HSDPA_AVERAGE_THROUGHPUT_PER_USER formula ?

[fahmi]

Hi,
please I need formula for this KPI HSDPA_AVERAGE_THROUGHPUT_PER_USER .

vendor huawei.

many thkss.

[romagdinio]

VS.HSDPA.MeanChThroughput:

Definition:

When data is transmitted in an HSDPA cell, the RNC measures the total data transfer time of the UE and the total number of bits in the received data in the cell. At the end of the measurement period, the RNC divides the total number of bits by the total data transfer time to obtain the mean downlink throughput of MAC-d flow in the cell. The RLC header overhead and the retransfer data are excluded.

[fahmi]

VS.HSDPA.MeanChThroughput:

Definition:

When data is transmitted in an HSDPA cell, the RNC measures the total data transfer time of the UE and the total number of bits in the received data in the cell. At the end of the measurement period, the RNC divides the total number of bits by the total data transfer time to obtain the mean downlink throughput of MAC-d flow in the cell. The RLC header overhead and the retransfer data are excluded.

Thank you,
But I tried this one and I have gotten some strange values like 214555 kbit/s .
I think something is wrong !!

[zeezzoo]

try this one :

HSDPA RLC Throughput =

VS.HSDPA.MeanChThroughput.TotalBytes*8/{SP*60}/1024

Unit: kbps
The unit of SP(Statistic Period): Minute


description : This KPI can be used to evaluate the DL RLC average throughput of HSDPA Service in one cell, it excludes the RLC retransmit data and RLC header.

[fahmi]

try this one :

HSDPA RLC Throughput =

VS.HSDPA.MeanChThroughput.TotalBytes*8/{SP*60}/1024

Unit: kbps
The unit of SP(Statistic Period): Minute


description : This KPI can be used to evaluate the DL RLC average throughput of HSDPA Service in one cell, it excludes the RLC retransmit data and RLC header.

the problem here is with SP. When working with parameter SP, you considering both :transmission time and no transmission time.

if we consider an SP of 24 hours, so this will cause an important error.
so what do you think ?

[zeezzoo]

the problem here is with SP. When working with parameter SP, you considering both :transmission time and no transmission time.

if we consider an SP of 24 hours, so this will cause an important error.
so what do you think ?

no dear this is a suitable kpi to get the throughput value . for lower SP ==> you can get more accurate throughput value in terms of realtime throughput .
as an example you cans use 15min in M2000 for this throughput kpi .

you cannot get the realtime throughput by counters . i think there's live monitoring in LMT that show you the realtime throughput . you can also check realtime throughput by doing a drive test for certain areas .

[jan74]

HSDPA Throughput per user can get fairly complicated and should not be really used as an accurate reflection of user experience. Personally I use it to show me a trend. The NSN formula is for Throughput per User is based on the total cell throughput divided by the average number of HSDPA subscribers associated with actual data transmission period duration. Here it is....

(RECEIVED_HS_MACD_BITS + (MC_HSDPA_ORIG_DATA_PRI + MC_HSDPA_ORIG_DATA_SEC)*8 - DISCARDED_HS_MACD_BITS)* 500)
/
(
HSDPA_BUFF_WITH_DATA_PER_TTI
*
((HSDPA_USERS_0_1_IN_CELLS+HSDPA_USERS_1_0_IN_CELLS)
+2*(HSDPA_USERS_0_2_IN_CELLS+HSDPA_USERS_1_1_IN_CELLS+HSDPA_USERS_2_0_IN_CELLS)
+3*(HSDPA_USERS_0_3_IN_CELLS+HSDPA_USERS_1_2_IN_CELLS+HSDPA_USERS_2_1_IN_CELLS+HSDPA_USERS_3_0_IN_CELLS)
+4*HSDPA_USERS_0_4_IN_CELLS+HSDPA_USERS_1_3_IN_CELLS+HSDPA_USERS_2_2_IN_CELLS+HSDPA_USERS_3_1_IN_CELLS+HSDPA_USERS_4_0_IN_CELLS)
+5*(HSDPA_USERS_0_5_IN_CELLS+HSDPA_USERS_1_4_IN_CELLS+HSDPA_USERS_2_3_IN_CELLS+HSDPA_USERS_3_2_IN_CELLS+HSDPA_USERS_4_1_IN_CELLS)
+6*(HSDPA_USERS_0_6_IN_CELLS+HSDPA_USERS_1_5_IN_CELLS+HSDPA_USERS_2_4_IN_CELLS+HSDPA_USERS_3_3_IN_CELLS+HSDPA_USERS_4_2_IN_CELLS)
+7*(HSDPA_USERS_1_6_IN_CELLS+HSDPA_USERS_2_5_IN_CELLS+HSDPA_USERS_3_4_IN_CELLS+HSDPA_USERS_0_7_IN_CELLS+HSDPA_USERS_4_3_IN_CELLS)
+8*(HSDPA_USERS_2_6_IN_CELLS+HSDPA_USERS_3_5_IN_CELLS+HSDPA_USERS_0_8_IN_CELLS+HSDPA_USERS_1_7_IN_CELLS+HSDPA_USERS_4_4_IN_CELLS)
+9*(HSDPA_USERS_3_6_IN_CELLS+HSDPA_USERS_1_8_IN_CELLS+HSDPA_USERS_2_7_IN_CELLS+HSDPA_USERS_4_5_IN_CELLS))
/
(HSDPA_USERS_0_1_IN_CELLS+HSDPA_USERS_1_0_IN_CELLS
+HSDPA_USERS_0_2_IN_CELLS+HSDPA_USERS_1_1_IN_CELLS+HSDPA_USERS_2_0_IN_CELLS
+HSDPA_USERS_0_3_IN_CELLS+HSDPA_USERS_1_2_IN_CELLS+HSDPA_USERS_2_1_IN_CELLS+HSDPA_USERS_3_0_IN_CELLS
+HSDPA_USERS_0_4_IN_CELLS+HSDPA_USERS_1_3_IN_CELLS+HSDPA_USERS_2_2_IN_CELLS+HSDPA_USERS_3_1_IN_CELLS+HSDPA_USERS_4_0_IN_CELLS
+HSDPA_USERS_0_5_IN_CELLS+HSDPA_USERS_1_4_IN_CELLS+HSDPA_USERS_2_3_IN_CELLS+HSDPA_USERS_3_2_IN_CELLS+HSDPA_USERS_4_1_IN_CELLS
+HSDPA_USERS_0_6_IN_CELLS+HSDPA_USERS_1_5_IN_CELLS+HSDPA_USERS_2_4_IN_CELLS+HSDPA_USERS_3_3_IN_CELLS+HSDPA_USERS_4_2_IN_CELLS
+HSDPA_USERS_1_6_IN_CELLS+HSDPA_USERS_2_5_IN_CELLS+HSDPA_USERS_3_4_IN_CELLS+HSDPA_USERS_0_7_IN_CELLS+HSDPA_USERS_4_3_IN_CELLS
+HSDPA_USERS_2_6_IN_CELLS+HSDPA_USERS_3_5_IN_CELLS+HSDPA_USERS_0_8_IN_CELLS+HSDPA_USERS_1_7_IN_CELLS+HSDPA_USERS_4_4_IN_CELLS
+HSDPA_USERS_3_6_IN_CELLS+HSDPA_USERS_1_8_IN_CELLS+HSDPA_USERS_2_7_IN_CELLS+HSDPA_USERS_4_5_IN_CELLS)
)

[dk2000]

I think,

In Huawei system, throughput is calculated like that:
VS.HSDPA.MeanChThroughput = A/B

With, A = Total HSDPA data is transfered per time. ( for example: 5 minutes).
B = TTI*U1 + TTI*U2 + ….. + TTI*Un

U1, U2,...,Un: user is served in each TTI.


But, this counter can't show the felling of customer.

Someone has anther way, please share!

[jan74]

I think,

In Huawei system, throughput is calculated like that:
VS.HSDPA.MeanChThroughput = A/B

With, A = Total HSDPA data is transfered per time. ( for example: 5 minutes).
B = TTI*U1 + TTI*U2 + ….. + TTI*Un

U1, U2,...,Un: user is served in each TTI.


But, this counter can't show the felling of customer.

Someone has anther way, please share!

I have looked in detail at this and came to the conclusion that regular stats are not an adequate way to see customer experience. Every customer is different and their requests are different depending on requested service etc. I have always looked for a way of seeing "What did the customer request and what did the customer get". Short of tracing, I have not yet found a suitable counter/KPI.