For those unaware, there are proposals across the world to sunset 3G services. In some areas, this has already happened. For others it may happen this year or at dates in the rest of the decade. It may be that 2G services are retained for a while longer to offer at least fallback call and texting services, but in many areas, they will close with 3G.

For most, this will go relatively unnoticed as they will have fairly recent devices capable of at least 4G. Problems arise with old, embedded or specialist systems - a similar situation to the EoL of windowsXP. Systems not capable of anything more than 3G service or too expensive to easily upgrade and those whose manufactures don't even exist anymore will have problems.

I think it may be possible to upgrade Touchpads to true 4G (rather than the H+ fast 3G of the Pre3) with a modem swap, but for everything else, your device is about to lose data connectivity - aside from all the problems arising from outdated SSL and certificates etc.

It seems to me that there are three possible solutions:

Replicating a 3G network locally that can connect to the main telephone system.
A coded work around to use another connection option (e.g. wifi / bluetooth hotspot).
Another hardware / software solution.

The obvious solution is to locally replicate a 3G tower. The upside is that once set up, the phone should work normally, without any changes. Mobile companies in fact used to do this in areas of poor coverage. A box called a "Femtocell" was issued to customers, which (presumably) plugged into the landline, routing nearby mobile calls into the network. Thoughts of acquiring old femtocells were quickly shot down: Use of this equipment requires a license from the authorities and it's unlikely mobile phone companies will issue them now unless there was significant need. The main reason for the shutdown of 3G is to free up bandwidth and unless fairly isolated, a 3G network may well interfere with new services filling the newly created frequency gap.

In most places, certainly urban areas mobile coverage is pretty good these days, but there are still some poor reception areas, including inside buildings. As wifi is also now fairly ubiquitous and smart phones can all connect to it, most mobile service companies now offer "wifi calling". This is essentially VOIP, but using a slightly different system and one that also allows connection out of the internet via a mobile tower or landline gateway. In theory, a SIM card allowing this service could be inserted into a legacy device and allow calls via home or even public wifi access points.

The problem of course is that legacy webOS devices pre-date wifi calling, so would almost certainly need some kind of code to at least interact with the SIM code to call over wifi. However, as a now common option, it's likely this is at least possible and if the feature is available on any open-source phone projects, it may be possible to port it to legacy webOS. It may even be that such code already exists in LuneOS, but I have no idea. Of course, LuneOS can run on devices that at least have 4G, so this option may be useful but not vital for that system.

A further option pointed out to me by Codepoet was this project: https://mbro95.github.io/PortableCellNetwork/ this might be possible for reasonably technically capable people. a 12 min video is here.

It's not clear what the hardware cost of this will be and I'm not sure how this is not a local 3G network - basically a homemade femtocell. The project seems to be aimed at emergency situations and services when commercial services are down and a local communication network will be sufficient. However, it might work. Licensing regulations are usually intended to prevent interference and if you could show that your set up was sufficiently isolated or local, the system might be overlooked.

My preliminary conclusion is that deploying wifi calling code to legacy webOS is the best and cheapest solution and adds greater mobility than a local (home) 3G access point, but I posted this to stimulate some discussion of possible options, so if you have opinions on what might or what won't work and ways to implement solutions, post below.