You are on page 1of 5

文 档 密 级 :

文档名称 Internal only

Author Ahsan Raza Staff Number 00725065


Wireless Performance &
Department NTS Product Family:
RNP & RNO
Update Time 2011-06-29 Product Vension: V200R012C00SPC300
Approver
Title: RRC Rejection due to Radio Link Failure because of high WBBP CPU Load
Phenomenon Some Sites started having RRC Rejection due to Radio Link Failure, however,
Description: when analyzed there was no congestion on any cell of these sites. And all
(power,code, Channel Element) utilization was also low. When the 2nd WBBP
board was added to these sites, RRC Rejection reduced to zero. So the main
question arised that why even with low CE utilization we are having RRC
Rejections and how adding another WBBP Board removed the RRC Rejection.

Alarm None
Information:
Cause None
Analysis:
Handling The first thing to understand is that Each WBBP card has 1 CPU and there is no
Process: counter to see the CPU load. However, the CPU load can be approximately
calculated using following procedure.

With BTS3900 Equipment 1 WBBP CPU can support 45 CNBAP messages per
second.

Version V200R012C00SPC110

WMPT+1*WBBP 45 CNBAP /s (WBBP capacity)

3900 WMPT+2*WBBP 100 CNBAP /s

WMPT+3*WBBP 100 CNBAP /s

WMPT+4*WBBP 100 CNBAP /s

CNBAP messages are basically contributed by below messages:


 RL SETUP Request
 RL ADD Request,
 RL Reconfigure Prepare and Commit,
 NBAP measurement report

The actual live traffic CNBAP is as below:


Since we cannot calculate the real time traffic per each seconds, we are using
mean CNBAP to estimate the peak.
Normally peak traffic/average traffic ≈ 2, so the max CNBAP=2* mean
CNBAP.

2014-2-12 华为机密,未经许可不得扩散 第 1 页, 共 5 页
文 档 密 级 :
文档名称 Internal only

The formula for CNBAP messages is

Now, if the CPU load reaches to the value of 90-100%, NodeB starts to reject
the RRC due to RL failure. When further analyzed using CHR, the cause of
failure comes as NodeB resources unavailable. However, when the CE
utilization is checked is around below 80%. The cause of RRC rejections arte
basically the WBBP CPU is maximum utilized and has no more capacity to
process any more rrc requests. Below is one Such Case.

DXB2315 is having high RRC rejections specifically on weekends.

When analyzed in CHR the cause of RRC Failure was NodeB Resources
unavailable as shown in below snapshot.

2014-2-12 华为机密,未经许可不得扩散 第 2 页, 共 5 页
文 档 密 级 :
文档名称 Internal only

When CE utilization was calculated, CE utilization was observed to be around


maximum upto 60% on the weekend.

Now When WBBP CPU load was calculated using CNBAP message from
below formula. The CPU load was reaching to 90%.

2014-2-12 华为机密,未经许可不得扩散 第 3 页, 共 5 页
文 档 密 级 :
文档名称 Internal only

Also, when the SHO Factor was calculated for this site. It was found that
some of the cells are reaching upto 100% SHO Factor. This is very high causing
too many CNBAP messages and causing high WBBP CPU utilization.

Also, when compared to other cells with high SHO Attempts. It was found that
traffic of this site is not high as compared to the sites with same SHO attempts.

Hence, when the 2nd WBBP board was added the capacity of handling CNBAP
message increased from 45 messages per second to 100 messages per second
and RRC rejection reduced to zero.

Suggestions This issue can be solved using following methods:


and summary: 1. By addition of another WBBP Board.
2. By hard and soft parameter optimization to reduce the SHO Factor.

Attachments:
Related None
document
link:
2014-2-12 华为机密,未经许可不得扩散 第 4 页, 共 5 页
文 档 密 级 :
文档名称 Internal only

2014-2-12 华为机密,未经许可不得扩散 第 5 页, 共 5 页

You might also like