Thinking the Unthinkable: The End of Circuit Switching

Quite a number of network operators are in the processes of shutting down their 2G or 3G networks or have already done so. A major example is AT&T who has switched off their 2G network back in 2017. In other words, they are now only running on their UMTS and LTE networks. So what will happen once they switch off UMTS as well?

Perhaps the thought might still seem a bit far fetched as many would argue that they still need their 3G network for voice telephony. I would hold against this view, however, that AT&T’s LTE coverage is probably pretty close or even exceeds their UMTS coverage area by now. Also, they’ve been at the forefront of rolling out VoLTE for IP based voice telephony on LTE over the past few years and I would be surprised if they are still selling any non-VoLTE devices today. In addition, the average usage period of a smartphone before it gets replaced is 18 months. Put all of this together and in, let’s say 2 years from now, there will only be a few non-VoLTE capable devices of their subscribers in their network. At this point they could switch-off their UMTS network and in the process they would get rid of a lot of very old legacy core network infrastructure, i.e. the whole circuit switched MSC and IN infrastructure. In addition they could get rid of a lot of CS/VoLTE interworking procedures such as all the different flavors of SRVCC (Single Radio Voice Call Continuity) that are making life difficult.

Yes, circuit switching has been virtualized many years ago and all circuit switched telephony traffic is transferred over IP right up to the base stations by now. But it’s a crutch, it should go when the time has come.


So while all of this sounds nice and well there is one little problem: Roaming. While the industry has made tremendous progress with LTE and VoLTE for ‘home subscribers’, there is close to zero progress when it comes to VoLTE roaming. Except for a few examples such as VoLTE roaming agreements between Japanese and Korean network operators, pretty much all others around the world rely on CS-fallback to GSM or UMTS for voice telephony when their subscribers roam in other LTE networks.

It’s not rocket science to introduce S-8 based VoLTE roaming but it seems not to be a high priority for many network operators so far. After all, a critical mass needs to be reached, i.e. enough other network operators have to support S-8 based VoLTE roaming before both GSM and UMTS can be switched-off in one’s network without loosing a significant amount of roaming revenue. So a lot of network operators driven by their CFO rather than innovation might wonder why they should be one of the first, it just costs money they won’t get back in this quarter or the next. And who cares about a few years down the road?

The way I see it, roaming is going to be THE major stumbling block to switching-off both GSM and UMTS in a network. Unless, of course a network operator is not interested in the roaming revenue, which I think is unlikely.


There actually ARE a few network operators that are not getting a lot of roaming revenue today! Take Verizon for example. Their 2G/3G network is based on CDMA so they probably only have a handful of inbound roaming agreements as only very few devices sold elsewhere around the world support legacy CDMA. So perhaps Verizon or one of the few others that never made the jump to 3GPP GSM/UMTS for 2G/3G will go to LTE&5G-only in the not too distant future!?

4 thoughts on “Thinking the Unthinkable: The End of Circuit Switching”

  1. Hi Martin,
    interesting article, and I fully agree. I recently read that an operator in NZ announced to shut down both, 2G and 3G… Have they considered loss of roaming traffic if they don’t introduce VoLTE Roaming asap? Another stumbling block on VoLTE Roaming would be device interoperability. Parameters and sometimes even worse, stacks would need to be switched straight when crossing the border since VoLTE networks today are quite heterogeneously configured… Still a lot of work for us walking down this route…

  2. Nice thoughts!
    I describe this as almost(!) getting to the end of the ‘Great Eastern’ phase. The ‘Great Eastern’ being that ship that came in the transition from the sailing era to the steam era:

    What concerns me is as you describe that there is ‘still a lot of work for us walking down this route’, the Whatsapp/Facetime/etc. calling solutions don’t have to deal with this at all, and can concentrate on the new ‘steam-powered services’…

  3. One additional point for retaining 2G should also be considered:
    It is not only about smartphones. There are already many IoT devices in the field which cannot be replaced easily.

  4. Hello,
    I wonder why each mobile network operator has its own specific VoLTE configuration? If you would like to use VoLTE you need to buy operator branded phone or phone with software provided for specific market that has operator specific configuration. Why you cannot use any VoLTE enabled phone in any VoLTE enabled network as it is with 2G, 3G or LTE-data mobile devices (assuming common radio bands are supported)?
    Why an specific configuration cannot be provided to the user device automatically via SMS (as it is presently done for MMS or internet access), NAS, DHCP (on SIP signalling bearer) or even written on SIM card?
    So IMO VoLTE roaming standard is far away if currently VoLTE has many standardization gaps.

    PS. Internet based VoIP application are not replacement solution because sometimes voice transmission needs support from lower layer like TTI bounding or SPS that cannot be provided via network neutral Internet access.

Comments are closed.