Professional Documents
Culture Documents
Application Paper: iRAT Handover Analysis
Application Paper: iRAT Handover Analysis
Analysis
Introduction
NSA Software allows you to monitor all incoming and outgoing legs of:
3G -> 2.5G iRAT cell changes for PS services (combined RAU / LAU)
all 2.5G -> 3G iRAT cell changes for PS services (combined RAU /
LAU) except for Gs interfaces
How to find an iRAT handover and cell change in the NSA Call Table
9-Dec-13, Page 1
Step 5: The target BSC allocates radio resources, especially a time slot
for the connection, in the target cell. The signaling of this procedure
can be monitored on the Abis interface.
9-Dec-13, Page 4
9-Dec-13, Page 5
9-Dec-13, Page 6
9-Dec-13, Page 7
9-Dec-13, Page 12
9-Dec-13, Page 13
iRAT HO Analysis
In the iRAT 3G-2G Handover Matrix different panes/chapters are used to
display the data. Main chapters are:
1. iRAT 3G-2G Handover Matrix
2. Cell Radio Parameters (3G cells only)
3. Mobility Indicators (for 2G cell only)
After a carefully case study of 3G-3G relocations scenarios it was decided that
a matrix view of 3G-3G relocations (inter-RNC hard handover) does not
provide meaningful results, because
Since best correlation results are expected with cell information provided by
operators the import of a combined 2G/3G cell info file is required as
prerequisite for the matrix algorithms.
9-Dec-13, Page 14
Source Cell 3G
This is the unambiguous identity of the source cell where the UE receives the
HO Command.
In the RANAP Init.Mgs. for Relocation Preparation the SAC of the source cell
is found together with the global cell ID (GCI) of the target ID. In the matrix
these IDs are displayed as source cell = RNC-ID+LAC+SAC (where RNC-ID
is derived from topology or cell info file).
Cell name, uARFCN and Primary Scrambling Code (Primary SC) are also
derived from cell info file.
9-Dec-13, Page 15
In this section the cell global identity is shown as derived from Init.Mgs.
Relocation Preparation = LAC+CI.
The BCCH ARFCN as well as BCC and NCC belong to a target TRX of a 2G
cell and are found in RRC Measurement Control and DMTAP Handover
Command (HCOM). (Refer to the figure below.)
Figure 10. Message Example 2: RRC Measurement Control for iRAT measurements
9-Dec-13, Page 16
BSC
TRX 0
TRX 1
Cell 1
BTS 1
TRX 2
Cell 2
TRX 3
BTS m
Cell n
TRX 255
Figure 11. General Relation between TRX and CGI
>>>Band indicator
MP
Enumerated (DCS
1800 band used,
PCS 1900 band
used)
Indicates how to
interpret the
BCCH ARFCN
>>>BCCH ARFCN
MP
Integer (0..1023)
[45]
9-Dec-13, Page 17
Figure 12. Message Example 3: RRC Measurement Report with iRAT measurement results
9-Dec-13, Page 18
Figure 13. Message Example 4: RRC Handover from UTRAN Command/DMTAP Handover Command
There is a fix mapping between BCCH ARFCN and the used GSM frequency
band:
GSM Frequ. Band
ARFCN
Comment
GSM 400
Dynamic
Not used
GSM 400
Dynamic
Not used
GSM 400
259293
Tansania only
GSM 400
306 340
Tansania only
GSM 700
Dynamic
Not used
GSM 700
438511
Not used
Dynamic
Not used
GSM 850
128251
America
GSM 900
1124
Worldwide
GSM 900
GSM 900
GSM 900
Dynamic
Not used
GSM 1800
512885
Worldwide
GSM 1900
512810
America
BCCH
LAC+CI
ARFCN/BCC/NCC
[dBm]
75/5/4
GSM 900
74
308-371
9-Dec-13, Page 19
Figure 14. Relocation Preparation Failure and Successful Relocation in same call
Reloc. Prep. Attempts. This is the number of RANAP Initiating Message for
procedure code relocation preparation
Distinct calls with Relocation Preparation Failure: This is the number of global
call IDs in which one or more Relocation Preparation Failures have been
found. Since the Relocation Preparation procedure in case of failure is
repeated multiple times in the same call a single problem of an individual
subscriber can lead to high failure ratios of the procedure KPI while subscriber
impact is low. This column helps to validate the subscriber impact.
Top Cause Reloc. Prep. Failure: These are the cause values seen most often
in Relocation Preparation Failure message (RANAP Unsuccessful Outcome for
procedure code relocation preparation.
For the sample call in Figure 14 the matrix delivers the following results:
9-Dec-13, Page 20
9-Dec-13, Page 21
9-Dec-13, Page 23
9-Dec-13, Page 24
The contents of the columns in the 3G-2G PS CCO section of the matrix is
defined as follows:
PS 3G-2G CCO Att. This is the number of RRC Cell Change Order from
UTRAN messages.
Top Cause CCO Failure reported by UE. This is the cause value seen most
often in RRC Cell Change Order Failure.
9-Dec-13, Page 25
9-Dec-13, Page 26
9-Dec-13, Page 27
Traffic Load
The Traffic Load pane, in the iRAT 3G-2G Handover Matrix, displays the
following KPIs (for the 3G source cell only).
iRAT CRS Mobility Ratio %. This formula works the following way: each interRAT cell reselection of an UE changing from 2G to 3G leads to a combined
Location/Routing Area Update (Loc. Update cause = normal LUP). The
number of iRAT cell reselections is the number of successfully established
RRC connections for establishment cause inter-RAT-cell-reselection. The
result of the formula shows how many normal Location Updates caused by
mobility of UEs are due to iRAT cell change 2G-3G. A value > 50% indicates
that there are more UEs toggling between 2G and 3G than UEs moving from
one 3G LAC to another.
NOTE: During system test phase it was observed that in short sessions defective
handsets can leverage the equation result to values higher than 100%. If such
values are seen this is a clear indication to make a more detailed investigation of
the Location Update performance in the source cell, e.g. by filtering on all calls
with RRC Establishment Cause = "inter-RAT cell reselection" of this cell and make
column statistics on subscriber IDs.
Outgoing iRAT PDP Cntx. Mobility Ratio %. This KPI shows how many active
PDP Contexts from the source cell are leaving to 2G.
Incoming iRAT PDP Cntx. Mobility Ratio %. This KPI shows how many active
PDP Context in the cell are coming from 2G.
CS outgoing iRAT Mobility Ratio %. This KPI shows how many CS RABs that
9-Dec-13, Page 28
9-Dec-13, Page 29
PS vs. CS iRAT Handover Ratio %. This KPI shows the percentage of PS iRAT
Cell Change compared to all iRAT handover (CS + PS). It allows determining if
more packet calls or more voice calls are handed over to 2G.
9-Dec-13, Page 30
9-Dec-13, Page 31
Problem discovery
As it is already known, the TrendNavigate tool provides more than one starting
points for discovering the network issues. One such report is the Top Cells
report.
In our case the Top Cells report can be generated based on several KPIs, e.g.
3G-2G CS Handover Failure Ratio.
The method of getting this report is simple. From the Reports menu, click Top
Cells. In the displayed screen, in the right pane, select the report criteria and
the KPI 3G-2G CS Handover Failure Ratio from the KPI list and click Apply.
The following figure shows a view of the Top Cells Report.
This report contains all the cells with 3G-2G Handover Failure, and gives
details on the number of Attempts recorded and for each attempt the sub9-Dec-13, Page 32
An equally good and valuable way of discovering iRAT issues is the KPI Cause
CPT Report. This report can be opened from the toolbar menu KPI Cause. In
the displayed report window select the criteria from the right navigation panel.
In the Additional right navigation panel, select the 3G-2G CS Handover Failure
Ratio in the KPIs list.
9-Dec-13, Page 33
Figure 27. KPI Cause Report for "3G-2G CS Handover Failure Ratio"
In this graphic we can see the causes and sub-causes for our selected KPI.
The main cause is Relocation Cancel and the sub-causes are Failure in the
Radio Interface Procedure and Radio Connection with UE lost, along with one
case where the sub-cause is mentioned as Cause not distributed.
Having seen the Iu Report for 3G-2G CS Handover Failure analysis, further
investigations into the messaging of call flow are possible, to find out the root
cause. The root cause can either be in the hardware (RNC, UE) or in the
planning of the network, (iRAT neighbor definitions, missing or one-way
neighbors, or poor GSM signal of the 2G neighbor).
All of this requires a detailed analysis in the messaging to compare the values
reported in the Measurement reports for various 2G Cells.
For this purpose, the first need is to look at all those calls that contributed to
the KPI 3G-2G CS Handover Failure Ratio in the UMTs network.
TrendNavigate offers this possibility to getting the call table containing only our
intended calls.
9-Dec-13, Page 34
Figure 28. Right-Click on the bar graph gives the option to Open Call Table
Click the Open Call Table option to open the NSA Call Analysis application. It
gets started with the auto-filtered call table for the desired KPI.
It is worth mentioning here that if you opened the call table from the top left
graph in the KPI Cause report, the call table will contain calls for all subcauses. You can also open the call table for one individual sub-cause by rightclicking on the graph for the particular sub-cause. Refer to figure 27.
The call table has the flexibility to show RANAP Release Causes. In this case
the Call Table looks like in figure below:
Each row in the call table represents one Global Call and is capable of being
drilled down to the single message level.
9-Dec-13, Page 35
Root Causes
b. There are 2G neighbors in the vicinity; they are the right candidates for
iRAT Handover. The RNC gives the Relocation Preparation Command,
a signal to go ahead and perform the Handover to 2G system. This is
shown in the figure 31 below.
9-Dec-13, Page 36
Figure 33. Handover from UTRAN Failure, cause "Physical Channel Failure"
9-Dec-13, Page 37
Figure 34. RNC sends "Relocation Cancel" Command, cause "Failure in the Radio
Interface Procedure"
This process repeats several times itself (Figure 35 below) and finally the call
is released with the cause Radio Connection with UE Lost (see Figure 37)
Figure 35. UE tries several times during the call to perform iRAT HO to 2G, but fails
each time.
9-Dec-13, Page 38
Figure 36. All of the reported GSM neighbors are too weak to perform the HO.
Figure 37. Finally, RNC send the Iu-Release command, cause "Radio Connection with
UE lost"
9-Dec-13, Page 39
Figure 38. RSCP, EcNo and TrCh BER before the call dropped
Figure 38 shows clearly that the RSCP of the serving cell is deteriorating,
EcNo is below -20 dB and the TrCH BER is ~ 25%. At such Radio
environment, it is impossible for the communication to continue.
Following figure (39) shows the RSCP value only just before the call drop:
Figure 39. RSCP Value of the serving cell before the call dropped.
9-Dec-13, Page 40
Some of the 2G neighbors are reported having good GSM RSSI, but
still the iRAT HO fails.
A snapshot of the excel sheet for iRAT HO Failure Analysis shows the GSM
RSSI is generally very low in the network, although there are some instances
of good GSM RSSI also reported in iRAT HO Failures.
9-Dec-13, Page 41
In this case, it is obvious that the 2G coverage is not so good, but to give it a
quantitative aspect, pull out some reports on RNC level and then on some
selected cell levels to see how the GSM RSSI looks like.
Figure 41 reveals that the general RMS value of the GSM RSSI is rather on the
lower side. Majority of the GSM carriers appears to be below -90 dBm, with
some of them being as low as -110 dBm. This trend indicates that the GSM
values reported by UEs are not favorable in most of the cases to support the
iRAT HO to 2G carriers.
One recommendation is to add more 2G sites to the network so as to bring the
average around -85 dBm. That will enhance significantly the chances of iRAT
HO to 2G being successful.
GSM RSSI value histogram for cell-id 61722 is shown below in the figure 19
below.
Figures 42 and figure 43 show the Histogram for GSM RSSI for two selected
cells.
9-Dec-13, Page 42
From the sample, cells selected for GSM RSSI histogram, it is shown that the
majority of the measurements report GSM values below -90 dBm. This coupled
with the average GSM RSSI for the whole RNC (Figure 41) is sufficient to
believe that the network lacks the absolute required level of GSM coverage
and the call will keep dropping due to unsuccessful iRAT HO to 2G, unless
some measures are taken to boost the average value of 2G signal in the
network.
PS Data Calls
9-Dec-13, Page 43
Figure 44. iRAT Hanover Matrix: Set Filter for 3G-2G HO Execution Failure Ratio
c) Scroll to the left side to see names of source cell/target cell and radio
conditions in target cell. Based on this information the RAN engineers
or consultants can prepare a list of executable actions.
9-Dec-13, Page 44
the
failure
ratios
really
low
after
successful
Figure 46. Source/ Target Cell and Radio Conditions View at HO decision
9-Dec-13, Page 45
9-Dec-13, Page 46
Problem Statements: During Location Update procedure, the mobile does not
listen to received Paging, so we should count the number of Paging sent to a
mobile when it is doing the LUREQ procedure (Paging received between
rrcConnectionRequest (registration) and LUACC). This sometimes is called
collision. Especially it is often seen that one IMSI makes 60 to more than 100
LUREQ/hour in one cell, because it is toggling in IDLE mode (not signaling
sent) between 2G and 3G LAC or between two different 3G LACs. This kind of
LAC toggling creates a lot of unwanted signaling load on RNCs.
Optimization Target: Reduce Signaling Load in RNC by eliminating ping-pong
Location Updates improve customer QoE in parallel, because ping-pong LUP
leads to high number of paging failures in turn
How to proceed in TrendNavigate:
a) When looking into iRAT Matrix you see iRAT CRS Mobility Ratio, which
is number of successful RRC Conn. Setups with est. cause inter-RAT
Cell Reselection vs. number of Loc. Upd. normal.
As we know, each inter-RAT Cell Reselection triggers a combined
LUP/RAUP normal and subscribers can only come from 2G to 3G cell
or form different 3GLAU to our monitored 3G LAU (RNC). Looking at
example below it is seen 3971 LUP normal and 75.8% of the are due
9-Dec-13, Page 47
b) Now there is a possibility to verify this thesis in the following way using
the call table:
a. Filter on the first UTRAN cell = cell-ID of Lafitte* and RRC
Establishment Cause = inter-RAT Cell Reselection (3G-2G PingPong) or registration (3G-3G Ping-Pong)
b. Run column statistics on IMSI to see if same subscribers pop up all
the time, filter on single IMSIs for more details.
Sample: it can be seen same subscriber in same cell registering again and
again
9-Dec-13, Page 48
If you need stats how often IMSI register in same cell you can run XML KPI. If
you further want to know if they have old LAC = new LAC you can also build
XML counter for this.
9-Dec-13, Page 49
9-Dec-13, Page 50
Thus any call that contains entries in the Handover Procedure column contains
intersystem cell changes.
9-Dec-13, Page 51
For packet switched calls the handover history is not populated. Hence, calls
with intersystem cell changes can be detected via the SGSN, BSC and RNC
columns in the Call Table. A call that has had an intersystem cell change
includes the BSC and RNC columns populated accordingly.
9-Dec-13, Page 52
Also in this view, a call that has had an intersystem cell change shows entries
in the BSC and RNC columns.
9-Dec-13, Page 53
Abbreviations
2G
3G
ALCAP
AMR
ARFCN
BCC
BCCH
BSIC
BSSAP
BSSGP
CPT
CS
Circuit Switched
CGI
DCH
Dedicated Channel
DL
Down link
DMTAP
Ec/N0
GSM
HO
HS-DSCH
ID
Identifier
IMSI
LAI
LAPD
LAU
MSC
NAS
NBAP
NEM
NSA
NSAP
NSAPI
PCU
PDP
PLMN
PS
Packet Switched
PSC
P-TMSI
Packet TMSI
QoS
Quality of Service
QPSK
RAB
RAI
RAN
RANAP
RAU
RB
Radio Bearer
RL
Radio link
RLC
RNTI
RRC
RSSI
SAC
SC
SGSN
SRNC
TMSI
TRX
u-ARFCN
UE
User Equipment
UL
Up link
Um
UP
User Plane
UTRAN
VLR
9-Dec-13, Page 56
Technical Support
For support contact your regional Technical Assistance Center:
EMEA
phone
+373 22 879020
fax
+373 22 879001
Tektronix-nd-tac-emea@tek.com
Americas
phone
e-mail
Asia Pacific
phone
e-mail
9-Dec-13, Page 57
9-Dec-13, Page 58