3G Cluster Packet Service Optimization Report

Security Level
3G Cluster Packet Service Optimization

Kickoff REPORT
DATE: 28-June-2012
Table of Contents
2016-02-11
Huawei Proprietary - Restricted Distribution
Page1, Total72
Security Level
Introduction............................................................................................................................................................................3
Tools.......................................................................................................................................................................................3
Contact Information...............................................................................................................................................................3
Analysis Procedure.................................................................................................................................................................3
4.1
PS CSSR Optimization.......................................................................................................................................................4
4.1.1
Pre Analysis....................................................................................................................................................................4
4.1.2
PS CSSR Optimization...................................................................................................................................................5
4.2
PS CDR Optimization......................................................................................................................................................19
4.2.1
Introduction..................................................................................................................................................................19
4.2.2
FACH Congestion........................................................................................................................................................19
4.2.3
F3 Strategy Modification.............................................................................................................................................20
4.3
PS Throughput Optimization...........................................................................................................................................21
4.3.1
Fundamental Check......................................................................................................................................................21
4.3.2
RTWP Optimization.....................................................................................................................................................21
4.3.3
TCP Accelerator...........................................................................................................................................................32
4.3.4
CQI Adjustment Based on Dynamic BLER Target.....................................................................................................36
4.3.5
HSUPA Adaptive Re-Transmission..............................................................................................................................37
Smart Phone Optimization...................................................................................................................................................38
5.1
Enhanced Fast Dormancy.................................................................................................................................................38
5.2
Multi-RAB Optimization.................................................................................................................................................46
5.3
Call Re-establishment......................................................................................................................................................59
Summary and Conclusion....................................................................................................................................................61
Schedule...............................................................................................................................................................................66
2016-02-11
Page2, Total72
Security Level
1 Introduction
According to Ongoing Service contract for 2012, Cluster Packet Service Optimization is one of the
defined work packages for which Huawei will take the responsibilities.
PP12 was selected for 3G packet service optimization
as an example. The Analysis and
recommendations will be provided to improve PS CSSR, CDR and throughput, etc. Meantime, some
improvement might be obtained on Circuit Switching Domain as well. As agreed, the best practices and
experiences verified in PP12 will be rollout in entire network by Vodacom.
This document is intended to explain the contracted 3G cluster packet service optimization process,
tools & the deliverables attached with this service. The RF design related issues and optimizations are
out of the scope of this activity.
2 Tools
1) Huawei Configuration Management Express (CME) for configuration data collection Integrated on
M2000
2) Huawei PRS (Performance Reporting System)
3) Sonar
3 Contact Information
Following will be the contact person from Huawei for this work package:
NO.
Name
Mail
Cell Phone
Sagar
sagar.bajaj@huawei.com
+27 715172241
Bob Sun
sunmingfa@huawei.com
+27 763186113
4 Analysis Procedure
This activity is divided into 4 categories:
2016-02-11
PS Call Setup Success Rate Optimization

PS Call Drop Rate Success Rate Optimization
PS Throughput Optimization
Smart Phone Optimization
Page3, Total72
Security Level
4.1
PS CSSR Optimization
4.1.1 Pre Analysis
CSSR (Call Setup Success Rate) = RRC Setup Success Rate (Service) * RAB Setup Success Rate
The PS CSSR is around 99%.
The RRC Setup Success Rate (service) is around 99.8%, so no big room for further optimization. So
the CSSR optimizations will be focusing on PS RAB Setup Success Rate.
2016-02-11
Page4, Total72
Security Level
From the graphs above, we can see the main contributor to the PS RAB failures are congestions.
On Huawei RAN, the congestions are pegged during Admission stage, admission failures caused
by Code, Uplink/Downlink power, Uplink/Downlink Channel Element, Iub bandwidth should be
2016-02-11
Page5, Total72
Security Level
checked.
The pie chart above indicates that about 85% of the congestions are caused by UL Channel
Element and Uplink Power.
These two aspects should be the main optimization targets.
4.1.2 PS CSSR Optimization

a) Fundamental Check
Before any fine tunings on some parameters, some fundamental checks as per the project
parameter baseline are required.
Items:
All Co-antenna cells should be configured with Blind Handover Inter Frequency Neighbors
(BlindHoFlag=TRUE)
Non Co-antenna cells cant be configured with Blind Handover Inter Frequency Neighbors
(BlindHoFlag=FALSE)
The EcNo threshold for candidate target DRD cell is -13dB (DRDEcN0Threshhold).
MO
UINTERFREQNCELL
Parameter ID
DRDEcN0Threshhold
Parameter Name
Drd Ec/No Threshold
NE
BSC6900
MML Command
ADD UINTERFREQNCELL
MOD UINTERFREQNCELL
Meaning
DRD Ec/No threshold for determining whether to perform the blind handover. If
the measured Ec/No of the current cell is greater than this parameter, this cell
can be the candidate cell for DRD.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-02040001
Feature Name
Intra System Direct Retry
Value Type
Interval Type
2016-02-11
Page6, Total72
Security Level
GUI Value Range
-24~0
Enumeration
Number/Bit
Not involved
Unit
dB
Actual Value Range -24~0

Default Value
-18
Recommended
Value
-18
Impact
Cell
Parameter
Relationship
None
Access
Read & Write
Service Interrupted No
After Modification
Interruption Scope
Not involved
Interruption
Duration(Min)
Not involved
Caution
None
Validation of
Modification
The parameter modification has no impact on the equipment.
Impact on Radio
Network
Performance
None
Introduced in
Version
Versions earlier than BSC6900 V900R011
Attribute
Radio
SIB11 should be ON
MO
UINTERFREQNCELL
Parameter ID
SIB11Ind
2016-02-11
Page7, Total72
Security Level
Parameter Name
SIB11 Indicator
NE
BSC6900
MML Command
ADD UINTERFREQNCELL
MOD UINTERFREQNCELL
Meaning
Indicates whether to send the system information block 11 (SIB11) including

neighboring cell information. The value "FALSE" indicates that the neighboring
information is not included in the SIB11. The value "TRUE" indicates that the
neighboring information is included in the SIB11.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-010801
WRFD-010802
Feature Name
Intra RNC Cell Update

Inter RNC Cell Update
Value Type
Enumeration Type
GUI Value Range
FALSE(Do not send), TRUE(Send)
Enumeration
Number/Bit
FALSE~0, TRUE~1
Unit
None
Actual Value Range FALSE, TRUE

Default Value
TRUE
Recommended
Value
TRUE
Impact
Cell
Parameter
Relationship
None
Access
Read & Write
After Modification
Interruption Scope
Not involved
2016-02-11
Page8, Total72
Security Level
Interruption
Duration(Min)
Not involved
Caution
None
Validation of
Modification
Impact on Radio
Network
Performance
None
Introduced in
Version
Attribute
Radio
The Priority for Coverage based interfreq handover between Co-antenna cells is 0
MO
UINTERFREQNCELL
Parameter ID
HOCovPrio
Parameter Name
Priority of Coverage-Based Inter-Frequency Handover
NE
BSC6900
MML Command
ADD UINTERFREQNCELL
MOD UINTERFREQNCELL
Meaning
Priority for neighboring cell supporting coverage-based inter-frequency

handover. The value "0" indicates that the coverage-based inter-frequency
handover is not supported. The value "1" indicates that the cell is assigned with
the highest priority for the handover. The value "3" indicates that the cell is
assigned with the lowest priority for the handover. If a higher priority is assigned
to a cell, the probability for selecting the cell as the measurement object and the
target cell for the handover increases. For example, the cell with priority 1 has
more chance to select as the measurement object and the target cell for the
handover than the cell with priority 2.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-020302
Feature Name
Inter Frequency Hard Handover Based on Coverage
2016-02-11
Page9, Total72
Security Level
Value Type
Interval Type
GUI Value Range
0~3
Enumeration
Number/Bit
Not involved
Unit
None
Actual Value Range 0~3

Default Value
Recommended
Value
Impact
Cell
Parameter
Relationship
None
Access
Read & Write
After Modification
Interruption Scope
Not involved
Interruption
Duration(Min)
Not involved
Caution
None
Validation of
Modification
Impact on Radio
Network
Performance
None
Introduced in
Version
Attribute
Radio
Findings on PP12:
2016-02-11
2 cells are not configured with Blind Handover Neighbors.

10 neighbors SIB11 are OFF.
Page10, Total72
Security Level
38 cells: The Priority for Coverage based interfreq handover between Co-antenna cells is
2.
40 cells: The EcNo threshold for candidate target DRD cell is -18dB.
Benchmark:
Slight improvement is seen on PS CSSR after implementation on these cells.
b) UL Power Congestion Optimization

As default, uplink power admission is based on equivalent user number (ENU): ALGORITHM_SECOND.
MO
UCELLALGOSWITCH
Parameter ID
NBMUlCacAlgoSelSwitch
Parameter Name
Uplink CAC algorithm switch
NE
BSC6900
MML Command
ADD UCELLALGOSWITCH
MOD UCELLALGOSWITCH
Meaning
The algorithms with the above values represent are as follow: ALGORITHM_OFF:
Disable uplink call admission control algorithm. ALGORITHM_FIRST: The load
factor prediction algorithm will be used in uplink CAC. ALGORITHM_SECOND: The
equivalent user number algorithm will be used in uplink CAC.
ALGORITHM_THIRD: The loose call admission control algorithm will be used in
uplink CAC.
IsKey
NO
Mandatory
YES
Feature ID
WRFD-020101
2016-02-11
Page11, Total72
Security Level
Feature Name
Admission Control
Value Type
Enumeration Type
GUI Value Range
ALGORITHM_OFF, ALGORITHM_FIRST, ALGORITHM_SECOND, ALGORITHM_THIRD
Enumeration
Number/Bit
ALGORITHM_OFF~0, ALGORITHM_FIRST~1, ALGORITHM_SECOND~2,

ALGORITHM_THIRD~3
Unit
None
Actual Value Range ALGORITHM_OFF, ALGORITHM_FIRST, ALGORITHM_SECOND, ALGORITHM_THIRD

Default Value
None
Recommended
Value
None
Impact
Cell
Parameter
Relationship
None
Access
Read & Write
After Modification
Interruption Scope
Not involved
Interruption
Duration(Min)
Not involved
Caution
None
Validation of
Modification
Impact on Radio
Network
Performance
None
Introduced in
Version
Attribute
Radio
ENU Calculations:
2016-02-11
Page12, Total72
Security Level
ENU Counter:
Counter VS.RAC.UL.EqvUserNum can indicate the real ENU for each cell.
Worst performing cells in PP12 with Uplink Power congestion:
Recommendations:
2016-02-11
F3 Upgrade if possible
Page13, Total72
Security Level
Increase admission threshold (impact all types of admission control)

Disable uplink CAC (Call Admission Control)
Benchmark for disabling UL power CAC on severe congested cells:
Uplink Power Congestions are cleared, and PS CSSR is around 99.5% for these cells. CE, Iub,
Code and DL power admission controls are still valid.
Practices to relieve UL power congestion in SGA region:

In SGC region, current CAC algorithm parameters were fine tuned in order to relieve the congestion
problem. Please note, in this study, the algorithm itself is still active and the triggering conditions
have been optimized only.
For this study, most congested 16 sites were selected as candidate sites from entire region. The
changes were implemented for 49 cells from 5 different RNCs. Candidate cells were selected
according to the UL Power Congestion rates. It was the key measure that cell has major UL Power
Congestion.
In current network policy, the max allowed ENU percentage is set 65, which is equal to 130 ENU out
2016-02-11
Page14, Total72
Security Level
of total allowed 200 for PS service in UL direction for all cells. With fine tuning of parameters, allowed
UL ENU percentage is set to 90, which increased the total allowed number from 130 ENU to 180 ENU.
In case of sudden traffic increase, since algorithm is still active but with an optimized set of
thresholds, in case of high load, required CAC actions will be triggered and system stability will be
kept constant.
Below are old and new values regarding to the change. Please note, the threshold values for Conv
and non-Conv AMR services have been changed to satisfy the
UlNonCtrlThdForHo > UlNonCtrlThdForAMR; UlNonCtrlThdForAMR > UlNonCtrlThdForOther
The rejects due to other types of congestion (i.e. Code, CE and Iub) have been monitored. In case of
an excessive load (because of the increased threshold) actions are expected to be triggered by this
algorithm.
It is possible to increase the thresholds further, but LDM parameters also have to be modified to do so
(because of parameter relation)
Results:
2016-02-11
Page15, Total72
Security Level
c) UL CE Optimization
CE congestions are mostly on uplink as on downlink the dedicated Mac-hs element processes
HSDPA traffic.
As per Huawei equipment specifications, one Resource Group can support 6 cells, for S333 sites,
9 cells have to be divided into 2 Groups. CE resources cant be shared between different Uplink
Resource Groups, so there is a possibility that one group is congesting and another one is not
fully utilized. Furthermore, as congestions are only pegged during admission stage, so its
possible that the overall NodeB CE utilization is not high but the CE congestions are seen due to
data service behavior (resource is always reserved for each user).
Worst performing cells in PP12:
Top 3 sites are already configured with 4*WBBP, but only one for F3. Many congestions are seen
on F3 cells. It is recommended to move one WBBP to Group1 (configuration).
RED Line: Uplink resource group reconfiguration
2016-02-11
Page16, Total72
Security Level
GREEN Line: Disable cell level CE CAC
After Uplink resource group adjustment, congestions moved from F3 to F1&F2 (even more). The
root cause is still that the different uplink resource groups cant be shared.
Disabling Cell level credit (CE) CAC (NodeB level CE CAC still valid), CE congestion reduced.
The graph above shows very low CE utilization for these sites.
2016-02-11
Page17, Total72
Security Level
Conclusion:
For 4*WBBP sites, disabling cell level CE CAC is a better solution.
The UL CE congestion cant be cleared after disabling cell level CAC as NodeB level CE CAC is
still valid.
Some other features and functions can be considered to reduce CE congestion, such as Fast
Dormancy, HSUPA TTI reconfigurations and HSUPA GBR adjustment.
HSUPA GBR Adjustment:
The main contributor to the uplink CE congestion is HSUPA traffic.
According to the statistics, the average HSUPA throughput per user is <40kbps.
The GBR (Guaranteed Bit Rate) for HSUPA is 64kbps in live network, so some resources are
wasted for admission.

GBR is only used for admission control, users can get higher throughput via DCCC (Dynamic
Channel Configuration Control)

Recommendation: Change the HSUPA GBR from 64kbps to 32kbps.
Less PS congestions after GBR changed from 64kbps to 32kbps.

d) Enable DRD Failure Rollback function
DRD (Direct Retry Decision):
If the UE requests a service setup or channel reconfiguration in an area covered by multiple
frequencies, the RNC sets up the service on a carrier with a light load to balance out the load
among the cells on the different frequencies.
Current PP12s DRD Successful Rate is around 99%.
2016-02-11
Page18, Total72
Security Level
Recommendation:
Enable DRD failure rollback function to avoid some RAB setup failures.
MML command:
SET URRCTRLSWITCH:PROCESSSWITCH2=RNC_RBRECFG_DRD_FAIL_ROLLBACK_SWITCH1&RNC_RBSETUP_DRD_FAIL_ROLLBACK_SWITCH-1;
Description:
RNC_RBRECFG_DRD_FAIL_ROLLBACK_SWITCH (RB Reconfiguration DRD Rollback Switch):
When the switch is turned on, the RNC supports rollback caused by failed DRDs of RB
reconfiguration. When the switch is turned off, the RNC does not support rollback caused by failed
DRDs of RB reconfiguration.
RNC_RBSETUP_DRD_FAIL_ROLLBACK_SWITCH (RB Establishment DRD Rollback Switch) :
When the switch is turned on, the RNC supports rollback caused by failed DRDs of RB
establishment. When the switch is turned off, the RNC does not support rollback caused by failed
DRDs of RB establishment.
Benchmark on PP12:
Slight improvement on PS RAB Setup Success Rate
2016-02-11
Page19, Total72
Security Level
4.2
PS CDR Optimization
4.2.1 Introduction
Besides of some RF design issues, there are two main contributors to the PS Call Drop Rate in live
network:
FACH Congestion causing PS Call Drops on Common channels

Correlation between PS CCH drop rate and FACH congestion:
F3 idle barred strategy

As Cell_FACH and Cell_PCH is enabled, if F3 is idle barred, when UE transit to Cell_FACH on
F3, it will reselect back to F1&F2 (SIB12 is not enabled). Ping-pong reselections between
different layers increase the possibility of drops, negatively impact subscribers perception
as well.
4.2.2 FACH Congestion

As default, only one SCCPCH (Secondary Common Control Physical Channel) is configured for each
cell. Paging channel (PCH) and FACH are mapped onto the same physical channel SCCPCH as per 3GPP
specifications. And, PCH has higher priority to occupy this channel comparing with FACH. To expand
FACH capacity, 2nd SCCPCH dedicated for FACH is recommended.
2nd SCCPCH addition has already been rolled out in entire network.
4.2.3 F3 Strategy Modification

Current F3 Strategy:
2016-02-11
Page20, Total72
Security Level
Randomly camping on F1&F2

F3 is idle barred
DRD algorithm based on User Number
Load based blind handover between different layers
Proposed Strategy for PP12 (non-continuous F3 footprint):

Change from F3 barred to inclined camping on F1&F2 in case F3 is not continuous.
- Set a big Qoffset2 between (F1,F2) & F3 to make idle UE mainly camp on F1&F2 to reduce the
mobility CS drops on F3.
Inclined camping on F1&F2

F3 is not idle barred
Benchmark for F3 strategy modification:

Improvement on F3 PS CDR after implementation (from 1.9% to 1.3%)
2016-02-11
Page21, Total72
Security Level
As for the areas with continuous F3 footprint such as SGS region, the proposed strategy is:
4.3
Randomly camping on F1&F2&F3

F3 is not idle barred
PS Throughput Optimization
4.3.1 Fundamental Check
Items:
HSDPA/HSUPA function should be activated.

64QAM, Downlink Enhanced Layer2, Dual-Carrier HSDPA should be activated.
4.3.2 RTWP Optimization

These section only targets at high RTWP caused by high traffic, abnormalities caused by external
interference, engineering issues need specific inspections.
a) Background Noise Alignment
As default, the background noise is 61, which means -112+61/10=-105.9dBm.
2016-02-11
Page22, Total72
Security Level
As default, the maximum target load is 75%, means 6dB lift.
Under these settings, when RTWP=-105.9dBm+6dB=-99.9dBm, the uplink load has already
2016-02-11
Page23, Total72
Security Level
reached maximum load target. Consequently, the HUSPA throughput cant be scheduled more. In
the live network during the peak hours, many RTWP spikes can exceed this value which are causing
low HSUPA throughput.
This activity is to align the configured background noise with actual RTWP. We can regard the
minimum RTWP during the early morning (02h00-03h00) as real background noise.
b) CQI Feedback Period Optimization
More and more HSDPA connections are active according to the traffic growth trend. HSDPA uplink
feedback channel HS-DPCCH generated a big part of uplink interference which limits the uplink
capacity and HSUPA throughput as well.
Huawei Lab test result:
2016-02-11
Page24, Total72
Security Level
Highest HSDPA User Number vs Mean RTWP in PP12: High RTWP caused by HSDPA traffic.
Recommendation:
Prolong CQI Feedback period from 2ms to 8 ms
Pros:
Improve the RTWP, uplink capacity and HSUPA throughput
Cons:
HSDPA scheduling will be slower in fast fading scenario
MML command:
ADD UCELLHSDPCCH:CELLID=11111,CQIFBCK=D8,CQIFBCKFORSHO=D8;
Benchmark for CQI feedback period change on PP12:

2016-02-11
Page25, Total72
Security Level
About 3dB improvement on average RTWP during peak hours, which means we have almost
50% gains on uplink capacity.
c) RACH Access Parameter Optimization

RACH access preamble adopts open-loop power control method, when the coverage is not so
good or with massive RRC attempts, UE TX power will increase quickly, lead to RTWP spikes. In
order to reduce RTWP spikes caused by this, RAC access parameters could be optimized.
Top cells with highest RRC attempts in PP12, the average RTWP is also high:
Recommendations:
2016-02-11
Page26, Total72
Security Level
Parameter Description:
MO
UPRACHUUPARAS
Parameter ID
Constantvalue
Parameter Name
Constant Value for Calculating Initial TX Power
NE
BSC6900
MML Command
MOD UPRACHUUPARAS
Meaning
This parameter specifies a constant used at calculation of the initial transmit power of the first
preamble, to be used in the random access procedure. The formula is as follows:
Preamble_Initial_Power = Primary CPICH DL TX power-CPICH_RSCP + UL interference +
Constant Value. Where, Preamble_Initial_Power is the preamble initial TX power, Primary
CPICH DL TX power is the downlink transmit (TX) power of PCPICH, CPICH_RSCP is the receive
signaling code power of the PCPICH measured by UEs, and UL interference is the uplink
interference. For detailed information of this parameter, refer to 3GPP TS 25.211.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-020501
Feature Name
Open Loop Power Control
Value Type
Interval Type
GUI Value Range
-35~-10
Enumeration
Not involved
Number/Bit
Unit
dB
2016-02-11
Page27, Total72
Security Level
Actual Value Range
-35~-10
Default Value
-20
Recommended Value
-20
Impact
Cell
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Modification
Impact on Radio
None
Network Performance
Introduced in Version
Attribute
Radio
MO
UPRACHUUPARAS
Parameter ID
PreambleRetransMax
Parameter Name
Max Preamble Retransmission
NE
BSC6900
MML Command
MOD UPRACHUUPARAS
Meaning
The maximum number of preambles transmitted in a preamble ramping cycle. For detailed
information of this parameter, refer to 3GPP TS 25.211.
IsKey
NO
Mandatory
NO
2016-02-11
Page28, Total72
Security Level
Feature ID
WRFD-020501
Feature Name
Value Type
Interval Type
GUI Value Range
1~64
Enumeration
Not involved
Number/Bit
Unit
None
Actual Value Range
1~64
Default Value
20
Recommended Value
20
Impact
Cell
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Modification
Impact on Radio
If the value of Max preamble retransmission is too small, the preamble power may fail to ramp
Network Performance
to the required value. This may result in UE access failure. If it is too large, the UE will
repeatedly increase the power and make access attempts, which may result in interference to
other UEs.
Attribute
Radio
MO
2016-02-11
UPRACHUUPARAS
Page29, Total72
Security Level
Parameter ID
PowerRampStep
Parameter Name
Power Increase Step
NE
BSC6900
MML Command
MOD UPRACHUUPARAS
Meaning
The power ramp step of the random access preambles transmitted before the UE receives the
acquisition indicator in the random access process. For detailed information of this parameter,
refer to 3GPP TS 25.211.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-020501
Feature Name
Value Type
Interval Type
GUI Value Range
1~8
Enumeration
Not involved
Number/Bit
Unit
dB
Actual Value Range
1~8
Default Value
Recommended Value
Impact
Cell
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
2016-02-11
Page30, Total72
Security Level
Validation of
Modification
Impact on Radio
If the value of Power ramp step is too large, the access procedure will be shortened, but it is
Network Performance
more likely to cause power waste. If it is too small, the access procedure will be extended in
time, but transmit power resources will be saved. This parameter must be set carefully. In
addition, the higher the parameter value is, the smaller impact "Constant Value for
Calculating Initial TX Power" has on the network performance.
Attribute
Radio
Benchmark:
2016-02-11
Page31, Total72
Security Level
Significant RTWP improvement on High RRC Attempts sites, HSUPA throughput improved
as well. No degradation on RRC Setup Success Rate.
d) Adaptive PO for HSUPA Traffic Channel

Current HSUPA power parameter configuration is more reasonable for high speed transmission,
in order to adapt to the scenario with many low throughput online users, this feature enables
the RNC to configure an optimal power offset for HSUPA 10 ms TTI UEs with a small target
number of retransmissions based on the rate of the UEs.
If the data rate is lower than 300 kbit/s, a large power offset of E-DPDCH and DPCCH can be
configured. Configuring a larger offset decreases power on the DPCCH and reduces power
consumption on the uplink control channel. In this way, more power can be used to transmit
user data in the uplink, thereby increasing capacity of HSUPA cells.
If the data rate is higher than 300 kbit/s, a low power offset of E-DPDCH and DPCCH can be
configured. Configuring a smaller offset increases power on the DPCCH, thereby meeting power
requirements in cases of multipath searching and channel assessment and improving
performance of HSUPA services.
MML command:
SET UCORRMALGOSWITCH: PcSwitch=PC_HSUPA_DATA_CH_PO_ADAPTIVE_ADJ_SWITCH-1;
Benchmark:
2016-02-11
Page32, Total72
Security Level
No obvious improvement on statistics, as the main gain is obtained when the HSUPA
throughput is greater than 300kbps as per the feature description. In the live network, this
scenario could have a small proportion. Field HSUPA test should be getting more benefit.
4.3.3 TCP Accelerator

The Transmission Control Protocol (TCP) is a reliable transport layer protocol to provide reliable
end-to-end streams of bytes for the Internet. It is widely used in email and file transfer services.
The TCP was initially designed for wired communications.
After the TCP is introduced in wireless communications, the original TCP mechanism cannot well
adapt to the wireless network's characteristics such as high bit error rate and long delay, which
leads to dissatisfactory transmission performance. To improve the performance, the TCP
Performance Enhancer (TPE) is introduced. The TPE is also known as TCP accelerator or TCP
proxy.
The TPE is a functional entity in the RNC. It works as a proxy between the sender and the
receiver. It buffers and processes data to accelerate data transmission and to improve
transmission performance. The position of the TPE entity on the TCP connection is shown as
below.
In uplink transmission, the server is the receiver and the UE is the sender. In downlink
2016-02-11
Page33, Total72
Security Level
transmission, the UE is the receiver and the server is the sender.

The TPE works independently of other network elements (NEs).
The TPE must be implemented before Packet Data Convergence Protocol (PDCP) header
compression.
After the TPE function is activated, only the new TCP connections can benefit from the feature.
TCP Terms:
Term
MSS
Definition
The maximum segment size (MSS) is the largest amount of data (excluding TCP headers) sent
through TCP from the source end to the peer end in a single segment. The MSS is negotiated
when a TCP connection is established between two ends.
Receive
The receive window (rwnd for short) specifies the largest amount of data that can be received
window
by the receiver.
Congestion
The congestion window (cwnd for short) limits the total amount of data that can be sent by a
window
TCP connection. It frequently changes throughout the communication process of the TCP
connection.
Offered
The offered window indicates the largest amount of data that can be received at a moment by
window
the receiver. It is used to limit the largest amount of unacknowledged data allowed by the
sender.
Send
The send window indicates the largest amount of data that can be sent by the sender at a
window
moment. It is equal to the minimum value of cwnd and offered window.
Slow start
When cwnd is larger than or equal to the slow start threshold (ssthresh for short), the TCP
threshold
connection enters the congestion avoidance phase.
RTT
The round trip time (RTT) is the time during which the sender sends a packet to the receiver and
the receiver returns an acknowledgement (ACK).
DupACK
The duplicate acknowledgement (DupACK) is an ACK whose sequence number is the same as
that of the previous ACK. When packet misordering or packet loss occurs during transmission,
the receiver responds the sender with a DupACK.
TCP Reno
Data can be transmitted in sequence only after a TCP connection is established through three
2016-02-11
Page34, Total72
Security Level
handshakes. During the transmission, congestion is controlled through transmission rate

adjustment. The adjustment is based on three variables: cwnd, rwnd, and ssthresh. When the
sender detects packet loss due to retransmission timeout or repeat reception of DupACK, it
adjusts the transmission rate.
The TCP Reno is the most widely used TCP protocol version. It consists of four phases: slow start,
congestion avoidance, fast recovery, and fast retransmission. Here, assume that ssthresh is set
to 64 KB.
Slow start and congestion avoidance:
Fast retransmission and fast recovery:
In the slow start phase, cwnd is set to the two times the MSS. Each time an ACK is received, the
2016-02-11
Page35, Total72
Security Level
value of cwnd is incremented by one MSS (unit: byte). In addition, the TCP connection maintains
the threshold ssthresh. When the value of cwnd reaches this threshold, the TCP connection
enters the congestion avoidance phase.
RTT Timeout
In this case, the sender sets ssthresh to max {cwnd /2, 2 x MSS} and then sets cwnd to 2 x
MSS. The TCP connection enters the slow start phase again.
Repeat reception of DupACK
In this case, the TCP connection enters the fast retransmission and fast recovery phases. The
sender sets ssthresh and cwnd to max {cwnd/2, 2 x MSS}. Each time a DupACK is received, the
value of cwnd is incremented by one MSS. This process is accomplished until the retransmitted
packets are received. Then, the sender sets cwnd to the value of ssthresh. The TCP connection
enters the congestion avoidance phase again.
TPE Technologies
The TPE module function is in the RNC. When establishing a TCP connection, the TPE module
sets up a TPE entity for the TCP connection to process TCP packets. In uplink transmission, the
server is the receiver and the UE is the sender. In downlink transmission, the UE is the receiver
and the server is the sender.
In downlink transmission, the TPE adopts the following technologies to improve the
transmission performance:
ACK Splitting
DupACK Duplication
Downlink Packet Sorting
In uplink transmission, the TPE adopts technologies such as ACK Splitting and Uplink
Packet Sorting. The working principles of uplink transmission are the same as those of
downlink transmission.
ACK Splitting:
In TCP, the congestion window is updated according to the number of received ACK messages
and is expanded by increasing the number of ACK messages. When a slow startup occurs at the
transmitting end, ACK splitting can quickly recover the congestion window; when the
transmitting end works in congestion avoidance mode, ACK splitting can accelerate the
2016-02-11
Page36, Total72
Security Level
expansion of the congestion window.

DupACK Duplication:
In TCP, a lost TCP packet is retransmitted after three DupACK are received. With this feature,
after the TPE receives the ACK message from the UE, the TPE immediately duplicates three
DupACK messages and sends them to the Server if it detects that the packets requested by the
ACK are not in the buffer. This shortens the time for packet retransmission.
Downlink Packet Sorting:
The TPE sorts and transmits the disordered DL packets to avoid unnecessary transmission of
DupACKs in the uplink and to prevent TPE local retransmission caused by disordered packets. In
this way, transmission resources are saved.
Uplink Packet Sorting:
The TPE sorts the UL packets and transmits them to the core network (CN) in sequence. This
avoids the deterioration of the UL data transmission performance caused by the disordered UL
packets.
TCP Accelerator was enabled on 20th June. Slight improvement is seen on throughput
statistically.
2016-02-11
Page37, Total72
Security Level
4.3.4 CQI Adjustment Based on Dynamic BLER Target

Without this feature, the NodeB determines a transmission block size (TBS) based on the channel
quality indicator (CQI) reported by the UE, system resources, and the transport format and resource
combination (TFRC) policy. If the reported CQI and related conditions remain the same, the NodeB
does not change the TBS because it does not consider the ever-changing radio environments. The
constant changes in radio environments, caused by multipath effects and UE mobility, lead to
fluctuating channel quality. Under these circumstances, choosing a TBS based on the reported CQI
makes it difficult to always achieve the optimum downlink throughput.
With this feature, the NodeB monitors the channel quality fluctuations for HSDPA users in a cell in
real time and dynamically selects a proper BLER target based on the monitoring result. The NodeB
then uses the BLER target to adjust the CQI reported by the UE. Based on the adjusted CQI, the
NodeB determines an appropriate TBS to achieve higher downlink throughput for HSDPA users and
higher cell throughput.
2016-02-11
Page38, Total72
Security Level
Note: This feature needs NodeB license support.
4.3.5 HSUPA Adaptive Re-Transmission

HARQ retransmission number is used as the target value of HSUPA uplink outer loop power control.
When UE signal quality is good and uplink transmission power is not limited, a small retransmission
can improve single user throughput. However, when capacity is limited and cell uplink power
becomes a bottleneck, increasing retransmission number can improve cell throughput. Increasing
retransmission number can also boost user cell edge throughput, where UE uplink power is limited.
Therefore theres a need to realize the adaptive adjustment of retransmission number.
With comprehensive considerations of cell uplink power load, CE resources, and limited uplink
coverage, this feature enables the adaptive adjustment of the number of target uplink
retransmissions to improve the throughput per user and cell uplink capacity.
Pros:
In a limited uplink coverage scenario, a users uplink cell edge throughput can be increased, in
order to enhance user experience. According to simulation results, single user throughput has been
show to increase by 15%-60%.
In a scenario where the cell uplink power load is limited, increasing the retransmission number can
2016-02-11
Page39, Total72
Security Level
improve cell throughput and cell uplink capacity. Simulation results have shown an increase of 53%
in cell throughput under multi-user scenarios.
Cons:
UL CE consumption will increase
MML Command:
//RNC MML command
SET URRCTRLSWITCH: PROCESSSWITCH=NODEB_PRIVATE_INTERFACE_SWITCH-1;
SET UCORRMALGOSWITCH: PcSwitch=PC_HSUPA_HARQNUM_AUTO_ADJUST_SWITCH-1;
// NodeB MML command (execute on all the NodeBs under PP12)

SET ADPRETRANSSWTCH: SWITCH =OPEN;
5 Smart Phone Optimization

5.1
Enhanced Fast Dormancy

Some intelligent UEs on a live network send a Signaling Connection Release Indication (SCRI)
message to the RNC after PS data transmission is complete. By sending the SCRI message,
intelligent UEs request for a transition to idle mode or the CELL_PCH or URA_PCH state to reduce
battery consumption. Upon receiving the SCRI, the RNC can release the signaling connection and
put the UE into idle mode, or the RNC can maintain the signaling connection and put the UE to
CELL_FACH or CELL_PCH/URA_PCH state. RSVDBIT1_BIT29 of the RsvdPara1 parameter in the SET
URRCTRLSWITCH command controls the UE through CELL_FACH to CELL_PCH/URA_PCH or directly to
CELL_PCH/URA_PCH. These procedures can reduce battery consumption, depending on the UE
capability and feature activation on the RNC. By default, the RNC puts the UE into idle mode. A UE
setting up PS services in idle mode consumes more signaling resources than that in the CELL_PCH
or URA_PCH state. If there are a large number of UEs transited between idle mode and CELL_DCH
state, signaling storms may occur. The EFD feature puts UEs into the CELL_FACH or CELL_PCH state
to reduce signaling resource consumption and eliminate signaling storms.
Huawei provides a comprehensive end-to-end solution based on network characteristics to respond
to network challenges. EFD is one component of the solution. This feature alone is not enough to
eliminate signaling storms caused by intelligent UEs. However, when applied with professional
2016-02-11
Page40, Total72
Security Level
services, EFD can be used to optimize the quality of service (QoS) for the entire network.
The UEs on a live network can be divided into three types:
UEs that do not send SCRI messages because there is no PS data to be transmitted.
Pre fast dormancy UEs. This type of UE does not read the T323 information element (IE) carried in
the system information block type 1 (SIB1) message. After PS data transmission is complete, this
type of UE sends an SCRI message to the RNC, but the SCRI message does not carry the IE
Signaling Connection Release Indication Cause. Both T323 and Signaling Connection Release
Indication Cause are introduced in 3GPP Release 8.
R8 fast dormancy UEs. This type of UE reads the T323 IE carried in the SIB1 message and sends an
SCRI message to the RNC after PS data transmission is complete. The SCRI message carries the IE
Signaling Connection Release Indication Cause that is set to "UE Requested PS Data session end."
The T323 IE carried in the SIB1 message controls the frequency at which the R8 fast dormancy UEs
send SCRI messages with the IE Signaling Connection Release Indication Cause set to "UE
Requested PS Data session end". The T323 timer starts after an SCRI message is sent by the R8
fast dormancy UEs, and the second SCRI message with the IE Signaling Connection Release
Indication Cause set to "UE Requested PS Data session end" can be sent only after the T323 timer
expires.
The EFD feature applies only to the pre fast dormancy UEs and R8 fast dormancy UEs. Users can
configure EFD to be enabled for only the pre fast dormancy UEs or R8 fast dormancy UEs.
When this feature is disabled, the RNC releases the signaling connection and puts the UE into idle
mode after receiving the SCRI message regardless the UE types.
When this feature is enabled, the RNC performs different processing based on the UE type:
For the pre fast dormancy UEs, the RNC checks for compatibility issues after receiving an SCRI
message. If there are no compatibility issues, the RNC puts the UE into the CELL_FACH or PCH
state. If there is a compatibility issue, the RNC puts the UE into idle mode.
For the R8 fast dormancy UEs, the RNC puts the UE into the CELL_FACH or PCH state after receiving
an SCRI message.
Note:
2016-02-11
Page41, Total72
Security Level
Some pre fast dormancy UEs may have compatibility issues after moving to the CELL_FACH,
CELL_PCH, or URA_PCH state. As a result, new services cannot be initiated. This feature has been
tested by using iPhones and no compatibility issue occurred.
When there is no PS data to be transmitted, the intelligent UE sends the RNC an SCRI message to
request the UTRAN to release the signaling connection. This procedure saves power. In addition, the
UE periodically sends heartbeat messages to the core network, for example, to check for new mails
on the mail server. If the EFD feature is disabled, the RNC releases the signaling connection and
puts the UE into the idle state. In this case, sending heartbeat messages leads to the procedures
for RRC connection setups, authentication, encryption, and RAB setups. These procedures increase
the RNC signaling processing load.
EFD addresses this problem. With EFD, the RNC transits a UE to the CELL_FACH or PCH state
instead of idle mode upon receiving an SCRI message from the UE. In this case, when the UE
periodically sends heartbeat messages, the signaling procedure between the UE and RNC is
simplified because the RRC connection is maintained. As the number of signaling exchange
messages decreases, significant CPU resources can be saved for the RNC, and the UE consumes
about the same low amount of battery power as a UE in idle mode.
Signaling processing procedures for pre fast dormancy UEs and R8 fast dormancy UEs before and
after EFD is enabled
2016-02-11
Page42, Total72
Security Level
Identifying Fast Dormancy UEs

The RNC identifies the pre fast dormancy UEs and R8 fast dormancy UEs using the following
mechanisms:
R8 fast dormancy: The T323 parameter that can be set in the SET UCONNMODETIMER
command is sent to the UE by the SIB1 message. The R8 fast dormancy UEs send an SCRI
message with the IE Signaling Connection Release Indication Cause set to "UE Requested PS
Data session end" after PS data transmission is complete. The RNC identifies the R8 fast
dormancy UEs by reading the value of IE Signaling Connection Release Indication Cause.
This mechanism is standardized by 3GPP Release 8.
Pre fast dormancy: The RNC identifies the pre fast dormancy UEs using the Type Allocation
Codes (TACs). This mechanism, which is not standardized, is used for the early proprietary
implementations of fast dormancy.
The international mobile equipment identity (IMEI), which consists of 14 decimal digits and one
check digit, contains TAC and SNR. TAC indicates a particular type of device that has been approved
by a national GSM/WCDMA approval body. SNR is the serial number identifying the UE. The
structure of the IMEI, which is shown as follows, is specified in 3GPP TS 23.003.
2016-02-11
Page43, Total72
Security Level
To identify pre fast dormancy UEs, FD_TAC_MATCH_SWITCH of the PROCESSSWITCH parameter on

the RNC side needs to be turned on. The TAC parameter has been set for the pre fast dormancy UEs
by the ADD UIMEITAC command. The RNC checks the TAC of the IMEI in the IDENTITY REQUEST
message. If the TAC is included in the EFD list, the UE can use EFD.
State Transition for Fast Dormancy UE
When EFD is disabled, upon receiving an SCRI message from a UE, the RNC releases the UEs RRC
connection. The EFD switches are specified by the RNC-level sub-parameter
FAST_DORMANCY_SWITCH of the parameter PROCESSSWITCH.
When EFD is enabled, state transition procedures vary according to the UE type as follows:
For R8 fast dormancy UEs and pre fast dormancy UEs in EFD list:
-
If the RNC received the SCRI, The RNC transfers the UE to CELL_FACH or CELL_PCH/URA_PCH. The
RSVDBIT1_BIT29 sub-parameter of the RsvdPara1 parameter in the SET URRCTRLSWITCH
command controls whether the UE moves to CELL_FACH or CELL_PCH/URA_PCH.
After PS data transmission is complete, If UE is in CELL_DCH, the PS inactivity timer

(PsInactTmrForFstDrmDch) for EFD in CELL_DCH is started. If the PS inactivity timer for EFD in
CELL_DCH expires, the RNC transfers the UE to CELL_FACH or CELL_PCH/URA_PCH.
After PS data transmission is complete, If UE is in CELL_FACH, the PS inactivity timer

(PsInactTmrForFstDrmFach) for EFD in CELL_FACH is started. If the PS inactivity timer for EFD in
CELL_FACH expires, the RNC transfers the UE to CELL_PCH/URA_PCH.
If UE is transferred to CELL_PCH/URA_PCH, the PS inactivity timer (PsInactTmrForPreFstDrm) is

started. If the PS inactivity timer for EFD in CELL_PCH expires, the RNC transfers the UE to IDLE.
If UE is in CELL_PCH/URA_PCH and has something to send and the condition for P2D is not satisfied,
the RNC instructs the UE to enter the CELL_FACH state.
2016-02-11
Page44, Total72
Security Level
If UE is in CELL_FACH and send messages whose traffic volume is higher than

FastDormancyF2DHTvmThd, The RNC will transfer the UE from CELL_FACH to CELL_DCH.
When the switch RSVDBIT1_BIT20 of RsvdPara1 in the SET URRCTRLSWITCH command is turned
on and the UE tries to originate or terminate a CS service and the FACH is congested, the RNC
instructs the UE to perform a P2D (instead of P2F) state transition.
When the switch FACH_DTCH_CONGEST_P2D of PROCESSSWITCH2 in the SET

URRCTRLSWITCH command is turned on and the UE tries to originate or terminate a PS service
and the FACH is congested, the RNC instructs the UE to perform a P2D (instead of P2F) state.
For pre fast dormancy UEs not in EFD list:

-
If the RNC received the SCRI, the RNC transfers the UE to IDLE state.
The D2F/F2P/P2F/F2D/P2D state transition for pre fast dormancy UEs not in EFD list is the same to
ordinary UEs.
2016-02-11
Page45, Total72
Security Level
Data configuration for state transitions for a fast dormancy UE:

Typ
e
Description
D2P
D2P indicates a transition from CELL_DCH to CELL_PCH, F2P from

CELL_FACH to CELL_PCH, and D2F from CELL_DCH to CELL_FACH.
The RNC starts the PS inactivity timer for fast dormancy when inactivity on
PS data is detected.
- PsInactTmrForFstDrmDch specifies the PS inactivity timer for
CELL_DCH.
- PsInactTmrForFstDrmFach specifies the PS inactivity timer for
CELL_FACH.
- PsInactTmrForPreFstDrm specifies the PS inactivity timer for CELL_PCH.
D2F
F2P
If the UE is a fast dormancy UE and is not processing CS services, when the

PS inactivity timer
for fast dormancy expires or the RNC receives an SCRI message or 4B event
from the UE, the
UE changes states as follows:
1) For a UE in the CELL_DCH state:
If RSVDBIT1_BIT29 of RsvdPara1 in the SET URRCTRLSWITCH
command is turned off,
the RNC triggers a D2F state transition.
If RSVDBIT1_BIT29 of RsvdPara1 in the SET URRCTRLSWITCH
command is turned on,
the RNC triggers a D2P state transition.
Note:
The D2F and F2P state transitions are recommended for the following reasons:
- During an IOT test, some terminal issues occur. Occasionally, terminals fail to establish services
after a transition from CELL_DCH to CELL_PCH/URA_PCH. This failure has a bad impact on user
experience.
- Compare with a state transition from CELL_DCH to CELL_FACH and then to CELL_PCH/URA_PCH,
the state transition from CELL_DCH to CELL_PCH/URA_PCH does not improve the network
performance or user experience.
- Some terminals send CELL UPDATE message to transfer data immediately after a state transition
from CELL_DCH to CELL_PCH/URA_PCH, which increases the signaling messages and leads to a
heavy load on the Uu interface.
2) For a UE in the CELL_FACH state, the RNC trigger an F2P state

transition.
2016-02-11
Page46, Total72
Security Level
D2I
D2I indicates a transition from CELL_DCH to idle.

When the FACH is congested, D2F is not triggered for inactive CELL_DCH
users. Instead, the
RNC triggers a D2I transition to release the UE's RRC connection. The switch
for D2I is turned off by default. Run the following command to switch it on:
SET UCORRMALGOSWITCH:
ReservedSwitch0=RESERVED_SWITCH_0_BIT16-1;
P2I
P2I indicates a transition from CELL_PCH to idle.

If the UE is in the CELL_PCH state, when the PS inactivity timer for fast
dormancy (specified by
PsInactTmrForPreFstDrm) expires, the RNC releases the UE's RRC
connection and then the
UE enters idle mode.
P2D
P2D indicates a transition from CELL_PCH to CELL_DCH.

When the switch RSVDBIT1_BIT20 of RsvdPara1 in the SET
URRCTRLSWITCH command is
turned on and the UE tries to originate or terminate a CS service and the
FACH is congested, the RNC instructs the UE to perform a P2D (instead of
P2F) state transition.
When the switch FACH_DTCH_CONGEST_P2D of PROCESSSWITCH2 in
the SET
URRCTRLSWITCH command is turned on and the UE tries to originate or
terminate a PS service and the FACH is congested, the RNC instructs the UE
to perform a P2D (instead of P2F) state transition.
P2F
P2F indicates a transition from CELL_PCH to CELL_FACH.

When the UE has something to send and the condition for P2D is not
satisfied, the RNC instructs the UE to enter the CELL_FACH state. For details
about conditions for P2D, see above P2D description.
F2D
F2D indicates a transition from CELL_FACH to CELL_DCH.

When a fast dormancy UE is in the CELL_FACH state, When the fast
dormancy UE sends
messages whose traffic volume is higher than
FastDormancyF2DHTvmThd, The RNC will
transfer the UE from CELL_FACH to CELL_DCH. The
FastDormancyF2DHTvmThd is a new
threshold of event 4a for fast dormancy UE, This threshold is set to bigger
than traffic volume of most heartbeat messages so that to prevent state
transitions from being triggered by a small number of heartbeat messages.
Benchmark for PP12:
2016-02-11
Page47, Total72
Security Level
About 20% reductions on RRC and RAB attempts, which reduce the signaling load significantly
Less congestions are observed as well.
5.2
Multi-RAB Optimization
The Multi-RAB (mainly for CS+PS) services increased as smart phone penetration is increasing.
The trend of Multi-RAB attempts in NGA region:
2016-02-11
Page48, Total72
Security Level
The live network statistics shows that the Call Drop Rate in Multi-RAB scenario is much higher
than single RAB service.
General cause of drop rate of Multi-RAB higher than single CS RAB:
Higher RF signal quality requirement comparing with CS Single RAB
More complicated signaling procedure comparing with Single RAB
To improve the CS call drop:
Reduce the PS RF signal quality requirement for multi-RAB
Reduce the loss probability due to complicated signaling process.
Optimize the signaling procedures
10 parameter tunings were rollout in NPI 2011, some more activities are proposed in 2012
Parameter Change (MML)
Remark
3G_PP12_PPQ_
NGA
Initiative in 2012:
ON
ReservedSwitch0=RESERVED_SWITCH_0_BIT1
Downlink HSDPA to R99 reconfiguration
5-1;
on Multi RAB
Initiative in NPI 2011
ON
SET UFRC:ULDCHBEUPPERLIMITFORAMR=D64,
D64
SET UFRC:DLDCHBEUPPERLIMITFORAMR=D8;
D8
SET URRCTRLSWITCH:
OFF
SET USTATETIMER:RBSETUPRSPTMR=10000;
10000
SET USTATETIMER:HOASUTMR=10000;
10000
SET
10000
SET USTATETIMER:RBRECFGRSPTMR=10000;
10000
SET USTATETIMER:RBRELRSPTMR=10000;
10000
SET USTATETIMER:
10000
ReservedSwitch0=RESERVED_SWITCH_0_BIT1
1-1;
RsvdPara1=RsvdBit1_BIT26-0;
USTATETIMER:HOPHYCHRECFGTMR=10000;
2016-02-11
Page49, Total72
Security Level
HoCellUpdateRspTmr=10000;
SET URRCTRLSWITCH:
Initiative in 2012:
When Cell_Fach&PCH is enabled, avoid
SET URRCTRLSWITCH:
CS RAB failures caused by the
procedure overlap between cell update
OFF
OFF
and CS RAB establishment
Parameter Description:
MO
UCORRMALGOSWITCH
Parameter ID
ReservedSwitch0
Parameter Name
CORRM Algorithm Reserved Switch 0
NE
BSC6900
MML Command
SET UCORRMALGOSWITCH
Meaning
CORRM algorithm reserved switch 0.

RESERVED_SWITCH_0_BIT11: When the switch is turned on, the PS service must be carried by
DCHs in the uplink if both CS and PS services exist.
RESERVED_SWITCH_0_BIT15: When the switch is turned on, the PS service must be carried by
DCHs in the downlink if both CS and PS services exist.
IsKey
NO
Mandatory
NO
Feature ID
NA
Feature Name
NA
Value Type
Bit Field Type
GUI Value Range
RESERVED_SWITCH_0_BIT11, RESERVED_SWITCH_0_BIT15
Enumeration
RESERVED_SWITCH_0_BIT11~10, RESERVED_SWITCH_0_BIT15~14
Number/Bit
Unit
None
Actual Value Range
RESERVED_SWITCH_0_BIT11, RESERVED_SWITCH_0_BIT15
Default Value
None
Recommended Value
RESERVED_SWITCH_0_BIT11:OFF, RESERVED_SWITCH_0_BIT15:OFF
2016-02-11
Page50, Total72
Security Level
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Not involved
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
MO
UFRC
Parameter ID
UlDchBeUpperLimitforAmr
Parameter Name
Upper limit of UL BE data rate in AMR services
NE
BSC6900
MML Command
SET UFRC
Meaning
This parameter specifies the upper limit of uplink BE service rate in combined AMR services.
Combined AMR services consist of only AMR services and BE services. If there is a member of
a different traffic class, the combined services are regarded as combined non-AMR services.
IsKey
NO
Mandatory
NO
Feature ID
None
2016-02-11
Page51, Total72
Security Level
Feature Name
None
Value Type
Enumeration Type
GUI Value Range
D8, D16, D32, D64, D128, D144, D256, D384
Enumeration
D8~1, D16~2, D32~3, D64~4, D128~5, D144~6, D256~7, D384~8
Number/Bit
Unit
kbit/s
Actual Value Range
8, 16, 32, 64, 128, 144, 256, 384
Default Value
D384
Recommended Value
D384
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
MO
UFRC
Parameter ID
DlDchBeUpperLimitforAmr
Parameter Name
Upper limit of DL BE data rate in AMR services
2016-02-11
Page52, Total72
Security Level
NE
BSC6900
MML Command
SET UFRC
Meaning
This parameter specifies the upper limit of downlink BE service rate in combined AMR
services. Combined AMR services consist of only AMR services and BE services. If there is a
member of a different traffic class, the combined services are regarded as combined non-AMR
services.
IsKey
NO
Mandatory
NO
Feature ID
None
Feature Name
None
Value Type
Enumeration Type
GUI Value Range
D8, D16, D32, D64, D128, D144, D256, D384
Enumeration
D8~1, D16~2, D32~3, D64~4, D128~5, D144~6, D256~7, D384~8
Number/Bit
Unit
kbit/s
Actual Value Range
8, 16, 32, 64, 128, 144, 256, 384
Default Value
D384
Recommended Value
D384
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Modification
2016-02-11
Page53, Total72
Security Level
Impact on Radio
None
Network Performance
Attribute
Radio
MO
Parameter ID
Parameter Name
NE
URRCTRLSWITCH
RsvdPara1
Reserved parameter 1
BSC6900
MML Command
SET URRCTRLSWITCH
Meaning
RSVDBIT1_BIT20 (Reserved Parameter 1 Bit 20) When the switch is turned on, for UEs that
support fast dormancy and are in the CELL_PCH state, the RNC performs the CELL_PCH-toCELL_FACH (P2F for short) procedure if the UEs perform cell update with the cause value
"uplink data transmission" or "paging response." When the switch is turned off, for UEs that
support fast dormancy and are in the CELL_PCH state, the RNC performs the P2D procedure if
the UEs perform cell update with the cause value "uplink data transmission" or "paging
response."
RSVDBIT1_BIT21 (Reserved Parameter 1 Bit 21) When the switch is turned on, for UEs that are
establishing AMR services and shifting from the CELL_FACH state to the CELL_DCH state, the
RNC stops establishing AMR services to handle cell update if the RNC receives from the UEs a
cell update message containing the cause value "cell reselection." When the switch is turned
off, for UEs that are establishing AMR services and shifting from the CELL_FACH state to the
CELL_DCH state, if the RNC receives from the UEs a cell update message containing the cause
value "cell reselection," the RNC stops establishing AMR services to handle cell update and
resumes AMR services only after cell update is completed.
RSVDBIT1_BIT26 (Reserved Parameter 1 Bit 26) When the switch is turned on, the RNC does
not reestablish radio links for a UE if the UE reports to the RNC cell update caused by SRB
reset. When the switch is turned off, the RNC reestablishes radio links for a UE if the UE
reports to the RNC cell update caused by SRB reset.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-010101
2016-02-11
Page54, Total72
Security Level
Feature Name
3GPP Specifications
Value Type
Bit Field Type
GUI Value Range
RSVDBIT1_BIT20, RSVDBIT1_BIT21, RSVDBIT1_BIT26
Enumeration
RSVDBIT1_BIT20~19, RSVDBIT1_BIT21~20, RSVDBIT1_BIT26~25
Number/Bit
Unit
None
Actual Value Range
This parameter is set to 0 or 1 according to the related domains.
Default Value
None
Recommended Value
None
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Not involved
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
MO
USTATETIMER
Parameter ID
RbSetupRspTmr
Parameter Name
Wait RB setup response timer
2016-02-11
Page55, Total72
Security Level
NE
BSC6900
MML Command
SET USTATETIMER
Meaning
A timer to RNC wait for the RB setup response from UE in the RB procedure. Refer to the Note.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-010101
Feature Name
3GPP Specifications
Value Type
Interval Type
GUI Value Range
300~300000
Enumeration
Not involved
Number/Bit
Unit
ms
Actual Value Range
300~300000
Default Value
5000
Recommended Value
5000
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No (No impact on the UE in idle mode)
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
The retransmit mechanism is used if the signaling is sent by adopting RLC UM on Uu interface
during channel handover. In this case, the setting of the corresponding waiting response timer
for UE setup does not work. (For example, when the channel change form DCH to FACH or the
channel change from FACH to DCH, the timer does not work.) The retransmit timer is 2
second. The RB reconfiguration message may retransmit three times when the timer expires.
2016-02-11
Page56, Total72
Security Level
Validation of
Not involved
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
MO
USTATETIMER
Parameter ID
HoAsuTmr
Parameter Name
HO active set update response timer
NE
BSC6900
MML Command
SET USTATETIMER
Meaning
A timer to RNC wait for the response to active set update in soft handover procedure.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-020202
WRFD-020203
Feature Name
Intra Node B Softer Handover

Inter RNC Soft Handover
Value Type
Interval Type
GUI Value Range
1~300000
Enumeration
Not involved
Number/Bit
Unit
ms
Actual Value Range
1~300000
Default Value
5000
Recommended Value
5000
Impact
RNC
2016-02-11
Page57, Total72
Security Level
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Not involved
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
MO
USTATETIMER
Parameter ID
HoPhychRecfgTmr
Parameter Name
HO PhyCh reconfiguration timer
NE
BSC6900
MML Command
SET USTATETIMER
Meaning
A timer to RNC wait for the response to physical channel reconfiguration in hard handover
procedure.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-020302
WRFD-020304
WRFD-020301
2016-02-11
Page58, Total72
Security Level
Feature Name
Inter Frequency Hard Handover Based on Coverage

Inter Frequency Hard Handover Based on DL QoS
Intra Frequency Hard Handover
Value Type
Interval Type
GUI Value Range
1~300000
Enumeration
Not involved
Number/Bit
Unit
ms
Actual Value Range
1~300000
Default Value
5000
Recommended Value
5000
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Not involved
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
2016-02-11
Page59, Total72
Security Level
MO
2016-02-11
USTATETIMER
Page60, Total72
Security Level
Parameter ID
RbRecfgRspTmr
Parameter Name
Wait RB reconfiguration response timer
NE
BSC6900
MML Command
SET USTATETIMER
Meaning
A timer used to wait for the RB reconfiguration response from UE in the RB procedure. Refer to
the Note.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-010101
Feature Name
3GPP Specifications
Value Type
Interval Type
GUI Value Range
300~300000
Enumeration
Not involved
Number/Bit
Unit
ms
Actual Value Range
300~300000
Default Value
5000
Recommended Value
5000
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No (No impact on the UE in idle mode)
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
2016-02-11
Page61, Total72
Security Level
for UE reconfiguration does not work. (For example, when the channel change form DCH to
FACH or the channel change from FACH to DCH, the timer does not work.) The retransmit
timer is 2 second. The RB reconfiguration message may retransmit three times when the
timer expires. The parameter modification has no impact on the equipment.
Validation of
Not involved
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
MO
USTATETIMER
Parameter ID
RbRelRspTmr
Parameter Name
Wait RB release response timer
NE
BSC6900
MML Command
SET USTATETIMER
Meaning
A timer to RNC wait for the RB release response from UE in the RB procedure. Refer to the
Note.
IsKey
NO
Mandatory
NO
Feature ID
WRFD-010101
Feature Name
3GPP Specifications
Value Type
Interval Type
GUI Value Range
300~300000
Enumeration
Not involved
Number/Bit
Unit
ms
Actual Value Range
300~300000
Default Value
5000
2016-02-11
Page62, Total72
Security Level
Recommended Value
5000
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
for UE release does not work. (For example, when the channel change form DCH to FACH or
the channel change from FACH to DCH, the timer does not work.) The retransmit timer is 2
second. The RB reconfiguration message may retransmit three times when the timer expires.
Validation of
Not involved
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
MO
USTATETIMER
Parameter ID
HoCellUpdateRspTmr
Parameter Name
HO cell update response timer
NE
BSC6900
MML Command
SET USTATETIMER
Meaning
A timer to RNC wait for the response to the cell update confirm from UE in forward handover
procedure.
IsKey
NO
2016-02-11
Page63, Total72
Security Level
Mandatory
NO
Feature ID
WRFD-010801
WRFD-010802
Feature Name
Intra RNC Cell Update

Inter RNC Cell Update
Value Type
Interval Type
GUI Value Range
1~10000
Enumeration
Not involved
Number/Bit
Unit
ms
Actual Value Range
1~10000
Default Value
5000
Recommended Value
5000
Impact
RNC
Parameter
None
Relationship
Access
Read & Write
Service Interrupted
No
After Modification
Interruption Scope
Not involved
Interruption
Not involved
Duration(Min)
Caution
None
Validation of
Not involved
Modification
Impact on Radio
None
Network Performance
Attribute
Radio
2016-02-11
Page64, Total72
Security Level
5.3
Call Re-establishment
The drop calls mostly are caused by Radio link failure, RLC reset and SRB reset. Utran can
initiate a Cell Update procedure to re-establish the Radio Link in a new cell to save some drops.
Parameter Change (MML
Description
command)
SET URRCTRLSWITCH:
When the switch is turned off, the RNC
RsvdPara1=RSVDBIT1_BIT22-0;
triggers cell update with the cause value
Current Setting
Proposed
on PP12
Setting
ON
OFF
ON
OFF
ON
OFF
ON
OFF
OFF
OFF
"RL Failure" and reestablishes radio links if

the RNC detects interrupted downlink
transmission on SRB2.
SET URRCTRLSWITCH:
triggers cell update with the cause value

"RL Failure" reported by a UE and
reestablishes radio links, if the associated
NodeB reports to the RNC that all radio
links for the UE experience synchronization
loss.
SET URRCTRLSWITCH:
When the switch is turned off, for UEs using
CS services, the RNC triggers cell update

with the cause value "RL Failure" reported
by a UE and reestablishes radio links, if the
RNC detects interrupted downlink
transmission on SRB2, or if the associated
NodeB reports to the RNC that all radio
links for the UE experience synchronization
loss.
SET URRCTRLSWITCH:
When the switch is turned off, for UEs using
PS services only, the RNC triggers cell

update with the cause value "RL Failure"
reported by a UE and reestablishes radio
link, if the RNC detects interrupted
downlink transmission on SRB2, or if the
associated NodeB reports to the RNC that
all radio links for the UE experience
synchronization loss.
SET URRCTRLSWITCH:
reestablishes radio links for a UE if the UE

reports to the RNC cell update caused by
SRB reset.
2016-02-11
Page65, Total72
Security Level
SET
Timer for CS
15000
Timer for PS
30000
URRRSVDPARA:RSVDPARA1=15000
;
SET
URRRSVDPARA:RSVDPARA2=30000
;
Pros:
The subscriber perception has been improved from dropped calls point of view. There are less
dropped calls compared to before KPIs.

Improvement in mobility KPIs makes users have more continuous service.
Cons:
While executing cell update procedure, transmission gaps may occur.

The improvement in user experience is not as much as gained in KPIs, some network problems
may be hidden. The user experience has been improved after the change, but the change in
subscriber perception is not as big as reflected to the KPIs. After the change, there are less
dropped subscribers in the network, but some of the subscribers are experiencing bad quality
disconnections where it is not possible to recover RL by repeated cell updates.
Benchmark:
Significant improvement is seen on CS Call Drop rate and PS Call drop rate has a slight improvement.
2016-02-11
Page66, Total72
Security Level
6 Summary and Conclusion

Totally 19 Jos were implemented, some individual benchmark evaluations in the previous slides
might be some combined operation affect. So to conclude some parameter tunings or features, some
separate trials in other RNCs are required.
Total Improvement
PS CSSR is improved from 99% to 99.5%
CS CSSR is improved from 98% to 98.7%
2016-02-11
Page67, Total72
Security Level
CS CDR is improved from 0.65% to 0.3%.
PS CDR increase is caused by EFD feature (only statistical impact)
Significant improvement on average RTWP
2016-02-11
Page68, Total72
Security Level
Improvement on HSDPA throughput:
Improvement on HSUPA throughput:
Improvement on UL PS384 throughput:

2016-02-11
Page69, Total72
Security Level
Further Actions:
N Category
Item
o
1
Implementa
tion
PS CSSR
Blind Handover
29-May
Optimizatio Fundamental Check

2
Remark
Disable uplink CAC for
11-Jun
high congesting cells
Rollout
Scenario
Slight improvement on PS CSSR on
Rollout in
these cells
entire network
Uplink Power Congestions cleared,
Only for the
and PS CSSR is around 99.5%. Need cells with high

to monitor the
uplink power
"VS.RAC.UL.EqvUserNum" to
congestions
evaluate the impact. CAC threshold

adjustment can be the 1st step.
3
CE resource group
10-Jun
After Uplink resource group
Resource group
reconfiguration for 3 CE
adjustment, congestions moved from configuration
congestion sites
F3 to F1&F2 (even more). The root
audit for all
cause is still that the different uplink sites

resource groups cant be shared.
4
Disable CE admission
13-Jun
CE Congestion reduced
Not
control on
recommended
3G_Mooikloof_Ridge_NG
to rollout
A
5
HSUPA GBR change
4-Jul
PS Congestion reduced
from 64kbps to 32kbps
Not
recommended
to rollout
Enable DRD Failure
27-Jun
PS RAB Setup Success Rate has
Rollout in
slight improvement
entire network
Just parameter standardization for
Rollout in
Optimizatio cells are not
these sites (most of them are new
entire network
sites).
Rollback function
7
PS CDR
2nd SCCPCH audit (24

implemented)
F3 Strategy
Modification
2016-02-11
13-Jun
8-Jun
Improvement on F3 PS CDR after
Rollout in non-
implementation (from 1.9% to 1.3%) continuous F3
Page70, Total72
Security Level
coverage
9
PS
No obvious change on stats as it is
Rollout in
Throughput sites were not enabled
related to the DC-HSDPA capable
entire network
Optimizatio with DC-HSDPA
handset penetration and user
behavior.
10
Fundamental Check: 4
Background noise
7-Jun
11-Jun
adjustment
11
CQI Feedback Period
11-Jun
Slight improvement on HSUPA
Rollout in
throughput on stats.
entire network
About 3dB improvement on average Implement on
optimization from 2ms
RTWP, which means we have almost high RTWP
to 8ms
50% gain on uplink capacity.
caused by high
HSDPA traffic
12
RACH Access Parameter
13-Jun
Optimization
Significant RTWP improvement on
Implement on
High RRC Attempts sites, HSUPA
high RTWP
throughput improved as well.
caused by high
RRC attempts
13
Adaptive Power Offset
18-Jun
for HSUPA
No obvious improvement on
Need a
statistics, as the main gain is
separate trial
obtained when the HSUPA
in another RNC
throughput is greater than 300kbps

as per the feature description. In the
live network, this scenario could
have a small proportion. Field HSUPA
test should be getting more benefit.
14
TCP Accelerator
20-Jun
Slight improvement on throughput
Need a
on statistics.
separate trial
in another RNC
15
CQI Adjustment Based
9-Jul
on Dynamic BLER
Implemented on 9-July, no obvious
Need a
improvement on statistics.
separate trial
Target
16
HSUPA Adaptive Retransmission
in another RNC
4-July & 5- Implemented on 4-July (DBS sites)
July
Need a
and 5-July (BTS sites), no
separate trial
improvement is seen after
in another RNC
implementation.
17
Smart
Multi-RAB Optimization
31-May
CS CSSR is improved from 98.2% to Rollout in
Phone
99.0%, CS CDR is improved from
Optimizatio
0.7% to 0.55% after Multi-RAB
optimization parameter
entire network
standardization.
18
Enhanced Fast
Dormancy
20-Jun
About 20% reduction on RRC and
Rollout in
RAB attempts, which reduce the
entire network
signaling load significantly. Less

2016-02-11
Page71, Total72
Security Level
congestions as well.
19
Call re-establishment
11-Jul
Significant improvement on CS CDR, Rollout in

slight improvement on PS CDR
entire network
7 Schedule
According to the project plan, Huawei will conduct 2*run cluster packet service optimization in 2012.
Meanwhile, the best practices and experiences verified in PP12 will be rollout in entire network.
THE END
2016-02-11
Page72, Total72

3G Cluster Packet Service Optimization Report

Uploaded by

Copyright:

Available Formats

3G Cluster Packet Service Optimization Report

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

3G Cluster Packet Service Optimization Report

Uploaded by

Copyright:

Available Formats

Security Level

3G Cluster Packet Service Optimization

Huawei Proprietary - Restricted Distribution

CQI Adjustment Based on Dynamic BLER Target.....................................................................................................36

HSUPA Adaptive Re-Transmission..............................................................................................................................37

Smart Phone Optimization...................................................................................................................................................38

Enhanced Fast Dormancy.................................................................................................................................................38

Summary and Conclusion....................................................................................................................................................61

Huawei Proprietary - Restricted Distribution

as an example. The Analysis and

PS Call Setup Success Rate Optimization

Huawei Proprietary - Restricted Distribution

The PS CSSR is around 99%.

Huawei Proprietary - Restricted Distribution

Huawei Proprietary - Restricted Distribution

4.1.2 PS CSSR Optimization

Drd Ec/No Threshold

Intra System Direct Retry

Huawei Proprietary - Restricted Distribution

GUI Value Range

Actual Value Range -24~0

Read & Write

The parameter modification has no impact on the equipment.

Versions earlier than BSC6900 V900R011

Huawei Proprietary - Restricted Distribution

Indicates whether to send the system information block 11 (SIB11) including

Intra RNC Cell Update

GUI Value Range

FALSE(Do not send), TRUE(Send)

Actual Value Range FALSE, TRUE

Read & Write

Huawei Proprietary - Restricted Distribution

The parameter modification has no impact on the equipment.

Versions earlier than BSC6900 V900R011

Priority of Coverage-Based Inter-Frequency Handover

Priority for neighboring cell supporting coverage-based inter-frequency

Inter Frequency Hard Handover Based on Coverage

Huawei Proprietary - Restricted Distribution

GUI Value Range

Actual Value Range 0~3

Read & Write

The parameter modification has no impact on the equipment.

Versions earlier than BSC6900 V900R011

2 cells are not configured with Blind Handover Neighbors.

Slight improvement is seen on PS CSSR after implementation on these cells.

b) UL Power Congestion Optimization

Uplink CAC algorithm switch

Huawei Proprietary - Restricted Distribution

GUI Value Range

ALGORITHM_OFF, ALGORITHM_FIRST, ALGORITHM_SECOND, ALGORITHM_THIRD

ALGORITHM_OFF~0, ALGORITHM_FIRST~1, ALGORITHM_SECOND~2,

Actual Value Range ALGORITHM_OFF, ALGORITHM_FIRST, ALGORITHM_SECOND, ALGORITHM_THIRD

Read & Write

The parameter modification has no impact on the equipment.

Versions earlier than BSC6900 V900R011

Huawei Proprietary - Restricted Distribution

Worst performing cells in PP12 with Uplink Power congestion:

Increase admission threshold (impact all types of admission control)

Benchmark for disabling UL power CAC on severe congested cells:

Practices to relieve UL power congestion in SGA region: