Coverage Mapping Tools

[lilotimz]

For those looking to use the Cellmapper websites, Sprint is country 310 and network 120. Clear is country 311 and network 870.

 

311-490 is also Sprint / Clear. 

 

Sprint uses multi plm-nid broadcast for their prepaid users too and it's something in the 312 range IIRC. 

[mhammett]

https://plus.google.com/+MikeHammett/posts/KTCdsMPNFpb

Is Ivan's assessment correct? I don't have the time to learn more about this to directly help him.

[mhammett]

00,01,02 band 25 

03 split between second carrier Samsung and  first carrier

04,05 second carrier Samsung band 25

09,0A,0B second carrier band 25

0F,10,11 band 26

19,1A,1B band 26

 

 

 

band 41 

GCI 3rd is number from the left is "odd" if(GCI && 0x00100000){}//band 41 

00,01,02 First carrier

03,04,05 Second carrier

31,32,33 First carrier

39,3A,3B Second carrier

I couldn't come up with a GCI number in that format. SignalCheck presents an eight digit hex that translates to a nine digit decimal, which seems to match the length in the CellMapper logs, but you've got an eight digit number where the third is even. I've been noticing in Signal Check logs that Sprint B41 has always been 07F.

 

That said, I also saw a couple Clear 07F GCIs. Could those be 8T8R conversions?

[Flompholph]

I couldn't come up with a GCI number in that format. SignalCheck presents an eight digit hex that translates to a nine digit decimal, which seems to match the length in the CellMapper logs, but you've got an eight digit number where the third is even. I've been noticing in Signal Check logs that Sprint B41 has always been 07F.

 

That said, I also saw a couple Clear 07F GCIs. Could those be 8T8R conversions?

GCI in hex is broken down by market XXX tower XXX sector XX. So MMMTTTSS 07E is chicago market band 25/26 (E=14 even) 07F is band 41(F=15 odd). For some reason they did their market GCI counting by 2 so all of the band 25 is even.

 

There is also a 1450 hex offset between bands for the same site. Meaning 07E00100 and 07F45100 are the same site.

 

07F clear is most likly plmn sticking and most likely an 8t8r. If my memory is correct clear should start with 00A or 00B.

[AgathosAnthropos]

GCI in hex is broken down by market XXX tower XXX sector XX. So MMMTTTSS 07E is chicago market band 25/26 (E=14 even) 07F is band 41(F=15 odd). For some reason they did their market GCI counting by 2 so all of the band 25 is even.

 

There is also a 1450 hex offset between bands for the same site. Meaning 07E00100 and 07F45100 are the same site.

 

07F clear is most likly plmn sticking and most likely an 8t8r. If my memory is correct clear should start with 00A or 00B.

 

Here is my simplistic algorithm for band determination:

if (provider_.equals("Sprint")) {

              ///@todo Will RRPP B12 be native "Sprint" coverage?

                if ((getPlmn() == 311870) ||

                      (getPlmn() == 311490)) { ///@todo add more "Clearwire plmn"

                      return 41; // Clear 41

                  }

                  else if((getGci() & 0x00100000) != 0) { //market code byte has lsb set

                      return 41;

                  }

                  else if((getGci() & 0x8) == 0x8) { //sector id has 4th bit set

                      return 26;                     ///@todo also check 5th bit?

                  }

                  return 25;            

            }

 

One thing I'll add is that the GCI is technically only 28 bits so it's only 7 hex digits and not 8 - so the most significant nibble (i.e. 0 in the 07E or 07F in the examples) will always be zero - http://developer.android.com/reference/android/telephony/CellIdentityLte.html#getCi%28%29

 

MikeJeep or SpenceSouth (or anyone really) can chime in with fixes/changes for this.  This also doesn't account for the "Sticky" PLMN mentioned above.

Edited June 11, 2015 by AgathosAnthropos

[Flompholph]

...

else if((getGci() & 0x8) == 0x8) { //sector id has 4th bit set

return 26; ///@todo also check 5th bit?

}

return 25;

}

...

 

I don't think the band 26 would be that easy but close. 0F(0000 1111) 10(0001 0000) 11(0001 0001) 19,1A,1B. 0F would through it off.

[AgathosAnthropos]

I don't think the band 26 would be that easy but close. 0F(0000 1111) 10(0001 0000) 11(0001 0001) 19,1A,1B. 0F would through it off.

Sorry, I'm confused on what you mean. Is 0F a possible B25 sector? My logic would make it B26. As it would also mark 19, 1A, and 1B as B26. I'm in an Ericcson market so I am not familiar if other markets use a different sector format either.

[Flompholph]

Sorry, I'm confused on what you mean. Is 0F a possible B25 sector? My logic would make it B26. As it would also mark 19, 1A, and 1B as B26. I'm in an Ericcson market so I am not familiar if other markets use a different sector format either.

0F,10,11,19,1A,1B are all band 26. Yours would not work for 10,11. If you change it to check the 5th bit(0x10) 0F would not work. You would be better of doing something like if(0x0E<sector<0x1C)

[JossMan]

311-490 is also Sprint / Clear. 

 

Sprint uses multi plm-nid broadcast for their prepaid users too and it's something in the 312 range IIRC. 

 Sprint prepaid accounts use PLMN 312530

[lordsutch]

The Java code I use in my app for band detection is at the following link; feel free to swipe it. It may or may not handle the prepaid PLMN and I don't have any logic in there for second carriers, but whatevs. Also, the T-Mobile logic is missing because I haven't bothered to add it yet.

 

https://github.com/lordsutch/Signal-Strength-Detector/blob/master/src/com/lordsutch/android/signaldetector/SignalDetectorService.java#L321

[mhammett]

CellMapper has added hex representation to their site IDs. I think they need some deeper code changes to correctly identify the B41 sites as such.

[mhammett]

CellMapper now has a top users list. It is for sure missing some that I did, but oh well.

https://www.cellmapper.net/topUsers

#9 in cells discovered and uploaded data points. :-)

[Flompholph]

Don't you hate it when you lose data. That is still a lot of data you contributed though. I know in SignalCheck Pro I have 10404 data points I collect and I lost a lot of the data that I collected in NYC and Boston.