This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
HUAWEI TECHNOLOGIES CO., LTD.
www.huawei.com
Huawei Confidential
Security Level:
英文标题 :40-47pt
副标题 :26-30pt
字体颜色 : 反白内部使用字体 :
FrutigerNext LT Medium
外部使用字体 : Arial
中文标题 :35-47pt
字体 : 黑体 副标题 :24-28pt
字体颜色 : 反白字体 : 细黑体
23/4/13
Prepared by: GUL Network I&V and Maintenance
Department
Reviewed by: Qi Haofeng
GSM Speech Quality:Influence Factors + Troubleshooting Methods and
Tools + Deliverables
July 30, 2011
Abstract:
This document mainly discusses the main factors that affect the speech quality
of a GSM network, principles of improving the functions related to speech
quality, and suggested values of some key parameters. In addition, this
document lists the deliverables (see the attachments) that field engineers
should submit when reporting speech quality problems or evaluating the speech
quality, including drive test information, counter information, and guides to
related tools (see the operation guide). This document aims to quickly locate
and solve speech quality problems and to evaluate the speech quality and
prevent speech quality problems based on the collected information about the
For speech quality problems, we can provide trainings and 7x24 hour technical support.
List of R&D support engineers
Name Employee ID Phone
Yang Zhengjie (Wireless
Network)
00127669 See the phone book.
Yang Chunjie 00119951 See the phone book.
Feng Lei (Core Network) 00151560 See the phone book.
英文目录标题 :35-40pt
颜色 : R153 G0 B0
内部使用字体 :
FrutigerNext LT Medium
外部使用字体 : Arial
中文目录标题 :35-40pt
颜色 : R153 G0 B0
字体 : 黑体
英文目录正文 :28-30pt
子目录 (2-5 级 ) :20-30pt
颜色 : 黑色内部使用字体 :
FrutigerNext LT Regular
外部使用字体 : Arial
中文目录正文 :28-30pt
子目录 (2-5 级 ):20-30pt
颜色 : 黑色字体 : 细黑体
Contents
• Evaluation Standards and Principles of Speech
Quality (MOS) • Statistics and Analysis of Factors Affecting the MOS
• Subjective Speech Problem Handling
• Voice-Related Key Parameters:
Quality Parameters
Codec Parameters
Handover Parameters
AoIP Parameters
英文目录标题 :35-40pt
颜色 : R153 G0 B0
内部使用字体 :
FrutigerNext LT Medium
外部使用字体 : Arial
中文目录标题 :35-40pt
颜色 : R153 G0 B0
字体 : 黑体
英文目录正文 :28-30pt
子目录 (2-5 级 ) :20-30pt
颜色 : 黑色内部使用字体 :
FrutigerNext LT Regular
外部使用字体 : Arial
中文目录正文 :28-30pt
子目录 (2-5 级 ):20-30pt
颜色 : 黑色字体 : 细黑体
Page 5
Evaluation Standards and Principles of Speech Quality (MOS)
Subjective EvaluationThis method indicates that many people compare the original voice sample with the degraded file
processed by the system by their own subjective perceptions, then mark the mean opinion score
(MOS) (ITU-T 800) value (full score: five points), and finally obtain the average value.
Objective EvaluationThis method indicates that the score is obtained through comparing the degraded voice file after
transmission with the original voice sample file by using a certain algorithm, such as PAMS (ITU-T
P 861) and PESQ (ITU-T P 862.1).
Parameter EvaluationThis method indicates that the voice after transmission is not evaluated, and the original voice is
not obtained. Instead, the voice after transmission is evaluated through some parameters of
wireless transmission network, which has a promising prospect in wireless network, such as
RXQUAL, VQI of Huawei, and SQI of E///.
Currently, carriers all over the world treat the speech quality as the key indicator for network
acceptance. Among them, PESQ algorithm is the widely-used scoring standard. In this algorithm,
the PESQ calculation result is mapped into the MOS value, ranging from 1.0 to 4.5.
英文目录标题 :35-40pt
颜色 : R153 G0 B0
内部使用字体 :
FrutigerNext LT Medium
外部使用字体 : Arial
中文目录标题 :35-40pt
颜色 : R153 G0 B0
字体 : 黑体
英文目录正文 :28-30pt
子目录 (2-5 级 ) :20-30pt
颜色 : 黑色内部使用字体 :
FrutigerNext LT Regular
外部使用字体 : Arial
中文目录正文 :28-30pt
子目录 (2-5 级 ):20-30pt
颜色 : 黑色字体 : 细黑体
Page 6
Evaluation Standards and Principles of Speech Quality (MOS)
PESQ: Perceptual evaluation of speech quality
This method is used for E2E network speech quality test. It is to compare the
original voice sample on the transmitting end in the network (narrowband) with
the distorted degraded voice file received on the receiving end, evaluate the
difference between the two signals through complex signal processing, and
finally obtain the speech quality value using the PESQ algorithm.
After the PESQ algorithm is processed, the following four metrics are obtained:
• PESQ RAW SCORE (the raw score)
• P.862.1 (the score is obtained through the P.862.1 mapping mode based on the raw score)• PESQ-LQ (the score is obtained through the Psytechnics mapping mode) • PESQ-Ie (The score is obtained through the mutilation factor of instrumental models defined by
P.834)
Among them, the value of P.862.1 is widely regarded as the reference value in voice
Statistics and Analysis of Factors Affecting the MOS
Voice qualityVoice quality
CodeCode Bit error (frame erasure)
Bit error (frame erasure) HOHODirectly-related
factors
Directly-related factors
Indirectly-related factors
Indirectly-related factors
Traffic
Traffic
Full/half
rate
Full/half
rate
High/low
coding rate
High/low
coding rate
Air Interface Quality
Air Interface Quality
Traffic Busy Threshold
Traffic Busy Threshold
Rate adjust Threshold
Rate adjust Threshold
F2H HO Threshold
F2H HO Threshold
Threshold Self-Adaptive
Threshold Self-Adaptive
Speech
version
Speech
version
Frequent HOs, PingPong HOs,
and unreasonable HOs
Frequent HOs, PingPong HOs,
and unreasonable HOs
Too low PN Rule
Too low PN Rule
Inappropriate
Neighboring Cell
Inappropriate
Neighboring Cell
Too Small HO Hysteresis
Too Small HO Hysteresis
Frame Theft of Physical
Messages
Frame Theft of Physical
Messages
Parameters, algorithms, and
optimizing strategies
Parameters, algorithms, and
optimizing strategies
The channel is normal.The channel is normal.
Interference
Interference
Algorithms
Algorithms
Engineering
network
optimization
Engineering
network
optimization
3.5-Generation Power
Control
3.5-Generation Power
Control
DTX
DTX
VAD
VAD
Anti-Interference solution
Anti-Interference solution
Intermodulation
Interference Quick
Troubleshooter
Intermodulation
Interference Quick
Troubleshooter
TOP Optimization
TOP Optimization
HO Optimization Packet
HO Optimization Packet
Call drop
Call drop
Long Call Drop Timer
Long Call Drop Timer
CoBCCH Resident
Strategy
CoBCCH Resident
Strategy
Discarded Packets
Compensation
Discarded Packets
Compensation
The speech quality is mainly related to three factors: code, bit error, and handover (HO). The coding factor benefits the speech quality. The bit error and handover factors, however, damage the speech quality.
To optimize the speech quality, you need to select reasonable codes and reduce the effect of the bit error rate (BER) and handovers on the speech quality.
The prerequisite is that the channel is normal.
The methods for improving the call drop rate and handover success rate usually damage the speech quality and the experience of subscribers. Therefore, strategies that optimize the speech quality may affect the call drop rate and handover success rate.
1. Problem specifying Specify the details about the MOS problem, including the current test value, target value, and the gap
between the two. Specify the standard of the MOS appraisal. Specify whether it is MS-MS, or MS-PSTN, whether the MOS
appraisal value is the overall average score or percentage of high scores, and whether the up-link and down-
link are appraised separately. Contrast the test MOS values by using instruments or terminals, it is found that there is no change. Contrast the test time frame (start from what time point and to what time point the test ends), test route, and
test period (to ensure the comparability of MOS tests, the tests shall be performed on the same day in
different weeks).
2. Speech version and coding proportion analysis There is a large difference for the MOS in different speech versions. In normal conditions, the sequence for
the MOS baseline performance is: FAMR > EFR > HAMR > FR > HR. For example, the MOS for the EFR in
the MS-MS test can be 4.0, whereas that for the HR will be 3.0. Therefore, for MOS problems incurred before
and after migration, contrast the occupation proportion of each speech version in the drive test, and check
whether the proportion of half rate is increased.
Statistics and Analysis of Factors Affecting the MOS
3. Handover times and proportion analysis If the MOS problem occurs before or after migration (upgrade), check whether the ratio between the number
of MOS dotting and that of handover times is changed. The larger this ratio is, the smaller the effect made by
the handover on the overall MOS is. Analyze whether ping-pong handover exists in the drive test data, or whether the handover times in counters
are too much (usually, the number of handover times in each call is within 1 in existing networks). If such
case exists, modify the corresponding parameter configuration to reduce the effect of handover on the MOS.
If the PN for the PBGT (better cell) handover is added, and the PBGTSTAT (s) parameter is set to 5s and
the PBGTLAST (s) parameter is set to 4s, the judgment time for handover is delayed, and the handover is
reduced.
4. Comparative analysis of data configuration and transmission If the MOS problem occurs before or after migration (upgrade), check whether the data configuration and
transmission mode are changed, for example, whether parameters including cell handover and power control
are changed, and whether cells or frequency band is added or reduced. In addition, trouble-shoot the radio
frequency channels in the area where problems occur, and check whether KPIs in the traffic transmission are
incorrect, which affect the MOS test result. For new-built network, check whether alarm information is displayed on the NEs such as the BTS, BSC, and
transmission in the test, which affects the test result. Check whether the network KPIs are abnormal. Usually,
KPIs include TCH call drop rate, success rate of wireless handover, and TCH congestion rate.
Statistics and Analysis of Factors Affecting the MOS
Voice services are key services in a GSM network. The quality of voice services is determined by many factors. To report MOS problems or speech problems, you need to report all factors related to voice during the drive test.
For details about the distributed troubleshooting, see the Guide to Locating and Isolating GSM Speech Problems.
Statistics and Analysis of Factors Affecting the MOS
1. One-way audio One party of the two call parties cannot hear the voice from the peer end, or both the two parties cannot hear the voice on the
peer end, which is presented as one-way audio or no audio. When one-way audio occurs, mute also occurs.
Handling process: Specify the area scope where the problem occurs and the specific situations. Confirm it is uplink one-way audio (the party
who holds the MS cannot hear any voice, but the one who is on the PSTN side can hear the voice) or downlink one-way
audio (the party holds the MS can hear voices, but the one who is on the PSTN side cannot). Enable the one-way audio detection function (confirm whether the current version supports it or not first), analyze the one-
way detection logs of the whole day, and find out the suspicious resources for dialing test. Perform the dialing test on the
site where the problem occurs. For detailed dialing test procedures, see the attachment Operation Guide for Speech
Tests. During the dialing test, perform TC recording and single user tracking. Perform loopback when the problem
reoccurs, and confirm the NE where the problem occurs. Analyze the trunk performance measurement of the A interface, and fond out abnormal occupation timeslots (Rules: The
A interface has 31 timeslots in total, while the average busy hour of the 31 timeslots is less than 30s, and the number of
timeslots whose average busy hour is less than 30s is at least 28. However, networks charged by second are excluded,
which needs special treatment). Combined with specified CIC dialing test, hardware connection of interfaces, and data
configuration, check whether there are problems such as crossed pair on the A interface or incorrect connection of lines
(In TDM transmission mode, if the E1 line on the A interface is not configured with the SS7 signaling link, or the E1 line on
the Abis interface is not configured with the RSL and OML links, the E1 line on the corresponding port is incorrectly
connected, or no alarm is generated even if crossed pair are made (as long as it is not suspended). However, when users
occupy this port, one-way audio or no audio occurs).
2. NoiseDuring the call, abnormal voices such as bubbles, clicks, and metallic sounds occur. When it is at its worst, only
noise can be heard and the normal speech cannot be heard completely. Usually, noises can be divided into two
types: noise in normal conversation and handover noise. However, noises are mainly caused by bit errors,
including
bit errors caused by frequency interference, voice processing software, and equipment hardware.
Handling process: Specify the area scope where the problem occurs and the specific situations. Select the site where the problem occurs for dialing test. For detailed dialing test procedures, see the
attachment Operation Guide for Speech Tests. During the dialing test, perform TC recording and single user
tracking. Perform loopback when the problem reoccurs, and confirm the NE where the problem occurs. Check the alarm and transmission connection line of the site where the problem occurs to see whether there
is any looseness or damage. Check transmission indexes in the traffic statistics, for example, whether
problems including packet loss, jitter, and too-long delay exist.
3. Speech make-and-breakSpeech make-and-break mainly presents like this: there is a sense of pause in the call, and listeners
may miss half a word or several words. When the make-and-break is obvious, it may affect the normal
conversation.
For the troubleshooting procedures, see the noise handling process.
4. EchoEchoes are mainly divided into two categories: acoustics echoes and electrical echoes. Echoes caused by MS
calling MS are called acoustics echoes, whereas echoes caused by MS calling PSTN are called electrical
echoes.
Handling process: Check whether the handsfree function is enabled or the headset mode is used. Then, check whether the echo
is disappeared or lowered after the handsfree function is disabled or the volume is lowered. Acoustics echoes are usually caused by the noncompliance of isolation of terminals to the protocol
requirements. During the test, adjust the volume of the MS on the peer end. If the echo volume heard on the
local end is obviously changed, it indicates that the echo is produced by the MS on the peer end. You can
change another MS for re-test.
Usually, acoustics echoes are strongly relevant to MSs. The solution to acoustics echoes: Enable the AEC
function on the BSC side to help MSs to further eliminate echoes. Electrical echoes usually caused by configuration or engineering problems. For example, the call routing data
configuration is incorrect, hybrid coils on the fixed network side do not meet the relevant telecom standards,
and the produced echo volume exceeds the processing capability of the echo canceler.
5. CrosstalkIn the call, not only the speech of the called party can be heard, but also a third-party speech can be heard, or the
speech of the called party cannot be heard, instead, a third-party speech is heard. The common reasons for
crosstalk are Um interface crosstalk, core network crosstalk, incorrect data configuration, and abnormal
connection.
Handling process: Encrypt the Um interface: Enabling the Um interface encryption is the root solution to Um interface crosstalk. In BSS data configuration, the configured value of the T3109 timer must be larger than the value configured
in RLT. When the MSC equipment is not Huawei equipment, enable the Call Re-establishment switch on the BSC
side, and set the call re-establishment timer to 45s. Record information such as routing, equipment resources, and transmission about each crosstalk and
analyze them one by one. If it is found that all crosstalks occur in long distance or cross-network (a China
Mobile subscriber calls a China Unicom subscriber) calls, basically it can be concluded that the crosstalk has
something to do with the core network, and the core network engineers need to participate in the fault
location. According to the customers' complaint information, draw the CDR from the MSC and find out the
corresponding CIC to perform the designated dialing test to check whether the CIC timeslot appears
regularly. If it is regularly appears, trouble shoot the hardware connection or data configuration. Meanwhile,
check the data configuration and E1 connection of the problem points.
Subjective Speech Problem HandlingNo. Feedback on Speech Quality Problems Output Description
1Detailed descriptions of the problem, including the scenario in which the problem occurs and the probability that the problem occurs
For example, record whether noises periodically occur (namely whether the noise occurs once every x second (s) or every x minute (s)), and whether noises persist during the calls.
2Tracing signaling of a single user and descriptions of the calling and called MSs
For example, the information about the TEMS of the calling MS is as follows: The MSISDN is 13913140397 and the IMSI is 460512300000397.
3 Log data about the TEMS test Record the test log data when the problem occurs.
4TC recording files and Um interface frame data captured by the Probe
The files and data need to be configured before the test. Probe is a test tool of Huawei and is used to capture the information about the Um interface frames. The TC recording files are in the format of *.dat.
5 Loopback test results about the problemList of loopback test results of interfaces (see the following table))
6BSS version information and information about core network vendors, configuration data and information about test sites
For BSC6900V900R011C00SP720, it is a CME configuration file and a MML configuration file in the format of *.txt. For versions earlier than BSC6900V900R011C00SP720, it is a *.dat file. For the BTS3012, you need to specify whether the new DTRUs or old DTRUs are used. For the BTS3900, you need to specify the types of TRXs.
7Information about the engineering parameter table within the test area
The information is in the format of *.cel that is supported by the Nastar and TEMS.
8 Transmission mode on the entire networkThe transmission mode is all-TDM, all-IP, or hybrid.
9BTS log, alarm log, one-way audio log, and related alarm information about the problem site
Logs and alarm information
10 DSP, DEBUGg, GCSR, and CHR logs of the BSC Log files at the test segments
Typical Case 2: A leader of a customer of a office complains that one-way audio exists in a certain probability
during the call, and asks Huawei to solve the problem as soon as possible.
A. Perform the dialing test at the problem site, and perform speech loopback when the problem reoccurs. After the calling party A (external network) has a conversation with the called party B (under the problem site), A cannot hear B, which is
uplink one-way audio. Enable the remote loopback of the A interface on the B side, A can hear his/her own voice, indicating that the problem does not exist on
the MSC side or on routing nodes after the A interface. Enable the local loopback of the A interface on the B side, B cannot hear his/her
own voice, indicating that the problem exists on routing nodes before the A interface on the B side. Enable the BTS speech loopback, B can hear his/her own voice, indicating that the problem exists between the Abis interface and the A
interface, namely the BSC.
B. Analyze the TC recording file, and find that when the one-way
audio occurs, the call works properly when the uplink voice data
enters into the TC. However, when the uplink voice data goes
out of the TC, no voice data is available.
Upon analysis, it is concluded that the BTS sends abnormal
frames, which leads to the TC scheduling
memory error, leading to one-way audio. This problem is
solved after the code optimization.
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential
Typical Case 2: A user of an overseas office complains that one-way audio and no audio problems occur with a high reoccurrence probability.
Location process:
Because the Abis interface of this office adopts the HDLC transmission, and does not support the one-way audio detection function, the one-way
audio detection cannot be enabled. Upon the analysis on the trunk performance measurement of the A interface, the average occupied duration of
the three ports under the three BSCs is less than 30s, which is far below the average occupied duration of all BSCs. The distribution of occupied
duration has obvious time intervals. In addition, there are more than 28 timeslots that the average occupied duration is less than 30s for each port.
Perform CIC dialing test for specified A interface, the one-way audio occurs. Check the transmission, and find that the E1 lines of the A interface on
the three ports are incorrectly connected. After the transmission is adjusted, the one-way audio does not reoccur.
Quality Parameters Parameters related to improving the speech quality at a low CIR Parameters related to user experience Parameters related to power control Other quality-related parameters
Coding Parameters Speech versions Parameters related to VQE Parameters related to channel allocation
Handover Parameters Handover-related parameters
AoIP Parameters Mapping versions related to AoIP AoIP-related parameters
Voice-Related Key Parameters – Summary of Quality Parameters
For details about the mapping versions that support voice-related features, see the Reference List of Core Parameters Related to Speech Quality in the attachments.
Description of the Adaptive Level Control (ALC) Function:This function controls the level automatically. It evaluates the voice level of the input signals and
controls
the gain of the input signals. Specifically, it adjusts the output voice signals to the target level and
ensures that the level of the signals is stable and the signals can be understood. Therefore, the hearer
thinks that the volume is proper and has a good experience to the voice.
Suggested Parameter Settings:
Description of the Adaptive Noise Reduction (ANR) Function:This function is mainly used to reduce the background noise in the voice without damaging the voice. In
this way, it makes the voice acceptable to the hearer.
Suggested Parameter Settings:
Quality Parameters: User Experience-Related Parameters
Description of the Counter Function: Radio Link Timeout: This counter defines the time of radio link connection failure for downlink links. The criterion is that whether the
SACCH measurement report can be correctly decoded.
SACCH Multi-Frames: This counter defines the time of radio link connection failure for uplink links. The criterion is that whether the
SACCH measurement report can be correctly decoded.
Suggested Parameter Settings:
Note: The suggested parameter values are to end the call in the case that the UM interface quality is bad,
avoiding continuous impact of continuous bad quality on the speech MOS values. These suggested values
may affect the call drop rate. Therefore, you are advised to use them only when you handle speech
problems. For other KPI handling, see the parameter baselines.
TC CRC Check: According to GSM specifications, the BSC performs CRC check for each uplink data (TRAU frame) from the BTS. If the TRAU frame fails
to pass the CRC check, the BSC regards it as an invalid frame and smoothens it. This avoids the noise caused by parameter
transmission errors and improves the speech quality.
Quality Parameters Parameters related to improving the speech quality at a low CIR Parameters related to user experience Parameters related to power control Other quality-related parameters
Codec Parameters Speech versions Parameters related to VQE Parameters related to channel allocation
Handover Parameters Handover-related parameters
AoIP Parameters Mapping versions related to AoIP AoIP-related parameters
Voice-Related Key Parameters – Summary of Codec Parameters
Codec Parameters Parameter Name Commend Value Parameter Description
Speech versions FAMR/EFR/HAMR/FR/HR AMRAdjusts the speech coding mode on the uplink and downlink according to changes in Um interface quality, thereby improving the speech quality.
Enhancement of speech quality
TFO Switch ON Reduces the impact of speech transcoding on the speech quality, improving the speech quality.
RTPSWITCHOFF
Specifies whether to enable the delay function to be implemented between the BTS where the calling MS is located and the BTS where the called MS is located.
TFOOptSwitch OFF When the speech versions on the two sides are inconsistent, establish the TFO after optimization.
RATSCCHENABLED OFFSpecifies whether to enable the RATSCCH procedure during a call setup. In the RATSCCH procedure, the rate set of AMR calls can be dynamically adjusted during a call to improve speech quality.
EPLC Switch OFF Compensates the packets that are lost during the transmission.
AMR Uplink Adaptive threshold allowed YESReduce the impact of inaccurate estimation of Signal-to-Noise Ratio (SNR) or the changes in channel conditions following the time on the Adaptive Multi Rate (AMR).
Voice Quality report switch YES Uses the voice quality index (VQI) to monitor the speech quality on the network in real time.
TrFO Switch YES Reduces the impact of TC coding and encoding on the speech quality, improving the speech quality.
Speech Channel Alarm Threshold 10Specifies the threshold for reporting the speech channel alarm. If the number of one-way audio that occurs in an hour on the BSC exceeds this threshold, the speech channel alarm is reported.
Speech Channel Resume Alarm Threshold 6Specifies the threshold for reporting the speech channel resume alarm. If the number of one-way audio that occurs in an hour on the BSC is smaller than this threshold, the speech channel resume alarm is reported.
TCMUTEDETECTFLAG ON Specifies whether to enable the class-1 one-way audio detection function.
MUTECHECKCLASS1PERIOD 5Specifies the class-1 one-way audio detection period. If the FER within the period specified by this parameter exceeds the value of Exceptional Frame Threshold(%), you can infer that one-way audio occurs.
EXCEPFRAMETHRES 25Specifies the threshold for the proportion of the number of bad frames to the total number of TRAU frames. If the FER exceeds this threshold within the value of Period of Mute Detect Class1(s), one-way audio may occur.
MUTECHECKCLASS2SWITCH ONSpecifies whether to enable the one-way audio and no audio detection function to improve the accuracy of one-way report.
DETECTFRAMEPERIOD 2Specifies the period for sending the TRAU test frame after the class-2 one-way audio detection function is enabled. One TRAU test frame is sent in each period until the response from the peer end is received or the timer expires.
MUTECHECKPEIROD 4 Specifies the class-2 one-way audio detection period.
Channel allocation
CHALLOCSTRATEGY CAPABILITY Allocates the channel with good quality, improving the speech quality.
TCHBUSYTHRES 60 Improves properly the proportion of full rate occupation, improving the speech quality.
TCHTRICBUSYOVERLAYTHR 70 Improves properly the proportion of full rate occupation, improving the speech quality.
TCHTRIBUSYUNDERLAYTHR 60 Improves properly the proportion of full rate occupation, improving the speech quality.
Description of the Tandem Free Operation (TFO) Function:
During an MS-MS call, this function reduces one voice coding/decoding session.
This reduces the effect of voice coding on the speech quality and improves the
speech quality.
Suggested Parameter Settings: TFO Switch: Enable Measure Link Delay Switch: Disable Support TFO Codec Optimize: No Is RATSCCH Function Enabled: Disable
Codec Parameters: Parameters Related to VQE
Note:•The GBSS 8.1 and GBSS 9.0 use different TFO mechanisms. You are advised to use a BSC and BTSs of correct mapping versions when the TFO function is enabled. Incorrect version mapping leads to a decrease in TFO gains. •When the TFO function is enabled on the BSC6900 V900R011C00SPC750 or earlier versions, the proportion of half rate channels increases and the proportion of the MOS values greater than 3.0 may decrease. •When the TFO of a Huawei BSC interconnects with that of a BSC from another vendor, ensure that the onsite BSC version is BSC6900 V900R011C00SPC756 or a later version.
Description of the Enhance Packet Loss Concealment (EPLC) Function: If packet loss occurs during the transmission of voice packets in the network, the speech quality may drop
sharply. This function ensures that the sound effect after the decoding is similar to that before the decoding by
hiding and reducing the effect of packet loss on the speech quality when decoding the received voice packets.
Suggested Parameter Settings: EPLC Switch: Off
Description of the AMR Adaptive Threshold Function: This function reduces the effect of incorrect estimation of the C/I or changes of the channel conditions with
time on the AMR performance. You do not need to estimate the value of this parameter. The value of this
parameter at various rates are adjusted automatically according to the fluctuation of the C/I. In this way, the
AMR performance will not be affected by incorrect estimation of the C/I or by changes of the channel
conditions with time.
Suggested Parameter Settings: AMR Uplink Adaptive Threshold Allowed: Yes
AMR Downlink Adaptive Threshold Allowed: No
Codec Parameters: Parameters Related to Enhancement of Speech Quality
Description of the Voice Quality Index (VQI) Function This algorithm evaluates the speech quality through parameters that calculate the radio transmission quality
over the Um interface. The value of VQI is calculated after the Um interface link parameters are measured.
The value of VQI can correctly reflect the end-to-end quality of voice calls. Therefore, the network operator
can monitor the speech quality of the network in real time without knowing the experience of end users.
Description of the Transcoder Free Operation (TrFO) Function: This function prevents the TC from processing voice signals when a voice or multimedia call is established in
the A over IP scenario. As the TC does not exist in physical links for transmitting voice signals when the TrFO
function is enabled, this function reduces the damages to the speech quality by the TC and improves the
speech quality.
Suggested Parameter Settings: This function is defined in 3GPP R4 and needs the support of core network equipment.
Codec Parameters: Parameters Related to Channel Allocation
Parameters Related to Channel Allocation: Channel Allocate Strategy Traffic Busy Threshold Tch Traffic Busy Overlay/Underlay Threshold
Basic Principles: The channel allocation strategy helps to allocate high-quality channels and to improve the speech
quality. A full-rate channel is better than a half-rate channel in improving the speech quality. Therefore,
increasing the ratio of full-rate channels helps to improve the overall MOS. Note: When allocating channels, be sure to consider whether congestion exists in the cell. If no
congestion exists, you can set the traffic busy threshold to a larger value.
Quality Parameters Parameters related to improving the speech quality at a low CIR Parameters related to user experience Parameters related to power control Other quality-related parameters
Codec Parameters Speech versions Parameters related to VQE Parameters related to channel allocation
Handover Parameters Handover-related parameters
AoIP Parameters Mapping versions related to AoIP AoIP-related parameters
For AMR calls, it is allowed to perform AMR full/half rate handover according to the C/I value
of the current call. When the C/I value is great, you can switch full rate to half rate to increase
the traffic capacity. When the C/I value is small, you can switch half rate to full rate to improve
the experience of subscribers.
Suggested Parameter Settings: Intracell F-H HO Allowed: Yes TCH F-H Threshold: 16 Intracell F-H HO Stat. Time (s): 5 Intracell F-H HO Last Time (s): 4
Description of the CoBCCH Access Strategy Function:
To reduce unnecessary handovers between the UL and OL in a concentric cell and to avoid
the damage to voice caused by unnecessary handovers, you need to set the Assign Optimum
Layer and Pref. Subcell in HO of Intra-BSC parameters properly.
Suggested Parameter Settings: Assign Optimum Layer: System optimization Subcell in HO of Intra-BSC : System optimization Incoming-to-BSC HO Optimum Layer: Underlaid subcell
Parameters related to improving the speech quality at a low CIR Parameters related to user experience Parameters related to power control Other quality-related parameters
Codec Parameters Speech versions Parameters related to VQE Parameters related to channel allocation
Handover Parameters Handover-related parameters
AoIP Parameters Mapping versions related to AoIP AoIP-related parameters
Setting of the Huawei core network (for reference)
(1) The A over IP interface in BSC6900V900R011 is
specified in a standard 3GPP protocol. In the
MSC, configure BSC Bearer Type as IPSTD (IP
Type of standard).
AoIP-Related Parameters on the Core Network
(2) On the core network, configure HR AMR and FR AMR codec set as S1. The network-supported HR AMR list includes 12.20 kbit/s, 7.40 kbit/s, 5.90 kbit/s, and 4.75 kbit/s; and the network-supported FR AMR list includes 7.40 kbit/s, 5.90 kbit/s, and 4.75 kbit/s.(3) Set bit 8 in soft parameter P13 to 0 to enable the TrFO compatible with AMR FR/HR. Bit 8 in soft parameter P13 controls the TrFO compatible with AMR FR/HR (UMG). 0: Enables TrFO compatible with AMR FR/HR.1: Disables TrFO compatible with AMR FR/HR. This is the default.(4) Run the following command to set bit 9 in soft parameter P13 to 0:MOD SFP: ID=P13, MODTYPE=BIT, BIT=9, BITVAL=0; Bit 8 in soft parameter P13 controls the optimization for TrFO compatible with AMR FR/HR (UMG).(5) Run the following command to set related parameters: SET UPPARA: RC2CMR=OPEN, CMR2RC=OPEN; For detailed parameter configurations, see the Version Policies and Configuration Requirements for IP-Based GSM V1.26 in the attachments on page 72.
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential
Troubleshooting Procedure for Speech Problems in A over IP
The procedure for troubleshooting speech problems in A over IP
transmission mode is the same as that for other transmission modes
except for the speech version and parameters for the A over IP. The
suggested troubleshooting procedure is as follows:
(1) Check that the version mapping of BSC, BTS, and NEs on the core
network meets the version requirements.
(2) Check that the AoIP-related parameters are correctly configured.
(3) Check that the IP transmission quality meets the QoS.
(4) Based on the preceding check, troubleshoot the problem in the aspect