Cheesy ESMR/LTE base station tracker

[lordsutch]

I ended up backing out the map API change, so this release still uses Leaflet, but I did get the material design look implemented correctly. I also added the two columns of data that digiblur requested; hopefully I've gotten the band guessing logic right, but particularly for non-Sprint networks it could be wrong.

 

Linky to this release: https://www.dropbox.com/s/isejck79hskqqcl/SigDet2.apk?dl=0

[mikejeep]

I tried using the version at that link, and it doesn't work on my N5.. previous versions worked fine. The notification says "Loading..." with an exclamation point icon, and the app itself shows no information. I tried uninstalling/reinstalling with no changes:

 

EDIT: Nevermind.. after about 2 minutes, it suddenly started working. Now it opens on startup immediately. Android gremlins!

 

-Mike

[lordsutch]

I tried using the version at that link, and it doesn't work on my N5.. previous versions worked fine. The notification says "Loading..." with an exclamation point icon, and the app itself shows no information. I tried uninstalling/reinstalling with no changes:

 

uploadfromtaptalk1433206875477.jpg

 

EDIT: Nevermind.. after about 2 minutes, it suddenly started working. Now it opens on startup immediately. Android gremlins!

 

-Mike

It probably needed a GPS fix to start updating (which is probably a bug).

[dkyeager]

It probably needed a GPS fix to start updating (which is probably a bug).

 

I have seen that before.  Generally move a block or two away or change bands, then it works.

[lordsutch]

If anyone still cares, I've updated the app for Android 6.0 including a permissions prompt. Otherwise no big changes since the last release, I think, except moving the layer settings into preferences so they stick between launches. Same link as before (hopefully)...

 

https://www.dropbox.com/s/isejck79hskqqcl/SigDet2.apk?dl=0

[lordsutch]

More updates in this release; main changes:

• A few more LTE band guesses.
• Should be a bit more efficient when running in the background.
• Action button added so you can now exit directly from the notification.

Get it here, as always: https://www.dropbox.com/s/isejck79hskqqcl/SigDet2.apk?dl=0

[lordsutch]

I've been working on another feature lately: logging and displaying the LTE timing advance values reported by the device if the RIL reports them.

 

On my devices they seem to be in units of 16 Ts (just over 0.5 microseconds), corresponding to around 78 meters from the tower RTT per unit 16 Ts, but I've seen other documentation that says the API is reporting data in whole microseconds, which would be about twice as far (≈149.4 meters). The map will also show a circle that should intersect where the tower is likely to be located. So for now it's very experimental, but it may help the tower hunters a bit as they try to triangulate sites more accurately than using RSRP values.

 

Any reports of how this seems to correspond to the real world on non-Nexus devices would be greatly appreciated.

[dkyeager]

Not sure how that will work with small cells.  We have to find one of the new ones first

[davidtm]

Just loaded this app to help map some new sites in my (Shentel-nTelos-Sprint) area. I can't see how to export the log file; I found the folder for the app, but it's empty.

 

Any advice or help pointing me to what is probably obvious? ????

 

Thanks,

David

 

Sent from my 0PJA2 using Tapatalk

[Flompholph]

They are saved to the internal storage for my s7 the address is ...

 

/sdcard/Android/data/com.lordsutch.android.signaldetector/files/

 

"Sdcard" is just the internal storage of the phone. There is no export it saves them automatically. I normally save the files to dropbox then delete the files on my phone or they get to big for google sheets.

[davidtm]

Thanks. I find the Files folder, but it's empty, even after running the app today. I have the log sites option checked. Maybe it will appear later?

 

Sent from my 0PJA2 using Tapatalk

[lordsutch]

A long-overdue update with some code cleanups. Now needs Android 4.2 or later to simplify a few things (hopefully nobody is left behind).

 

The big update is that it will use the EARFCN data available with Android 7.0 or above on some handsets, rather than using the (possibly broken, certainly untested by me) root method for Qualcomm devices. I've also updated the band detection code to use the EARFCN instead of guessing using the MCC/MNC/GCI if possible.

 

It should also be able to use the full screen real estate on the LG G6 and Samsung S8. 

 

It definitely works on the Nexus 5X (Project Fi) and Pixel XL (Sprint). On other devices... who knows?

 

Get it here: https://www.dropbox.com/s/hucdx5lu36zp45c/Signal-Strength-Detector-debug.apk?dl=0

 

Source: https://github.com/lordsutch/Signal-Strength-Detector

[dkyeager]

Thanks very much.  Works on screen with LG V20 showing earfcn on unrooted phone.  I had heard that neighbors could also show earfcn, which would help a lot in Samsung markets.  Any other cool features mentioned like number of TX? At least a date stamp would also help in the logs.  Thanks!! 

[Flompholph]

S7 edge here

 

If EARFCN = 0 can you hide it and do band lookup by GCI? It takes a line by itself and reporting B1.

[lordsutch]

Thanks very much.  Works on screen with LG V20 showing earfcn on unrooted phone.  I had heard that neighbors could also show earfcn, which would help a lot in Samsung markets.  Any other cool features mentioned like number of TX? At least a date stamp would also help in the logs.  Thanks!! 

 

Excellent! It does log EARFCNs if it gets them from the radio for neighbors, but I haven't seen them show up in the logs on my phones - the cellinfolte.csv log includes neighbors while the ltecells.csv only includes the primary cell (technically the first cell that shows up as isRegistered).

 

I've pushed out a new release that adds a timestamp to each log entry; hopefully that'll help.

 

S7 edge here

 

If EARFCN = 0 can you hide it and do band lookup by GCI? It takes a line by itself and reporting B1.

 

I've added some code to ignore EARFCN 0 too - try it now.

 

Poking around in the Android O preview APIs, it looks like some additional signal information will be available, including Cqi, Rsnr, Rsrq, Rssnr for LTE and the timing advance for GSM.

[dkyeager]

Anything for channels on CDMA? This is quite useful for quickly figuring out where B26 LTE 800 5x5 is no longer possible, plus B25 refarming for wider LTE.

 

Sent from my LG-LS997 using Tapatalk

[Flompholph]

I've added some code to ignore EARFCN 0 too - try it now.

 

It IDs the bands 25 and 26 fine now. Thanks.

[Flompholph]

While you have it on your mind not sure if you want to fix sector 0x11 and mini micros. Sorry if syntaxs is wrong I am on my phone and never really messed with java.

 

if(sector == 0x11){

if(GCI >= 0x7600000 && GCI < 0xBA00000){ // Samsung

return 26;}

else{

return 25;}

}

 

 

//mini macros starts with what looks like B25 but b41 sectors

if(sector >= 0x31 && sector <= 0x43)

return 41;

 

 

I saw a mini macro west of atlanta so you may see them soon popping up.

[lordsutch]

I'll keep an eye out. This far south all we really have is Band 25 (1 carrier) and Band 26 for now.

[lordsutch]

Warning: semi-technical post follows. If you don't want to play with code, this post isn't for you.

 

Here's something I've been working on for a while to post-process the log files and help you find the location of towers. Obviously if you're a sponsor you don't need this functionality for Sprint generally speaking, but it may be helpful with new deployments (mini-macros, Clearwire site upgrades, Airspan and Mobilitie sites, LTE Airaves, etc.) and other carriers where that info isn't available.

 

It relies on your device reporting the LTE Timing Advance value, which (due to the laws of physics) gives you a reasonably accurate distance to the base station location, to a much greater extent than a guess based on the signal strength that is affected by terrain, ground clutter, etc.

 

Here's the code; it requires R and several R packages: the ones listed under the library() calls at the beginning, plus any dependencies.

 

Here's an example of the output map. Compared to ground truth, only one tower location (3A492) is wrong, which is understandable given that there are only 3 data points for it.

 

It's not purely plug and play because there's no general way to determine which cells are co-located across markets, but I figured I'd throw it out there in case anyone else finds it useful as a starting point for their tower hunting. You can ignore co-location and try to estimate locations too (that's what the "drawnmap3" maps are: example), but having the site co-location heuristics set for your market (the code involving "base2" does this) will improve accuracy a lot since you can constrain the location solutions better with data from multiple bands.

 

Anyway if it breaks both pieces are yours. Enjoy!

[dkyeager]

In the LTEcells file can you add mcc and mnc.  That would keep me from splitting the cellinfo file or making bad assumptions.  Thanks.

 

 

P.S.: Any way to get the earfcn for the neighbors?  Neighbor earfcn is there. My bad.

Edited June 22, 2017 by dkyeager

[Flompholph]

I second the adding of the plmn. With Roaming it is a little hard to see what is sprint coverage and what is roaming. I was looking for native service in an area and it made it a little difficult to filter the map. 

[ingenium]

 

 

P.S.: Any way to get the earfcn for the neighbors?

I browsed the source of GitHub and it looks like he retrieves the earfcn value for neighbors. But maybe doesn't store or use it?

 

Sent from my Pixel XL using Tapatalk

[dkyeager]

I browsed the source of GitHub and it looks like he retrieves the earfcn value for neighbors. But maybe doesn't store or use it?

 

Sent from my Pixel XL using Tapatalk

 

 My plan is to then sort through those for new PCIs for things like small cells, else you need to be withing a few blocks to find them.

[dkyeager]

An additional request, store the data time in an excel compatible format, typically decimal.  For example the date and time of this request is approximately:

42897.38

 

This could be in another column if needed.  Currently I pull apart the string and reformulate it.

 

Thanks again!