Worst cell in BSC

[electron]

Hi all,


Suppose we have KPI for busy hour .How practically we can find most worst cells which affecting BSC performance on TCH and SDCCH drop in E******* system?

Is there any macro or tool?

Thanks

[josogun2]

Any info on this from the expert?

I will add rep to any useful reply.

[mashi123]

Hi all
i'm not an expert but i will tell you what i do in my work to get worst cells :

1. first as you said we have to get the statistics of main KPIs for the busy hours . (like call setup success rate , handover success rate , dropped call rate , sdcch drop rate ... etc)

2. for example if i want the worst cells in terms of call setup success rate (CSSR) i will open the document of KPIs in excel and then sort the column that contain the CSSR stat from lowest to highest values

3. of course the lowest values represent the worst cells.

hope this is useful

also you can also make a macro in excel or any programing language that will do this work automatically for you

[telwind]

excel pivot table is good tool to do it.

[s52d]

Hi all,


Suppose we have KPI for busy hour .How practically we can find most worst cells which affecting BSC performance on TCH and SDCCH drop in E******* system?

Is there any macro or tool?

Thanks

Hi!
It all depends on:
- do you understand STS
- do you have access to raw data

What we do: parse all STS files from IOG/APG and generate number of .csv files.
(format is well described in Alex, any secondary school student can do parsers)
Then it is easy for engineers to do SW to look into details: down to last number if needed.
Even openoffice.org is good to make graphs - my tool is fpc and gnuplot.

Now, funny part is: what is worse?
Rural cell with 10 calls and 1 drop has 10% drops.
CIty cell with 10 000 calls and 10 drops has 0.1 % drops.

MBA boss would jump on rural cell: I would do city cell.

For me, problematic cell is: worse as expected for this traffic profile and cell density.
Quite often we find faulty HW by this approach.

So, if you already have them on unix/linux box (OSS can do it), use sort and tail and you have it.
Look at composite KPI, look at raw numbers and NEVER claim 1 drop out of 4 calls is 25% drops!


BR
s52d

[Processor]

Nice one s52d...i also look at numbers not rates cause if you work on the high numbers your KPIs improve.

[striker]

Nice one s52d...i also look at numbers not rates cause if you work on the high numbers your KPIs improve.

Yup I agree with you..but there are some projects has a target not only bsc wise but including the cellwise...:) ...Especially acceptance report for Cell KPI's..no choice but to optimize the cell which has high drop rates even with low calls..hehehe

[mashi123]

Optimizing the cells with high Drop Number affect the Drop Rate on BSC level and has a major impact on the subscribers perceived drops
but optimizing the cells with drop rate specially cells with low traffic will have no impact on BSC level

[striker]

Optimizing the cells with high Drop Number affect the Drop Rate on BSC level and has a major impact on the subscribers perceived drops
but optimizing the cells with drop rate specially cells with low traffic will have no impact on BSC level


yup it's true...but working with the contract of cell BBH kpi's it's a different story already...;)

[aircom_vn]

Hi all,


Suppose we have KPI for busy hour .How practically we can find most worst cells which affecting BSC performance on TCH and SDCCH drop in E******* system?

Is there any macro or tool?

Thanks

Yep,

could you use create an pivot excel marco to filter the cell list with high drop call ?

Ex. any cell which has Call Drop Rate >=10% and Call drop number >=10 calls ...
That is the way we use to concentrate to the worst cells.

BR.,