MPU Overload in RNC with few Nodes B

[ivanbrij]

Dear Experts,

I'm having a problem of CPU overload for MPU 0-4-0 in RNC with only 94 Nodes B. There are 6 pairs of SPUb boards in the RNC and 2 MPUs. DPUs and FG2 distribution between MPUs is balanced, SCTP/M3UA Links distribution between SPUs is balance. Nodes B distribution between SPUs is describe as below:

-----MPU 0-2-0---- Total 37 Nodes B



SPU 0-0 with 14 Nodes B
SPU 0-2 with 12 Nodes B
SPU 0-16 with 11 Nodes B
DPU 8

FG2 18 for Iu/Iur



---MPU 0-4-0---- Total 57 Nodes B
SPU 0-4 with 22 Nodes B
SPU 0-14 with 24 Nodes B

SPU 0-20 with 11 Nodes B

DPU 9

FG2 22 for Iub

37 Nodes B handled by MPU 0-2-0 have high traffic than the other ones. Alse there are a lot of problems of UL Power congestion in many Cell/ Nodes B.

Thanks in advance,

Ivan

[tikozo]

Dear Experts,

I'm having a problem of CPU overload for MPU 0-4-0 in RNC with only 94 Nodes B. There are 6 pairs of SPUb boards in the RNC and 2 MPUs. DPUs and FG2 distribution between MPUs is balanced, SCTP/M3UA Links distribution between SPUs is balance. Nodes B distribution between SPUs is describe as below:

-----MPU 0-2-0---- Total 37 Nodes B



SPU 0-0 with 14 Nodes B
SPU 0-2 with 12 Nodes B
SPU 0-16 with 11 Nodes B
DPU 8

FG2 18 for Iu/Iur



---MPU 0-4-0---- Total 57 Nodes B
SPU 0-4 with 22 Nodes B
SPU 0-14 with 24 Nodes B

SPU 0-20 with 11 Nodes B

DPU 9

FG2 22 for Iub

37 Nodes B handled by MPU 0-2-0 have high traffic than the other ones. Alse there are a lot of problems of UL Power congestion in many Cell/ Nodes B.

Thanks in advance,

Ivan

Ivan,

For CPU balancing (MPU), you need to see RRC Attempt KPI. The signalling caused by smartphones is worst then traffic for these CPUs.
In this case, you need to create another MPU, associate some NODEBs for this new MPU, and some interface boards (you can associate SPU 0-14 for this new MPU).
Each MPU supports 32 SPU Subsistems, it´s not a problem for your case (because you have 3 active SPUb = 24 subsystems for each MPU).

What´s your RNC software version, RAN12?

Another question, are these problems of UL Power congestion in many Cell/Nodebs associated with CPU overload?

Hope I could help.

[ivanbrij]

Hi Dear,

SW version for RNC with problems is RAN12.

I have only 1 subrack. Huawei said than only support 2 MPUs (2 SPUs as RUCP per subrack). Also Huawei said than MPU supports only 24 subsystems.

Checking RRC Reject I found than at peak hour there are around 45000 RRC Reject due to congestion (VS.RRC.FailConnEstab.Cong), but at same time RNC has only 900 RRC fail due to UL Power/CE Congestion and there are not RRC rejects due to flow control. What about the other 44000 RRC Rejects?

Any suggestion?

BR,

JB

[rfsupp]

Dear,

Beside the capacity issues mentioned above (that is correct) we need to consider the number of users in each subsystem:

Total - Each SPU subsystem supports 3000 connected users
1200 CELL_DCH users or CELL_FACH
1800 CELL_PCH users.

Have you checked?