Missing 4G only mode toggle in the SIM settings

oci3o

itsjpb It used to be a separate option, so you'd have a list of modes such as 4G and 5G only mode, 4G only mode etc, but it seems to be missing lately.

wizoatk

oci3o

Something changed and now some carriers can hide this option.

You should be able to change the preferred network type with the *#*#4636#*#* dialer code and then selecting Phone information v2 (or Phone information). Or you could try an app like ForceLTE.

Ladida

oci3o

Yes exactly

oci3o

wizoatk Just forced it using the way you suggested, thanks :)
Not sure why its not an option in the settings however.

GrapheneOS

wizoatk This won't work correctly. It doesn't save it in a way that's as persistent as the normal setting. Using an app isn't going to work properly and should not be done.

wizoatk

oci3o

The simple answer is that you and your chosen carrier have a difference of opinion about what options the end user should have access to. The SIM options UI is dynamic. For example on my primary phone, two out of three SIMs still have the preferred network option in the Settings UI. The third SIM has that option hidden. That same behavior exists on Stock OS. It's not obvious where GOS will end up in either the short or long term with respect to specific carrier overrides.

GrapheneOS

oci3o GrapheneOS is meant to ignore carriers disabling support for changing the cellular network mode but the feature was recently lost due to changes to how we implement device support. It has been added back for the next release. The change you made in the debug menu isn't going to persist properly and will get reset.

r134a

GrapheneOS Might i borrow some of your knowledge that is somewhat relates to this topic?

Yesterday i've came accross an app (Privacy cell, which indicated within the app that a device connected with 4g isn't secure from stingray IMSI mitm attacks, and while connected with 5g it claims the device is 'secure' against stingray IMSI attacks.

I've already found part of the answer here, that neither 4g and 5g aren't 'secure' against IMSI catchers, but the paper is dated 2 years back:
https://arxiv.org/abs/2405.00793.

I'm mostly in airplane mode with WiFi-calling routed through VPN.
For the times i need to connect to cellular towers, unless 5g is/becomes significantly more secure than 4g, i'd prefer staying on 4g only as i believe in general the triangulation 'range' is significantly 'less accurate' than 5g, depending on environment etc (urban, city, ...)

What is the case as of date? And what would be your recommendation for me or people like me with a threat model u can derive from my writing?

Thank u

oci3o

GrapheneOS I saw the recent change logs for this, thanks for the fix!

de0u

r134a For the times i need to connect to cellular towers, unless 5g is/becomes significantly more secure than 4g, i'd prefer staying on 4g only as i believe in general the triangulation 'range' is significantly 'less accurate' than 5g, depending on environment etc (urban, city, ...)

Is there detailed information about the degree of the difference?

r134a

de0u At present 5G does not localize to 8-10 meters,

After reading it a bit more in detail, they indeed claim that it is in principle possible to localize a static user with an accuracy of approximately 8–10 m. U mention at present 5g doesn't localize to 8-10 m, did u mean at publication of paper (2023), or do u know if this is still the case at present aswell?

de0u I think the techniques involve active participation (computation) on the device.

I think you are correct, the test was performed with a static receiver and then some computations were made as more of a POC on what could be possible.

de0u If so, then involuntary precise localization by carriers is not part of that picture.

I agree, in my original question i mentioned triangulation specifically:

r134a i believe in general the triangulation 'range' is significantly 'less accurate' than 5g, depending on environment etc

After lots of reading i agree that in terms of pure triangulation accuracy, as of date there probably isn't a significant difference compared to 4g.

To return to my initial question, do u perhaps know if 5g still is considered bleeding edge?
With this last piece of information i could form an answer.

A 'fun' fact about emergency calling with assisted GPS i came accross while reading lots about this: (https://ceur-ws.org/Vol-3980/paper5.pdf)

The adoption of 5G significantly
improved A-GPS performance, reducing the average assistance data acquisition time to 1.41 s, compared to 3.50
s in 4G, therefore a reduction of more than 50%. Location accuracy also improved, with the horizontal error
reduced to 1.05 m in 5G, compared to 1.36 m in 4G.

r134a

de0u Very good question, admittedly this is an assumption i made based on reading about this separated over several years. It was quite a search to find a single source that could motivate that claim.

I believe these 2 papers could motivate that claim:

https://pmc.ncbi.nlm.nih.gov/articles/PMC9268850/
https://arxiv.org/html/2209.01183v3

Disclaimer, i'm not a RF engineer, so my assumption could be based on interpreting data wrongly because lack of some specific terminology etc.
I believe these papers could motivate the claim i made, however someone more knowledgeable about this specific topic reading these papers could perhaps prove otherwise. In that case i guess i ride Cunningham’s law.

de0u

r134a I believe these 2 papers could motivate that claim:

https://pmc.ncbi.nlm.nih.gov/articles/PMC9268850/
https://arxiv.org/html/2209.01183v3

I glanced at the abstracts. To my eyes those both appear speculative, discussing things that carriers might do someday (one of them mentions 6G).

Even though there may be solid theoretical reasons why 5G localization could be much more precise and may be in the future, personally I would not conclude that there is a "significant" difference in current networks (my hunch is that the difference at present is very close to zero).

r134a

de0u i still believe that 5G networks have higher base station density and more accurate timing information, thus yielding higher location precision, in a nutshell.

When i have some more time i'll try to research it some deeper, perhaps papers that performed tests on 4g specifically and 5g specifically, so they can be compared.

If i find something, i'll post it here.

r134a

Ladida Doesn't 5G have like "beamforming technology " like some subprotocol in WiFi 6e+7 which follows you around as you move around...?

This means it is far more accurate than 3g/4g which is just antenna/cell triangulating

That's roughly what i went looking for in scientific papers to perhaps back up that assumption, but didn't came accross one that proves that is the case as of date, so in terms of pure triangulation, i concluded that currently i have to agree with:

de0u Even though there may be solid theoretical reasons why 5G localization could be much more precise and may be in the future, personally I would not conclude that there is a "significant" difference in current networks (my hunch is that the difference at present is very close to zero).

Perhaps u are aware of a source that could back up that as of date, triangulation is far more accurate on 5g than it's predecessors?

de0u

r134a I still believe that 5G networks have higher base station density and more accurate timing information, thus yielding higher location precision, in a nutshell.

As long as we're discussing beliefs, my personal belief is that a lot of 5G carriers emanate from a pre-existing 4G cell site, with part of the site's frequencies refarmed from the one protocol to the other. Over time, for sure, the plan is to add more/denser cell sites using the higher frequency bands, but that takes more time and money than refarming some of a site's spectrum to provide some users faster download speeds some of the time.

r134a When i have some more time i'll try to research it some deeper, perhaps papers that performed tests on 4g specifically and 5g specifically, so they can be compared.

If i find something, i'll post it here.

Great!

r134a

de0u I've found this paper:
https://pmc.ncbi.nlm.nih.gov/articles/PMC10504288/ which performed a test on a deployed 5g Vodafone network in Milan, Italy, and concluded that it is possible to localize with an accuracy of approx 8-10m.

I didn't read it fully though, just skimmed through it fast. On first glance the paper doesn't really compare it to 4g standards, so it is not clear from that specific paper if there is a significant improvement.
if i come across a test on the 4g network that concludes it's possible to localize down to say 50m at best, would that suffice to conclude its a 'significant' difference?

Probably9857

r134a if i come across a test on the 4g network that concludes it's possible to localize down to say 50m at best, would that suffice to conclude its a 'significant' difference?

If so, would that conclusion also apply to the actual locations and networks where you use your device?

de0u

r134a I've found this paper:
https://pmc.ncbi.nlm.nih.gov/articles/PMC10504288/ which performed a test on a deployed 5g Vodafone network in Milan, Italy, and concluded that it is possible to localize with an accuracy of approx 8-10m.

What I see there (admittedly based on a fast read):

At present 5G does not localize to 8-10 meters,
The researchers propose some techniques that someday might localize to that precision.
I think the techniques involve active participation (computation) on the device. If so, then involuntary precise localization by carriers is not part of that picture.

r134a

Probably9857 If so, would it be reasonable to expect from me to perform these tests on all locations and networks i use my device, periodically, to be able to answer that question?

At the time, perhaps or perhaps not, depending on location. As time goes on, gradually, probably so.

Ladida

r134a

Doesn't 5G have like "beamforming technology " like some subprotocol in WiFi 6e+7 which follows you around as you move around...?

This means it is far more accurate than 3g/4g which is just antenna/cell triangulating

With 5g just 1 cell can track your angle and distance instead of 3 like LTE does

de0u

r134a U mention at present 5g doesn't localize to 8-10 m, did u mean at publication of paper (2023), or do u know if this is still the case at present aswell?

That is a great question. I was reacting to that paper, and I am absolutely not an expert on 5G technology (and also absolutely not an expert on the state of deployment). Long ago I had a perhaps-fair non-engineer understanding of the original Qualcomm CDMA system, but that is many years and two whole generations ago.

That said, the way I understand what the authors of that paper wrote in 2023 suggests to me that the techniques they discussed are not in 2025 handsets.

r134a To return to my initial question, do u perhaps know if 5g still is considered bleeding edge?

In terms of deployment? My inexpert sense is that as of the end of 2025:

Some places are still turning 2G and 3G off,
LTE is very widely available,
At many cell sites, a reasonable fraction of LTE spectrum has been reallocated to 5G,
The prevalence of 5G in the newer high-frequency bands varies a lot by country and carrier, with more deployment in Asia (due in part to high population densities) and less in the rural U.S. and Canada (lots of space, not a lot of people),
People are negotiating about what "6G" will include, and there are some deployments of 5G plus some features that will probably be in 6G

de0u

r134a The adoption of 5G significantly improved A-GPS performance, reducing the average assistance data acquisition time to 1.41 s, compared to 3.50 s in 4G, therefore a reduction of more than 50%. Location accuracy also improved, with the horizontal error reduced to 1.05 m in 5G, compared to 1.36 m in 4G.

I'm not sure based on the quote what those error figures are measuring (angle might be good while range might not be), but vertical error is inherently much worse for cellular systems because the cell sites have less "vertical diversity" (especially vertical angular diversity) compared to horizontal. This is similar to GPS: regardless of where the satellites are in an around-the-horizon sense with respect to an observer, they are much more similar in terms of how high up they are.

I'm also not surprised about acquisition time. I think some of the 5G changes cut latency for small packets, though I don't recall details.

r134a

de0u In terms of deployment?

More in the code context, as per the GrapheneOS faq:

The intention of the LTE-only feature is only hardening against remote exploitation by disabling an enormous amount of both legacy code (2G, 3G) and bleeding edge code (5G).

I wondered if it's still considered bleeding edge code, but i suppose it still is.

Thanks for the productive discussion, i'll keep prefering LTE-only for now in the edge use cases i need cellular connection, and won't have any 'fears' regarding triangulation changing to 5g only if the recommendation changes in the near future.