Probability Density Function (PDF) counter to discrete value

[dannthemann]

Hi colleagues,

I am working with counters like

pmRadioRecInterferencePwr

which is expresed in terms of



The measured Noise and Interference Power on PUSCH, according to 36.214

PDF ranges:
[0]: N+I <= -121
[1]: -121 < N+I <= -120
[2]: -120 < N+I <= -119
[3]: -119 < N+I <= -118
[4]: -118 < N+I <= -117
[5]: -117 < N+I <= -116
[6]: -116 < N+I <= -115
[7]: -115 < N+I <= -114
[8]: -114< N+I <= -113
[9]: -113 < N+I <= -112
[10]: -112 < N+I <= -108
[11]: -108 < N+I <= -104
[12]: -104 < N+I <= -100
[13]: -100 < N+I <= -96
[14]: -96 < N+I <= -92
[15]: -92 < N+I



But, what I am trying is to convert that Probability Density Function (PDF) into a single value.

Anybody knows?

Thanks

[electron]

Hi colleagues,

I am working with counters like

pmRadioRecInterferencePwr

which is expresed in terms of



The measured Noise and Interference Power on PUSCH, according to 36.214

PDF ranges:
[0]: N+I <= -121
[1]: -121 < N+I <= -120
[2]: -120 < N+I <= -119
[3]: -119 < N+I <= -118
[4]: -118 < N+I <= -117
[5]: -117 < N+I <= -116
[6]: -116 < N+I <= -115
[7]: -115 < N+I <= -114
[8]: -114< N+I <= -113
[9]: -113 < N+I <= -112
[10]: -112 < N+I <= -108
[11]: -108 < N+I <= -104
[12]: -104 < N+I <= -100
[13]: -100 < N+I <= -96
[14]: -96 < N+I <= -92
[15]: -92 < N+I



But, what I am trying is to convert that Probability Density Function (PDF) into a single value.

Anybody knows?

Thanks

Hi,

Use weighted average calculation .

B
R

[dannthemann]

OK, thanks. I did the weighting and it worked.

BRs

[felixj.alvarez]

In my experience, it is not enough to take the Probability Density Function (PDF) counter to discrete values. The best way to interpret these counters is to create three-dimensional or surface graphs at different time scales. Below, I show two graphs of a counter very similar to the one you are talking about, but of 3G:

42066 42067

[dekili]

Hello,

for each range you have number of events being measured. You will have 16 values representing ranges 0 to 15. You need to multiply that number with average value of the range (e.g. -120.5, -119.5 etc), add all and divide by total number of events (sum of all 16 values). Nevertheless, it is rough estimate as it is average, not exact values would be used (e.g. -120.5 is average and it might be -120 or -121 in the real life) and [0] and [16] are not -121 or -92. They might be much higher or lower from the thresholds given in 36.214. As felixj.alvarez mentioned, you can plot distribution and decide what is acceptable to you regarding the performance. You can say that all that is above e.g. -105 is OK for you, you read counter for 15 minutes and you can assume that is 90% is above -105 that you are 90% of time e.g. 0.9*15 minutes OK and f0r 90 seconds you have outage.

dekili