CID data for CDMA

I’ve got a question about the context of the CID field for CDMA records. I was looking at Sprint data that I extracted. Here’s a snapshot of the CDMA data (see below). Here’s my question. Does the CID for CDMA in the OpenCellID database represent a unique BTS or is it a cell sector? For example, when I extract LTE data from the OpenCellID database the CID is a unique sector identifier. It’s clear that for the LTE data one can take the CID and do a shift right by 8 bits to extract the corresponding eNB. For this CDMA data, I really can’t tell if there’s a unique BTS embedded within the CID for if the CID actually represents a unique BTS. Any help/guidance would be greatly appreciated.

Best Regards,

Nick Damenti

Radio, MCC, MNC, Area, CID, Unit, Long, Lat, Range, Samples
CDMA, 311, 870, 313, 20561, -1, -106.461, 31.7812, 1000, 1
CDMA, 310, 120, 240, 20561, -1, -122.197, 47.4491, 1000, 2
CDMA, 310, 120, 5, 20561, 0, -88.2154, 30.5619, 1000, 1
CDMA, 310, 120, 240, 20562, -1, -122.197, 47.4348, 1000, 1
CDMA, 311, 870, 313, 20562, -1, -106.466, 31.7734, 1000, 1
CDMA, 310, 120, 5, 20562, 0, -88.1833, 30.5386, 4665, 5
CDMA, 310, 120, 301, 20562, -1, -85.2134, 32.822, 1000, 1
CDMA, 310, 120, 5, 20563, 0, -88.2425, 30.5432, 3859, 13
CDMA, 311, 870, 313, 20563, -1, -106.466, 31.7734, 1000, 1
CDMA, 310, 120, 301, 20563, -1, -85.3499, 32.6441, 1000, 1
CDMA, 310, 120, 309, 20569, -1, -91.4567, 38.9003, 1000, 1
CDMA, 311, 870, 307, 20569, -1, -81.0031, 29.1638, 1000, 2

Hey Nick,

We always report the approximate BTS position - not cell sector. Sometimes, the last digit represents the sector; but it is not universal rule.

Sagar-

Thanks for your reply. What I’m trying to do is consolidate the data to identify unique ‘towers’, meaning unique eNBs or unique BTSs. I’ve been able to solve the puzzle quite well for LTE, UMTS and GSM, but I’m still a bit puzzled as to how to consolidate CDMA sector information. Any suggestions would be appreciated.

Thanks, Nick

1 Like

The best approach would be to cluster cells that are close to each other and consider it as being on the same cell tower

Hi Sagar,

I am trying to do similar and not sure how ndamenti did for GSM/LTE. I would like to have a unique count for the number of cell sites per mcc-mnc (country-operator). Can I just use the count of the CID in the database? and is there a way to “de-duplicate” for any overlap between GSM/UMTS/LTE?

thank you!

You can combine ‘lac-cid’ into one string and then do a count.

There are no duplicate cells in the entire database

thank you very much!

Happy to help :slight_smile:

Hi Sagar - do you have a recommended way to “cluster” the cells together? Is there a recommended algorithm or certain parameters in the data we can use over others? I am also trying to get to ‘unique cell towers’. Curious how others have solved this. Thanks.

1 Like

I’m bumping this thread to see if anyone has solved this problem yet. :upside_down_face:

@EdS @Cindy_Forcia, there is a pretty good example of doing this in R posted to GIS Stack Exchange using hierarchical clustering: Clustering spatial data in R

I checked the distribution of the cell range values in my study area, and most are 1000m, so I applied the clustering algorithm from that example and set d = 1000. I retained the coordinates for the cluster centroids and called them “cell towers” - out of the 500 rows that I sampled, I wound up with 450 clusters i.e. unique “cell towers.”