Newer phone networks are much more efficient than old ones. For a certain amount of 3G spectrum the carrier can serve more customers at higher data rates with 4G or 5G. From their viewpoint it is expensive to keep serving 3G customers.
So realistically, this shutting down of 3G should have happened when 4G came in if speed demands for spectrum were what mattered. But of course back then there weren't enough users to really justify it even if the speed argument applied more then than now.
2G and 3G standards have their place and keeping a 2 MHz channel open for them is not going to get in anyone's way, on the towers, or spectrum-wise.
At low-band, it's often quoted as 20%, but when you get to mid-band it's a lot more. That's one of the reasons there's been a mad-dash for mid-band spectrum.
I'd also note that the radio standards are made to evolve. LTE wasn't nearly as efficient as it was when it was introduced which is why, for a time, HSPA+ networks were able to compete with LTE.
> but of course the telcos are now realizing this spectrum, even the half gigahertz they stole of C band, doesn't actually propagate well around Earth instead of straight up and down
The propagation isn't really a surprise and the idea it was stolen is terribly biased. The incumbents on that spectrum were paid huge sums for it and moving expenses.
It does propagate well enough that it will be very meaningful. T-Mobile is seeing amazing gains using the 2.5GHz spectrum that they got through their merger with Sprint. They already cover around 60% of the US population with mid-band 5G. 3.5GHz won't propagate as well, but there are ways to overcome that like having the towers transmit at higher power and then having lower-frequency spectrum handle the uplink from the handset.
The three carriers didn't spend nearly $100B for something worthless and they aren't realizing its propagation as some sort of surprise.
> 2G and 3G standards have their place and keeping a 2 MHz channel open for them is not going to get in anyone's way, on the towers, or spectrum-wise
Running a 3G UMTS network requires 10MHz of spectrum. 10MHz of spectrum can be pretty meaningful. CDMA uses 2.5MHz for voice and 2.5MHz for data and maybe you're thinking of that, but it's still 5MHz to have both voice and data on a CDMA network.
Things can certainly be kept around, but it means less for other stuff.
That's an interesting claim. What aspect of the 2-4 GHz stuff gives 5G spec a greater relative spectral efficiency up there? The only thing I can think of would be tighter specifications for equipment phase noise and phase noise going up with frequency and becoming the main problem.
At least thats what uk operators told me
Newer generations increase the spectral efficiency, so more throughput can be supported on the same spectrum. So the operators tend to move spectrum from old to new generations.
This is not immediate and there are a lot of differences from one operator to the next: in an area with low demand it's more cost efficient to keep using the old hardware as much as possible for example, until at some point using new more efficient hardware is cheaper than maintaining the old one. Then there are contractual obligations: some 2G is kept around to support industrial M2M applications with contracts covering 10 to 15 years of service for example.
Which is why there's quite a lot of differences from operator to operator. Still, the trend is always there. What changes is the transition speed.
I imagine there will be a fair bit of industrial equipment and smart meters needing replaced as a result!
As others have pointed out, 2G GSM uses less spectrum (not for the same usage, but as a base) with 0.2MHz channels rather than 5MHz channels. There's been some stuff talking about T-Mobile running GSM inside the guard bands of their LTE network (dead space between channels). Still, it seems like they're also going to be shutting it down.
T-Mobile's GSM network is pretty limited with maybe a third of the coverage of their LTE and 5G networks. It doesn't use any low-band spectrum so it performs pretty poorly. They also haven't been expanding it over the past 7 years or so, preferring to put their effort into LTE and now 5G.
They did keep GSM alive a lot longer than AT&T (6 years longer), but it seems like the reasons for keeping it alive are dwindling compared to the cost.
I think that greatly depends how much they can continue to milk those 2G / 3G equipment. And the pandemic has speed things up a little bit.