How accurate are Base Station coordinates?

[s52d]

Hi!

I just recalled never ending story: Is my database correct?
Do I know where my BSes are?
How about antenna type and direction?
We keep finding errors: somebody mistyped etc.

Some operators spend huge amount of time/work to have this data accurate.
Others do optimization based on stats/traces, not on radio predictions.

One nice trick: calculate average handover distance and direction.
Just take handover stats, get distance between sites, multiply by handover number,
and devide result witj total handover number.

It is easy to see wrong coordinates by little sorting: cells in the city center can not
have average over 1km etc. Simmilar with antenna directions.

Not really accurate, but it quickly points to suspicious data, which is then checked a bit
more seriously.

Some other nice tricks, beside going to sites and measure?

BR
s52d

[rfworld]

How to calculate avg HO distance and direction in E///

[s52d]

How to calculate avg HO distance and direction in E///

It is not vendor specific ;-)
If you have statistics like:
CellA-CellB 100 100 0 (relation, HO started, ok, failed)
Take coordinates in Gauss-Krueger coordinates (the one with kilometers),
and solve triangle.
distanceAB^2=(xA-xB)^2 + (yA-yB)^2 etc...

BR

[s52d]

Hi!

Interesting results - looks like some networks are more robust as others.

In optimization process, we do not depend heavily on coordinates, antena types, tilts and directions.

Typical work order is: put this antenna 3 degrees lower, not change tilt to 5 degrees.
So, make changes relative to existing situation, not absolute.

Best indication on quality of data and models is to compare handover statistics
from the system with planning tools. Funny.

So, what we do:
Neighbors are based on mobile measurement reports:
With E///:
- 2g/2g is based on BA list recording
- 2g/3g and 3g/2g also (scanning all possible candidates from GSM)
- 3g/3g intrafrequency is GPEH
- 3g/3g interfrequency is using BA list recording from BSS and GPEH
we define all candidates and monitor stats and remove what is not needed.

For BSS, we build inter cell dependancy matrix from BAR and handover stats.
Thus, we have good picture how cells interfere and where users might feel it, and
use this data for AFP.

On the end, if coordinates etc are not accurate, if there is some new building not in our 3D model, if people decide to make calls where we do not expect them: we just adapt.

Of course, before we have any stats for new sites, we use planning tools based
on radio coverage predictions.

BR
s52d