You are on page 1of 7

RNC PS call drop rate increase after migration to Huawei SGSN

RNC PS call drop rate increase after


migration to Huawei SGSN

2016-6-17

RNC PS call drop rate increase after migration to Huawei SGSN

1 Brief Description of the Issue


Issue occurred in H project when Huawei SGSN swap out NSN SGSN. Two
RNC(Huawei) were migrated to Huawei SGSN and got same problem on both. DT is
implemented for both two RNC.
After RNC migration (from NSN SGSN to Huawei SGSN), PS call drop rate increase
from 1% to 5%.

Figure 1 PS call drop rate increase after PS migration in BDG RNC


After RNC migration (from NSN SGSN to Huawei SGSN), PS call drop rate increase
from 1% to 8%.

Figure 2 PS call drop rate increase after PS migration in PK RNC


2016-6-17

RNC PS call drop rate increase after migration to Huawei SGSN

2 Alarm inforation
No additional alarm.

3 Root Cause Analysis


3.1

VS.RAB.Loss.PS.GTPULoss description
VS.RAB.Loss.PS.GTPULoss detailed description is listed below:
The measurement is triggered at point A as shown in the figure below. In the case of
GTPU failure(RNC receive ERROR_IND message), RNC sends to the CN a RAB
RELEASE REQUEST message. If RNC receives a IU RELEASE COMMAND
message(with the cause other than "User Inactivity","Normal Release","Successful
Relocation","Network Optimization" or "UE Generated Signalling Connection
Release", reference to 3GPP TS 25.413), RNC measures the items according to
different domains and service attributes in the best cell that the UE camps on.

Figure 3 VS.RAB.Loss.PS.GTPULoss detailed description

2016-6-17

RNC PS call drop rate increase after migration to Huawei SGSN

3.2

ERROR_IND message description in 3GPP Specs

3.2.1

ERROR_IND message is described in 3GPP 23.060:

When the SGSN receives a GTP U PDU from the RNC for which no PDP context exists, the SGSN
shall discard the GTP U PDU and send a GTP error indication to the originating RNC. The RNC
shall locally release the RAB.
When the GGSN receives a GTP U PDU for which no PDP context exists, it shall discard the GTP
U PDU and return an error indication to the originating SGSN. The SGSN shall mark the related
PDP context as invalid and send a Deactivate PDP Context Request message to the MS. The MS
may then reactivate the PDP context.

3.2.2

And the action if RNC receive that kind of message is described as well in
3GPP 23.007:

GTP error indication message shall be handled as follows:


-When the RNC receives GTP error indication from the SGSN, it shall initiate the RAB Release
procedure with the error cause "GTP Resources Unavailable" and shall immediately locally release
the RAB (i.e. without waiting for a response from the SGSN).
-If the RNC receives a GTP error indication from the GGSN (i.e. if Direct Tunnel is established), it
shall initiate the RAB Release procedure with the error cause "GTP Resources Unavailable" and
immediately locally release the RAB (i.e. without waiting for a response from the SGSN).

Considering one tunnel is implemented for BDG RNC, we suspect GGSN send
ERROR_IND message to RNC some times.

4 Handling process
Compare the signaling flow from HW SGSN and NSN SGSN, the difference is
discovered.

4.1

Signaling Handing of GTPULoss when RNC connect to HW SGSN


Huawei traced subscribers in the list and summarized the whole PS call flow which
have such issue. Below is the whole flow chart:

2016-6-17

RNC PS call drop rate increase after migration to Huawei SGSN

Figure 4 BDG RNC PS call drop flow chart


During the PDP activation process, GGSN would sent CCR message to OCS for
authentication without checking balance. And once GGSN receive any data from
RNC, GGSN would send CCR-U message to OCS with corresponding RG for
balance checking. If subscriber dont have enough balance, OCS would send 5003
error code to GGSN. GGSN should send delete PDP request message to SGSN and
delete subscriber PDP context away. SGSN should send deactivate PDP request
message to MS since receive request from GGSN. But there is a chance that MS
keep sending data to network after PDP activation. If one data packet is sent to
GGSN after GGSN delete subs PDP context, GGSN would response with
ERROR_IND message according to 3GPP spec. Once RNC receive ERROR_IND
message and receive Iu release CMP message with error code no remaining RABID,
VS.RAB.Loss.PS.GTPULoss item would plus one. And since Iu connection is
released, Deactivation PDP request message cannot send to MS successfully.
According to above description, we can conclude that the pre-condition for
GTPULoss error are subs without enough balance + Direct Tunnel network + Online
Change enable and would not affect any normal service.
2016-6-17

RNC PS call drop rate increase after migration to Huawei SGSN

4.2

Signaling Handing when RNC connect to NSN SGSN


Huawei did a trace on PK-RNC and analyzed the flow when RNC connect to NSN
SGSN. Please refer to below flow:

Figure 5 NSN SGSN flow chart


The difference is when RNC send RAB_release_req to NSN SGSN, NSN SGSN
dont response this message immediately and keep sending Deactivation PDP Req
message to MS. After PDP deactivation flow finish, NSN SGSN would response
RAB_release_Req message with Iu release CMD which reason is normal-release. In
that case, RNC would not record this as GTPU.Loss error.

5 Summary
1. It will not affect the customer experience.
2. It doesnt affect the throughput since subs who have GTPULoss are no-balance
subs.
2016-6-17

RNC PS call drop rate increase after migration to Huawei SGSN

3. It wouldnt affect the billing since problem only occur on no-balance subs.

6 Next action
Huawei would release a SGSN patch for optimization. Patch number is SPH510
which would be release by 24th Jan 2010.

2016-6-17

You might also like