telecomHall.com | Hunter Blog | Tips Blog | Course Blog | Groups

Cross feeder detect from Drive test data


#1

Dear expert,
Do you have any idea to detect cross feeder from Drive test data 2G/3G/4G network (auto report site list/sector suspected swap/cross sector/feeder), please share the code if possible,
Thanks sir,


#2

Having the cells mapped with the corresponding sector and its bearing, it’s easy to spot a x-sector.

Just plot on the map…
2G - BCCH frequency
3G - PSC
4G - PCI

…and just by looking at it, you can understand if they are crossed or not.

There are tools that will make your life easier, and IMO the best come from crowd-source data.

You can shape the data to determine which cell was serving the customers, for the whole network at any point in time.
Then you’ll need to assign that serving cell to its corresponding sector and site.
Determine a geographical point, taking in to consideration all the users served by that cell (for example, use the center of an imaginary polygon of those users).
For reference, draw a fixed width line for every sector and its planned bearing.
Then draw a line from the middle of the sector line and middle point of the users polygon.
Lastly, create a routine to check if, taking in to account the point of your site, sector’s bearing and the bearing towards the served users, if it’s within the expected bearing or not.

Bear in mind that due to reflection, you might get some odd results. So give more attention to sites where at least 2 sectors were flagged for possible x-sectors issue.

All the lines I’ve talked about are only to help on post investigations, as a visual aid to get a better understand of what might be happening (x-sectors, all sectors rotated clockwise/anti-clockwise/possible reflections/ etc).

This is also a good way to visually identify possible overshooters, but you should move the point from the middle of the imaginary users polygon, to the further user being served by any given site middle point. And/or you might want to tweak it further by adjusting how far away you set it, depending on the users distribution per distance.
(Or just use TA/PRACH delay stats :slight_smile:)