure ("Network as Code"/QoS-API's, etc ). https://www.nokia.com/blog/taking-adaptiveness-in-qos-to-the-next-level-in-6g/
Hello, 🔹 For any network, you need to handle access resources, control the users, and connect the different components. So you need, in any case: RAN (or fiber, copper, or any option for connection to the end user), Core (to handle users, connections to other platforms, policies, profiles, and special features), and transmission in all possible ways to connect the core and RAN. 🔹 Since the first Gs, you have had a structure like that, so you could work on the different developments for each system (Core, RAN, and transmission) and define common interfaces between the systems. 🔹 The good point now is that with all the cloudification and softwarization, it is possible to deploy big parts of the functionalities in the cloud or virtualized ecosystem, making deployment easier. 🔹 I consider that for the next generation, the structure of cloud, CNF, and APIs will still be used. If you do not deploy a "box" for the core, you will still need core functions deployed somewhere. I have read about 6G, and the core is not yet defined; there are proposals to use cloud, add an A-Plane (AI-Plane), and include AI network functions. 𝒀𝒐𝒖 𝒘𝒊𝒍𝒍 𝒔𝒕𝒊𝒍𝒍 𝒏𝒆𝒆𝒅 𝒄𝒐𝒓𝒆 𝒇𝒖𝒏𝒄𝒕𝒊𝒐𝒏𝒔 𝒅𝒆𝒑𝒍𝒐𝒚𝒆𝒅 𝒔𝒐𝒎𝒆𝒘𝒉𝒆𝒓𝒆. BR,
Jim Forster
4mo
A related issue, or question, is: would these Private 5G/6G networks need to support telephony? Does it imply a new phone number? Or is it mostly about connecting to the Internet?
I am increasingly starting to believe the problem IS the tight coupling between access network to core network to authentication to one set of metrics to define industry success. I like Dean Bubley's earlier reclassification of a wireless network's job being to reach the closest fiber node possible. We see this already in fiber-first countries where traffic is primarily (80%) non-cellular.
In all honesty all we are doing is perpetuating a dialup access model in the age of broadband. Just like we are perpetuating a voice anchoring first mobility model in the age of data. The result is cost, complexity and enormous inefficiency to preserve legacy. The same is true in fixed networks albeit at lower complexity; 3GPP is trying to fix that too in BBF though 😉 Oh -BTW - we are now sobering up and realising that using a software model devised for very large scale apps terminating traffic to build distributed bumps in the wire was most likely a really bad idea….
Network technology and strategy leadership and consulting; 5G/6G standardization and open source Director
4moProbably makes more sense from the business perspective, but has no feasibility as of now, considering: 1. 3gpp org structure split into RAN, SA and CT, and absolutely no intention to restructure for 6G; 2. IP model as exists now which feeds up more and more useless developed functionalities; 3. The 3GPP already-discussed-6G reality, where, in the best case scenario, 6gc will be based on 5gc, but in the worst case scenario, a new even-more-"sophisticated" core will be developed for 6g; 4. AI, energy efficiency etc hype and "need" to be implemented everywhere; 5. ORAN RIC applicability and relevancy - while not being a part of 3gpp's 5G or, most likely, upcoming 6g (for all good, bad and geopolitical reasons).