110857247 3 GSM Speech Quality Influence Factors Troubleshooting Methods and Tools Deliverables 20110730

Security Level:

2013/2/18

GSM Speech Quality: Influence Factors + Troubleshooting Methods and Tools + Deliverables

Prepared by: GUL Network I&V and Maintenance Department Reviewed by: Qi Haofeng

July 30, 2011

HUAWEI TECHNOLOGIES CO., LTD.

Huawei Confidential

www.huawei.com

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 existing network.

R&D Support 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

00127669

See the phone book.

Yang Chunjie

00119951

See the phone book.

Feng Lei (Core Network)

00151560

See the phone book.

Network)


Huawei Confidential

Page 3

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

Evaluation Standards and Principles of Speech Quality (MOS) Subjective Evaluation This 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 Evaluation This 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 Evaluation This 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.

Page 5

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 evaluation.

Page 6

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


Huawei Confidential

Page 7

Statistics and Analysis of Factors Affecting the MOS Voice Voicequality quality

Directly-related Directly-related factors factors

drop Calldrop Call

Engineering Engineering network network optimization optimization

Algorithms Algorithms

Interference Interference

Speech Speech version version

High/low High/low rate codingrate coding Full/half Full/half rate rate

IndirectlyIndirectlyrelated relatedfactors factors

DD VV TT AA XX DD

HO HO

Frequent FrequentHOs, HOs, PingPong PingPongHOs, HOs, and andunreasonable unreasonable HOs HOs

CoBCCH Resident CoBCCH Resident Strategy Strategy HO Optimization Packet HO Optimization Packet Frame Theft of Physical Frame Theft of Physical Messages Messages Too Small HO Hysteresis Too Small HO Hysteresis Inappropriate Inappropriate Neighboring Cell Neighboring Cell Too low PN Rule Too low PN Rule Discarded Packets Discarded Packets Compensation Compensation Long Call Drop Timer Long Call Drop Timer TOP Optimization TOP Optimization

3.5-Generation Power 3.5-Generation Power Control Control

Anti-Interference solution Anti-Interference solution Intermodulation Intermodulation Interference Quick Interference Quick Troubleshooter Troubleshooter

Threshold Self-Adaptive Threshold Self-Adaptive

Rate adjust Threshold Rate adjust Threshold Air Interface Quality Air Interface Quality Traffic Traffic Traffic Busy Threshold Traffic Busy Threshold

F2H HO Threshold F2H HO Threshold

Parameters, Parameters, algorithms, algorithms,and and optimizing optimizing strategies strategies

Bit Biterror error(frame (frame erasure) erasure)

Code Code

The Thechannel channelisisnormal. normal.


Huawei Confidential

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.

Page 8

Statistics and Analysis of Factors Affecting the MOS The MOS handling process is as follows: In existing networks, two methods

MOS fails to reach the standard.

are available for the MOS acceptance standard: One is that the average

001 Make clear the actual situation of the problem (including average MOS or deterioration conditions of the percentage of high scores, testing methods, and instruments).

MOS for the drive test of the entire network shall reach a value. The other is that the proportion of high scores in the MOS shall be larger than the required value, or the proportion of low scores shall be

002

003

004

005

006

Trouble-shoot the occupation of voice version and encoding rate.

Trouble-shoot the percentage of the number of handover times and that of MOS dotting.

Trouble-shoot the data configuration and transmission quality.

Trouble-shoot the TC recording and air interface frame data.

Trouble-shoot the air interface quality.

lower than a certain value, and

During the processing, perform the troubleshooting from the easier to the more advanced following the dashed in red.

comparison between the two drive test data (such as migration and 007

version upgrade) shall be performed.

Analyze the preceding factors and specify the reasons that lead to the MOS problems.

No matter which method is used, when the MOS does not reach the standard, troubleshooting is carried out based on factors affecting the

008 After the preceding troubleshooting, optimization & adjustment, and verification, if the MOS problems are still not solved, perform the escalation processing.

001 No

After the optimization and adjustment, verify whether the problem is solved through drive test.

Yes

MOS in the process. End


Huawei Confidential

Page 9

Statistics and Analysis of Factors Affecting the MOS 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 downlink 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.


Huawei Confidential

Page 10

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.


Huawei Confidential

Page 11

Statistics and Analysis of Factors Affecting the MOS 5. Analysis on TC recording and Um interface frame capture Mainly check whether there is any problem on the TC and Probe terminals and the area where the problem occurs. Data

TC recording file

Probe file

Wave Mode Comparison Files

Whether Difference Exists in Wave Mode Comparision

MOS Compared with the Original Sample

UpIn vs. UpOut DnIn vs. DnOut Original sample vs. MS uplink voice data MS downlink voice data vs. Recording (degraded) file

6. Um interface quality analysis For comparison before and after migration (upgrade), analyze based on the quality before and after migration (upgrade), and check whether quality deterioration exists in all areas or part of areas. If quality deterioration exists in some areas (including cases that call drop or handover failure occurs due to poor quality of Um interfaces), perform analysis based on the drive test data, and make clear whether the poor quality is caused by cases such as interference, poor coverage, and missing cross and neighboring cells.


Huawei Confidential

Page 12

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.


Huawei Confidential

Page 13

Feedback Information About Counter Data Related to Voice Problems Good speech quality first depends on good network quality. The quality of the entire network, however, should be evaluated by counters. For details, see the following table.

For more detailed feedback information, see the attachment "Checklist for Data Provided for Speech Quality Problems".


Huawei Confidential

Page 14

Typical Case 1: In a site, it is required that the average value of MS-MS MOS should be greater than 3.50 according to the drive test result of the entire network after the migration. The value, however, is only 3.35 according to the drive test result. Therefore, the value of MOS fails to reach the standard. According to the statistics and analysis of the drive test information, the ratio of half rate channels reaches 75%. This is the main factor that affects the overall MOS. The details are as follows:

Through parameter configuration and counter analysis, it is discovered that the problem is caused by that the values of TCH Traffic Busy Threshold of many cells are set to be too small (30%). After adjusting the values of this parameter and performing another drive test, the ratio of occupied half rate channels is reduced to 47% and the value of MOS reaches 3.52, which exceeds the acceptance standard.


Huawei Confidential

Page 15

Typical Case 2: In a site, it is required that more than 95% MS-PSTN MOS values should be greater than 2.7 in the drive test of the entire network after the migration. According to the drive test result, only 90% MS-PSTN MOS values are greater than 2.7. Therefore, the MOS value fails to reach the standard. A: According to the analysis, no known speech problem exists in the BSS after the migration. The reason that the MOS value of the existing network fails to reach the standard is that the quality over the Um interface is low, many handovers occur, and the ratio of half rate channels is high. B. After optimizing the concentric handover parameters, adjusting the Assign Optimum Layer and the Pref. Subcell in HO of Intra-BSC parameters and their thresholds, and optimizing the number of handovers, the ratio between traffic and number of successful handovers rises from 68.4 to 71.9. In addition, the ratio between number of MOS values and number of handovers in the drive test rises from 1.86 to 2.76. Therefore, the ratio of MOS values that are grater than 2.7 of the entire network rises about 4%. C. After optimizing cells one by one and expanding the capacity of busy cells, the ratio of half rate channels in the test is reduced from 46% to about 35%. This improves the overall MOS. D. After optimizing problem sites one by one and take optimization measures at a low carrier-to-interference ratio (CIR), the ratio of Um interfaces whose quality is at level 0 to level 4 rises 2%. After taking a series of optimization measures, the MOS value in the drive test is improved obviously. The ratio of MOS values that are greater than 2.7 rises about 10% and reaches over 95%.


Huawei Confidential

Page 16

Typical Case 3: The MOS value of a migrated site is 0.1 less than that of the original network. 

According to the analysis, the smaller MOS value is

mainly caused by abnormal low MOS values. Based on the problem location and isolation process and the analysis of the TC recording over the A interface on the BSC and on the core network, engineers find that the TC recording is normal on both areas. However, packet loss occurs on the downlink recording data before the data enters the A interface and the core network, as shown in the figure. It is concluded that the PSTN causes the low MOS value. 

The DT data after processing on the PSTN shows that the MOS values are better than those of the original network.


Huawei Confidential

The upper figure: Downlink voice sample on the A interface The lower figure: Original voice sample on the PSTN

Typical Case 4: 

The TFO fails to be established sometimes when the TFO of a Huawei BSC interconnects with that of an Ericsson or NSN BSC.



Cause:



On a Huawei BSC earlier than BSC6900 V900R011C00SPC756, the TFO-related protocol content has not been updated in the implementations and therefore the TFO cannot interconnect with the

TFO of BSCs from other vendors.  

Problem Description: The TFO fails to be established when a Huawei HAMR channel interconnects with an Ericsson FAMR channel.



Upon receiving the speech version of the FAMR from the Ericsson BSC during a TFO negotiation, the Huawei BSC decides that the TFO frame type from the Ericsson BSC is AMR_TFO_16k and enters the TFO establishment process normally. However, the Huawei BSC keeps receiving TFO frames of AMR_TFO_8+8k from the Ericsson BSC, resulting in failures to establish the TFO.



As defined by the GSM protocols, when the HAMR speech version (excluding 7.95 kbit/s) is used during the TFO negotiation, the TFO frame type must be AMR_TFO_8+8k. Therefore, Huawei needs to change the frame search mode to resolve the problem.


Huawei Confidential

Contents 


Quality (MOS) 








Quality Parameters



Codec Parameters



Handover Parameters



AoIP Parameters


Huawei Confidential

Page 19

Subjective Speech Problem Handling Currently, for problems such as one-way audio and noise location, the main application methods on the BSS side are speech loopback test and TC recording. The speech loopback function can define the NE where the problem occurs, while the TC recording function can determine whether the problem is from the TC and the specific changes. For the analysis methods of loopback tests and the TC recording file analysis, see the attachment Operation Guide for Speech Tests. The following mainly introduces the handling methods of subjective speech problems such as one-way audio, noise, echo, crosstalk, and voice make-and-break.


Huawei Confidential

Page 20

Subjective Speech Problem Handling 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 oneway 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).


Huawei Confidential

Page 21

Subjective Speech Problem Handling 2. Noise During 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-break Speech 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.


Huawei Confidential

Page 22

Subjective Speech Problem Handling 4. Echo Echoes 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.


Huawei Confidential

Page 23

Subjective Speech Problem Handling 5. Crosstalk In 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.


Huawei Confidential

Page 24

Subjective Speech Problem Handling The table on the right side lists the information on handling subjective speech problems. It should be collected and submitted after the completion of the test. For details, see the attachment Checklist for Data Provided for Speech Quality Problems (deliverables).

No.

Feedback on Speech Quality Problems

Output Description

1

For example, record whether noises periodically Detailed descriptions of the problem, including the scenario occur (namely whether the noise occurs once in which the problem occurs and the probability that the every x second (s) or every x minute (s)), and problem occurs whether noises persist during the calls.

2

Tracing 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.

4

TC 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 problem

List of loopback test results of interfaces (see the following table))

6

BSS 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.

7

Information 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 network

The transmission mode is all-TDM, all-IP, or hybrid.

9

BTS 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


Huawei Confidential

Page 25

Typical Case 1: A user complains that the speech quality is bad under a BTS at a site. The case is that everything is normal in the outdoor coverage area of this cell, but uplink quality make-and-break and noise occur in the indoor coverage areas of this BTS. A. Perform TC recording and make analysis at the problem site, and find that the voice make-and-break exists before the voice enters into the TC. For the speech frame structure at the break-and-make points, the corresponding frames are all No-Data frames, namely, one or more No-Data frames appear before two continuous speech frames without the transition of SID frame. This usually is caused by the failure in BTS decoding. B. During the test at the problem site, the uplink noise is extremely serious and continuous. Through the TRX loopback, the calling party can hear its own speech with make-and-break, it demonstrated that the voice make-andbreak problem exists between the Um interface and the BTS DSP. C. According to the test log analysis, it is concluded that the quality of the Um interface in the area where the problem occurs is poor, and the proportion of uplink quality 5, 6, and 7 is 64%, which is the major cause for voice break-and-make. RxQual 0~4 5~7 RxLev AVG Whole Indoor

90.82% 36.05%

9.18% 63.95%

-79.7 -89.2

After the adjustment of optimization measures for uplink low CIR, the speech problem on this site disappears after several times of verification.


Huawei Confidential

Page 26

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 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.


Huawei Confidential

Page 28

Typical Case 2 (continued) Average Shortest Call Number of Call Duration in BSCs Duration Normal Cases

Office

Thailand

Chengdu Average Call Duration

1.

Shantou

Shijiazhuang

Thailand

167

62

3

Chengdu

68

36

49

Shantou

80

44

11

Shijiazhuang

68

32

16

Hangzhou

87

30

63

Hangzhou

Shortest Call Duration in Normal Cases

In normal cases, the average call duration of all evaluation offices is more than 60s (Upon the analysis on 136 BSCs, there is no super short call caused by crossed pair).

2.

In abnormal cases, the average call duration is less than 29s (Thailand).

3.

Upon the analysis on the A interface occupation measurement for 10 BSCs in Nigeria, a large number of trunk occupied durations of the A interface on the port are less than 30s, with the shortest one is 12.69s. This may be relevant to the strategy of charging by second in Africa. Therefore, in Africa, the case that checking the A interface connection based on the situation that the A interface trunk occupies the super short call may be altered according to the actual situation. HUAWEI TECHNOLOGIES CO., LTD.

Huawei Confidential

Page 29

Typical Case 3: An AoIP office reports that metallic sounds occur in the existing network, consequently, some users ask to cancel the network service, provide cause analysis, and immediately solve the problem.

Analyze the TC recording and Probe file, there is no noise in the uplink UM interface voice on the calling side and in the uplink voice before entering into the core network on the BSC side. However, after the voice passes through the core network, and when enters into the BSC downlink (the called side uses the EFR speech version), the noise appear. This, as a result, can be determined that the noise is caused during the processing of core network.

The core network confirmed that in some cases, the DSP cannot complete the call processing with 20 ms, and need to re-process it 120 ms later. In the 120 ms, the DSP will send the previous data again and again, causing the metallic sounds (which is complained by users) acoustically. After the core network engineers optimize the scheduling algorithm of the internal DSP, re-test the problem message on site.


Huawei Confidential

Page 30

Contents 


Quality (MOS) 








Quality Parameters



Coding Parameters



Handover Parameters



AoIP Parameters


Huawei Confidential

Page 31

Voice-Related Key Parameters: 

Quality Parameters    



Coding Parameters   



Speech versions Parameters related to VQE Parameters related to channel allocation

Handover 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

Handover-related parameters

AoIP Parameters  

Mapping versions related to AoIP AoIP-related parameters


Huawei Confidential

Page 32

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. HUAWEI TECHNOLOGIES CO., LTD.

Huawei Confidential

Page 33

Quality Parameters: Parameters Related to Improving the Speech Quality at a Low C/I 

Description of the Enhanced Interference Cancellation Combining (EICC) Function:

Among signals received by dual antennas, the interference is related to both the space (between antennas) and the time. The EICC function considers both the relationship between interference and space and the relationship between interference and time. In this way, it suppresses interference more effectively and improves the voice quality. Suggested Parameter Settings: 

STIRC Allowed: Yes when serious interference exists.


Huawei Confidential

Page 34

Quality Parameters: User Experience-Related Parameters VQE: mainly includes four sub-features, such as AEC, ALC, ANR, and ANC. Description of the Acoustic Echo Cancellation (AEC) Function This function cancels the acoustic echo generated during the call. It determines whether the signals input by the local and remote ends are echo to the local end. If the signals received by the local end are echo from the remote end, this function attenuates the echo and replaces the echo with comfortable noise. If the signals

received by the local end are voice of the speaker, this function keeps the voice unchanged. Suggested Parameter Settings:


Huawei Confidential

Page 35

Quality Parameters: User Experience-Related Parameters 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:


Huawei Confidential

Page 36

Quality Parameters: User Experience-Related Parameters Description of the Automatic Noise Compensation (ANC) Function: This function compensates the noise automatically. It evaluates the level of the background noise at the local end and the voice level at the remote end. When the background nose at the local end is great, this function turns up the volume of the voice input by the remote end. This improves the signal-tonoise ratio between the voice at the remote end and the background noise at the local end. Therefore, the hearer at the local end can hear the voice of the speaker at the remote end clearly. Suggested Parameter Settings:


Huawei Confidential

Page 37

Quality Parameters: Parameters Related to Power Control 

Parameters Related to Power Control:  III Power Control Algorithm Switch  III Power Control Optimization Algorithm Switch



Basic Principle:  Power control: When the uplink and downlink signals are strong, reduce the uplink and downlink transmitting power to reduce the interference of the entire network.  Remarks: The principle of 3.5-generation power control algorithm is advanced in the industry. This algorithm implements power control based on the quality.



Suggested Parameter Settings: 

Currently, 3.5-generation power control algorithm is widely promoted globally. It is required to enable both III power control algorithm switch and III power control optimization algorithm switch. The 3rd-generation power control algorithm, however, is not recommended.


Huawei Confidential

Page 38

Quality Parameters: Other 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.


TC CRC Allowed: ON


Huawei Confidential

Page 39

Voice-Related Key Parameters: 




Codec Parameters   












Huawei Confidential

Page 40

Voice-Related Key Parameters – Summary of Codec Parameters Codec Parameters Speech versions

Parameter Name FAMR/EFR/HAMR/FR/HR

Commend Value

Parameter Description

AMR

Adjusts the speech coding mode on the uplink and downlink according to changes in Um interface quality, thereby improving the speech quality.

ON

Reduces the impact of speech transcoding on the speech quality, improving the speech quality.

OFF

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

OFF

Specifies 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

YES

Voice Quality report switch

YES

Reduce 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). 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.

TFO Switch RTPSWITCH

Speech Channel Alarm Threshold

10

Specifies 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

6

Specifies 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.

Enhancement of speech quality

TCMUTEDETECTFLAG

Specifies whether to enable the class-1 one-way audio detection function.

MUTECHECKCLASS1PERIOD

5

Specifies 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

25

Specifies 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

ON

Specifies whether to enable the one-way audio and no audio detection function to improve the accuracy of oneway report.

2

Specifies 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.

CHALLOCSTRATEGY

CAPABILITY

DETECTFRAMEPERIOD

Channel allocation

ON

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


TCHTRIBUSYUNDERLAYTHR

60



Huawei Confidential

Page 41

Codec Parameters: Speech Versions Due to coding differences, different speech versions have different speech quality MOS values. The following table lists the MOS values at different encoding rates and at different C/Is. Overall, the MOS in FAMR is greater than that in EFR, the MOS in EFR is greater than that in HAMR, the MOS in EFR is greater than that in FR, and the MOS in FR is greater than that in HR.


Huawei Confidential

Page 42

Codec Parameters: Speech Versions The key idea of speech version 3 (AMR) is achieving the best balance between the speech quality and system capacity by continuously adjusting the uplink and downlink voice coding schemes according to the quality changes of the uplink and downlink signals on the GSM Um interface. For the recommended settings of AMR parameters, see the table on the right. Note: A bad quality cell indicates a cell where

over 5% of the uplink and downlink receiving qualities are at levels 5, 6, and 7 according to the counters or a cell where over 5% interference bands are interference band 4 and interference band 5 according to the

interference band statistics.


Huawei Confidential

Page 43

Codec Parameters: Speech Versions Description of AMR gain reflected in MOS and enabling scenarios 

Scenario 1: Compared with the HR speech version, the gain of HAMR speech version reflected in the MOS exists all the time and is obvious. If the FR and HR speech versions are used on site, the AMR function can be enabled to obtain better speech quality and MOS values.



Scenario 2: Compared with the EFR, the MOS gain is mainly reflected in low CIR scenario. When the UM interface quality is good, the gain is not obvious. Compared with common FR, in the same scenario, the gain of FAMR on the MOS exists all the time. If the on-site speech versions include the EFR and HR versions, see the following items for reference to determine whether it needs to enable the AMR function to obtain better speech quality: (1) Judge according to the half rate traffic. If half rate occupation exists in busy hours in the traffic statistics, it is recommended to enable the half rate HAMR (half rate version 3). (2) Judge according to the receiving quality:



If the Um interface quality is good, usually it is regarded that the proportion of uplink and downlink receiving quality at levels 0-5 exceeds 98%, the MOS gain of the EFR is not obvious after the FAMR is enabled. In this scenario, the FAMR function does not need to be enabled.



If the receiving quality of the cell is bad, usually it is regarded that the proportion of uplink and downlink receiving quality at levels 5-7 exceeds 5%, the proportion of interference band at levels 4-5 exceeds 5%, or it is found that the quality is bad in the drive test areas, or the receiving quality is bad at some cells (the proportion of receiving quality at levels 5-7 exceeds 5%), it is recommended to enable the FAMR function. The MOS gain in this scenario is reflected in the case that the AMR improves the speech quality through changing the encoding mode of the AMR rate in low CIR.


Huawei Confidential

Page 44

Codec Parameters: Speech Versions Description of gain to capacity and coverage produced by the AMR and enabling scenarios Capacity: Because both the uplink and downlink of the AMR have good immunity from interference, the AMR can tolerate larger identical and adjacent channel interference, achieving tighter multiplexing of frequency and improving utilization of frequency spectrum resources. In addition, the speech performance of the HR AMR equals that of the FR AMR. However, the occupied wireless interface bandwidth of it is only the half of that of the FR AMR. In this case, the AMR can effectively improve the frequency spectrum utilization, increase the system capacity, and reduce the cost through methods such as tight multiplexing and half rate AMR. When the AMR penetration is 100% through simulation, the system capacity can increase 140% compared with the EFR speech version. Coverage: The AMR code has good immunity from interference. Therefore, the AMR code can support lower CIR compared with non-AMR codes in the same frame error ratio (FER). In other words, the AMR can make a call in areas where non-AMR codes cannot make a call in the past. Therefore, the AMR code has better coverage performance, and has large gain for breadth coverage (spacious areas) and depth coverage (fade areas, shadow areas, and floor and indoor areas). However, because the robustness of the AMR speech frame and that of the SACCH frame, the actual covering power is decided by that of the SACCH channel. Therefore, in actual applications, the value of the RLT and the number of SACCH frames shall be set to a larger value for the AMR channel. In this way, the robustness of the SACCG channel is increased, improving the AMR network coverage performance and reducing the call drop rate. To improve the network capacity and coverage, you are advised to enable the AMR function.


Huawei Confidential

Page 45

Codec Parameters: Speech Versions Parameter

Mapping

Impact of the AMR function on KPIs:

Call Parameter

(1) In the existing network of non-AMR speech version, after the

AFRSAMULFRM

The value is identical to that of the SAMULFRM.

AFRDSBLCNT

The value is identical to that of the RLT.

AHRSAMULFRM

The value is identical to that of the SAMULFRM.

AHRDSBLCNT

The value is identical to that of the RLT.

AMR function is enabled, the ratio of uplink receiving quality at level 6 and level 7 deteriorates. The causes are as follows: 



HO Parameter

Different algorithms specified in the protocol lead to the different

RXLEVOFF

results of the AMR and non-AMRs.

DLQUALIMITAMRFR

The suggested values of AMR parameters and non-AMR parameters

ULQUALIMITAMRFR

in algorithms such as the handover, power control, and call control

DLQUALIMITAMRHR

are different, which may also lead to the deterioration of the quality statistics after the AMR is enabled. This can be avoided through configuring the AMR parameters and

0 The value is identical to that of the DLQUALIMIT. The value is identical to that of the ULQUALIMIT. The value is identical to that of the DLQUALIMIT.

ULQUALIMITAMRHR

The value is identical to that of the ULQUALIMIT.

INTRACELLFHHOEN

N/A

PC Parameter

non-AMR parameters to be consistent with each other in the

DLAFSREXQUALHIGHTHRED

algorithm. However, this may lose some gain brought about the AMR

DLAFSREXQUALLOWTHRED

feature. Considering from the overall performance, do not use the

DLAHSREXQUALHIGHTHRED

parameter mapping on the right side, unless otherwise to solve the

DLAHSREXQUALLOWTHRED

problem of the deterioration of uplink receiving quality at levels 6 and

ULFSREXQUALHIGHTHRED

7 after the AMR is enabled.

ULFSREXQUALLOWTHRED

The value is identical to that of the DLFSREXQUALHIGHTHRED. The value is identical to that of the DLFSREXQUALLOWTHRED. The value is identical to that of the DLHSREXQUALHIGHTHRED. The value is identical to that of the DLHSREXQUALLOWTHRED. The value is identical to that of the ULFSREXQUALHIGHTHRED. The value is identical to that of the ULFSREXQUALLOWTHRED. The value is identical to that of the ULHSREXQUALHIGHTHRED. The value is identical to that of the ULHSREXQUALLOWTHRED.

ULHSREXQUALHIGHTHRED ULHSREXQUALLOWTHRED Channel Parameter

The value is lower than that of the TCHBUSYTHRES.

AMRTCHHPRIORLOAD


Huawei Confidential

Page 46

Codec Parameters: Speech Versions 

Because when the statistics of BER of AMR and that of non-BMR are collected, the algorithm is different, which will lead to the case that the proportion of uplink receiving quality at level 6 and 7 for AMR is larger than that for non-AMRs. Currently, the calculation accuracy of receiving quality for products is improved by adjusting the corresponding software parameters through version optimization in BSS9.0. The specific parameters are as follows: Cell software parameter command: SET GCELLBTSSOFT: IDTYPE=BYID, CELLID=0, ITEMINDEX=57, ITEMVALUE=219; Note: After the AMR is enabled, the proportion of uplink receiving quality at level 6 and 7 deteriorates, which is only the change in statistics, and does not affect the actual user perception and UM interface quality. Other vendors (including Ericsson and Nokia Siemens Networks) also have the problem that the proportion of uplink receiving quality at level 6 and 7 deteriorates after the AMR is enabled. The figure on the right side lists the comparison data of receiving quality of Ericsson in existing networks before and after the AMR

is enabled. It can be seen that the proportion of uplink receiving quality at level 6 and 7 deteriorates, which reduces about 0.5%.


Huawei Confidential

Page 47

Codec Parameters: Speech Versions (2) Due to the difference of statistics formula for the threshold to define that the AMR and non-AMR traffic is busy, the half rate traffic is reduced after the AMR is enabled. The problem can be solved by modifying LOADSTATYPE (cell load calculation type) in GBSS9.0. The specific adjustment is as follows: SET GCELLCHMGAD: LOADSTATYPE=NODYNPDCH; However, the problem can be avoided only through lowering the AMRTCHHPRIORLOAD parameter in GBSS8.1. (3) When the Abis interface is in HDLC transmission mode, due to the allocation of half rate channel bandwidth algorithm in HDLC mode (The HAMR channel bandwidth is allocated according to the highest rate of HAMR rates in HDLC mechanism. In this way, when the centralized rate is larger than or equal to 5.9 kbit/s, the bandwidth

exceeds that of the common HR version), the transmission resources occupied by the HAMR after the AMR function increase. In case that there are much half rate traffic, transmission resource congestion may occur. (4) The drive test HQI decreases due to statistical differences of the TEMS when the AMR function is enabled. When the DTX is enabled, the TEMS measures the values of RXQUAL_FULL and RXQUAL_SUB simultaneously. Manual collection of the HQI only involves the RXQUAL_SUB values. For the non-AMR traffic, the TEMS collects the RXQUAL_SUB values based on eight voice frames and four SACCH frames. However, the TEMS collects the RXQUAL_SUB values only based on SACCH frames in the AMR traffic. As a result, the HQI results deteriorate when the AMR is enabled. On Dingli devices, the same problem occurs because the TEMS collects the RXQUAL_SUB values based on the SID_UpDATA and SACCH frames when the AMR is enabled. This problem is caused by a defect in the TEMS and no solution is available. HUAWEI TECHNOLOGIES CO., LTD.

Huawei Confidential

Page 48

Codec Parameters: Speech Versions (5) The MOS values have not achieved any gains during the DT when the AMR function is enabled. 

The major cause lies in the selection of the test points. This problem mainly occurs along the main streets or roads where the proportion of the half rate channels is low. In these scenarios, the voice quality almost reaches the maximum and therefore the AMR function cannot implement the signal gains. The gains of the MOS values can be achieved during the drive test only when the following requirements are met:



a. Perform a drive test along a route that consists of main roads and side roads.



b. Ensure that the proportion of half rate channel is not less than 15% during the drive test.



(6) The call drop rate increases a little. 

The AMR function helps improve the capability of voice frames to resist interference. With the AMR function, call duration can be prolonged within an area where the signal quality is poor. However, the capabilities of SACCH frames have not been improved with the AMR function. The determination of call drops is based on the fact whether the SACCH frames can be correctly demodulated. With the AMR function enabled, the call duration can be prolonged within an area where the signal quality is poor because users seldom terminate calls. As a result, there will be a high probability that call drops occur.



To resolve the problem, modify the values of radio link counter AFRDSBLCNT/AHRDSBLCNT and AFRSAMULFRM/AHRSAMULFRM for AMR calls to ensure that their values are the same as those for non-AMR calls. In this case, however, the AMR function cannot be fully achieved.

Note: For detailed description, see the 04 GSM BSS AMR Performance Technical Disclosure. HUAWEI TECHNOLOGIES CO., LTD.

Huawei Confidential

Page 49

Codec Parameters: Parameters Related to VQE 

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


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.

Huawei Confidential

Page 50

Codec Parameters: Parameters Related to Enhancement of Speech Quality 

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.




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.


AMR Uplink Adaptive Threshold Allowed: Yes



AMR Downlink Adaptive Threshold Allowed: No


Huawei Confidential

Page 51


Description of the Mute Detection Function: In the TDM scenario, you can set the Mute Detect Class1 Switch parameter to flexibly adjust the detection conditions and to increase the chance of detecting mute. Then, you can configure the Detect Class2 Switch parameter to confirm the mutes detected by Mute Detect Class1 Switch to improve the correctness of the mute report. This function provides additional information for locating problems, preliminarily determines the problem devices, and narrows the scope of problem devices. Suggested Parameter Settings:


Huawei Confidential

Page 52


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.

Suggested Parameter Settings:  



Report Speech Quality: Report Report Downlink VQI Allowed: Disable

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.


Huawei Confidential

Page 53

Codec Parameters: Parameters Related to Channel Allocation 

Parameters Related to Channel Allocation:   



Basic Principles:  





Channel Allocate Strategy Traffic Busy Threshold Tch Traffic Busy Overlay/Underlay Threshold

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.

Suggested Parameter Settings:    

Channel Allocate Strategy: Quality preferred Traffic Busy Threshold: 60 Tch Traffic Busy Overlay Threshold: 70 Tch Traffic Busy Underlay Threshold: 60


Huawei Confidential

Page 54

Voice-Related Key Parameters:


















Huawei Confidential

Page 55

Voice-Related Key Parameters – Summary of Handover Parameters Handover Parameters

Parameter Name

HOCTRLSWITCH

Handover parameters

Commend Value

Parameter Description

HOALGORITHM1 Indicates the handover algorithm.

PBGTSTAT(s)

5

Sets the triggering time of handover to a proper value to reduce the impact of handover on the speech quality.

PBGTLAST(s)

4


ULDATAFWDTMR

180

Delays the release time of the old channel, reducing the number of frames that are lost during the handover.

INTRACELLFHHOEN

YES

Triggers the AMR TCHF/TCHH handover.

INHOH2FTH

16

Enables the half rate to be handed over the full rate, improving the user perception.

INFHHOSTAT(s)

5


INFHHOLAST(s)

4


OPTILAYER

SysOpt

Sets the concentric cell-related parameters to proper values to reduce the number of handovers.

HOALGOPERMLAY

SysOpt


USubcell


ACCESSOPTILAY


Huawei Confidential

Page 56

Handover Parameters: Handover-Related Parameters 



Handover-Related Parameters: 

HO Algorithm Selection Switch



PN Rule of PBGT



Timer for UL Data Forward

Basic Principle 

When other KPIs are not affected, it is recommended to increase the PN of PBGT handover and to delay triggering a handover. This improves the speech quality.

Uplink data smooth timer: This timer delays releasing old channels and reduce the number of lost frames during the handover. 


HO Algorithm Selection Switch: HO I Algorithm

PBGTSTAT (s): 5; PBGTLAST (s): 4 Note: The suggested parameter values are to reduce the PBGT handover times, reducing the impact of handover on voice MOS values. These suggested values will impact on the success rate of handovers, you are advised to use them only when you handle speech problems. If you handle other KPIs, see the parameter baselines. 



Timer for UL Data Forward (ms): 180


Huawei Confidential

Page 57

Handover Parameters: Handover-Related Parameters 

Description of the Intracell F-H HO Function: 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


Huawei Confidential

Page 58

Voice-Related Key Parameters 
















Huawei Confidential

Page 59

Mapping Versions Related to AoIP NE Versions on the Wireless Network BTS BSC

China BTS3900

GBSS9. 0

Areas Outside China BTS3012

BTS3900

BTS3012

SingleRAN MBTS

BTS3000 BTS3000 BTS3900 BSC6900 BTS3000 BTS3000 V300R009C00SPC V300R009C00SPC V100R002C00SPC V900R011C00SPH73 V100R009C00SPC08 V200R009C00SPC00 030 and later 030 and later 250 and later 2 and later versions 9 and later versions 2 and later versions versions versions versions

BSC6900 BTS3000 GBSS12 V900R012C01SPH51 Not recommended. Not recommended. V100R012C00SPC04 .0 2 and later versions 2 and later versions

BTS3900 BTS3000 V100R003C00SPC V200R009C00SPC00 350 and later 2 and later versions versions

NE versions on the core network MSC (Huawei)

CPCI V100R008C03SPH209 (for areas outside China), CPCI V100R007C05SPH209 (for China), or ATCA V200R008C03SPH209 and later versions

UMG (Huawei)

V200R008C03SPH117 and later versions

IPCLK Server (Huawei)

IPCLK1000 V100R002C01SPC200 and later versions


Huawei Confidential

Page 60

AoIP-Related Parameters on the Wireless Network 

Setting of the wireless network



(1) All MSCs and BSCs are configured with AMR full-rate (FR) codec set 1 (12.20 kbit/s, 7.40 kbit/s, 5.90 kbit/s, and 4.75 kbit/s) and AMR half-rate (HR) codec set 1 (7.40 kbit/s, 5.90 kbit/s, and 4.75 kbit/s).



(2) All cells are configured with the same speech version and rate set.



(3) Ensure that bit 13 in reserved parameter 3 is set to 0 in versions earlier than BSC6900 V900R011C00SPH728 and set to 1 in BSC6900 V900R011C00SPH728 and later versions.



(4) Ensure that bits 14 and 15 in reserved parameter 3 are set to 00 in GBSS9.0. In the case of upgrade, however, set this parameter the same as the value before the upgrade. To query the parameter, run the following command: LST OTHSOFTPARA: LstFormat=VERTICAL; Bits 14 and 15 indicate the strategy by which the BSC selects the speech version during intra-BSC handover in the A over IP mode. The values are as follows:

00: MSC strategy 01: BSC strategy 10: Speech version in originating cell preferred 11: MSC strategy 

(5) In GBSS12.0, set the speech version selection policy during intra-BSC handover to MSC Strategy by running the following command: SET AITFOTHPARA: CNNODEIDX=XX, SpeechVerStrategyInAss=MSC_STRATEGY;



(6) Set the timer T25 (INTRABSCCODECHOCMDTIMER) to 5000 ms by running the SET GCELLTMR command.


Huawei Confidential

Page 61

AoIP-Related Parameters on the Core Network 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). (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 networksupported 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

Page 62

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 of the MOS and subjective user experience.


Huawei Confidential

Page 63

Typical Case 1: 

After the A over IP reconstruction is complete at a site, no audio occurs for a short or a long period of time.

 

Cause: The AoIP-related parameters in the wireless network and the core network are not correctly configured according to the requirements in the Version Policies and Configuration Requirements for IP-Based GSM V1.26. No audio may occur sometimes because the Um interface quality is not satisfactory.



To use the TrFo function during AMR handover between FR and HR in the A over IP mode, ensure the following settings:



(1) If the Huawei core network does not support the TrFO during AMR handover between FR and HR, bit 13 in reserved parameter 3 must be set to 0 in the wireless network. Otherwise, users of MSs on the half rate channel cannot hear any audio during calls.



(2) If the Huawei core network supports the TrFO during AMR handover between FR and HR, bit 13 in reserved parameter 3 must be set to 1 by default in the wireless network and bits 8 and 9 in soft parameter P13 must both be set to 0. Otherwise, no audio may occur for a short period of time.



In both scenarios, the SET UPPARA command must be executed to set RC2CMR and CMR2RC to OPEN in the core network.


Huawei Confidential

Page 64

Typical Case 2: 

After the A over IP reconstruction is complete and the AMR codec set is modified at a site, the call drop rate deteriorates.



The AMR codec set before the IP reconstruction consists of the ACS: {12.2 , 10.2 , 7.95 , 5.90} and the HACS: {7.40 , 6.70 , 5.90}. After the A over IP reconstruction, the AMR full-rate (FR) codec set 1 (12.20 kbit/s, 7.40 kbit/s, 5.90 kbit/s, and 4.75 kbit/s) and AMR half-rate (HR) codec set 1 (7.40 kbit/s, 5.90 kbit/s, and 4.75 kbit/s) are used as defined by the relevant GSM protocol.



The BSC assigns and sends AMR codec set configuration to BTSs in the following ways before and after the changes in the AMR codec set.



When a BSC in A over IP mode is configured with the standard AMR codec set 1, the MSC compares the capabilities of the BSC and MSs and then sends the standard AMR codec set 1 (including 12.20 kbit/s, 7.40 kbit/s, 5.90 kbit/s, and 4.75 kbit/s) to BTSs.



When a BSC in A over IP mode is configured with a non-standard AMR codec set, the MSC detects the nonstandard AMR codec set and sends the AMR single codec set to BTSs under the BSC.



Compared with the AMR multiple codec set (including 12.20 kbit/s, 7.40 kbit/s, 5.90 kbit/s, and 4.75 kbit/s), the

AMR single codec set (including only 12.20 kbit/s) cannot enable the BSC to select adaptable codec rates based on the Um interface quality. As to user experience, when the AMR multiple codec set is configured, users can maintain calls and call drop rate may increase if the Um interface quality is poor. When the AMR single codec set is configured, the voice quality will be the worst and users may have to hang up. Therefore, the call drop rate is different for two different AMR codec sets.


Huawei Confidential

Page 65

Typical Case 3: 

When a Huawei BSS in A over IP mode interconnects with the MSC from other vendors, such as NSN or ZTE, no audio can be heard sometimes by both parties when the speech version of any party is half rate.

 



Onsite BSC version: BSC6900 V900R011SPH722 Cause:

A Huawei BSC processes HR voice frames in a different way from some BSCs supplied by other vendors. In the A over IP mode, the Huawei BSC encodes HR voice frames into packets of 14 bytes. However, some BSCs supplied by other vendors have used the latest codec mechanism in related GSM protocols by adding a TOC byte to the front of the 14 bytes of HR voice frames. The TOC byte indicates a voice frame, non-voice frame, or null frame. When a Huawei BSC interconnects with an MSC from another vendor and the HR voice frames processed by the Huawei BSC are sent to the MSC, the MSC fails the frame format check and dumps the frames directly. Therefore, the problem occurs.



Solution: Upgrade to the BSC6900 V900R011SPH726



Occurrence scenario: A BSC6900 V900R011SPH726 in A over IP mode interconnects with the MSC from NSN or ZTE and the HR speech version calls are initiated.


Huawei Confidential

Page 66

Attachments


Huawei Confidential

Page 67

Thank you www.huawei.com

110857247 3 GSM Speech Quality Influence Factors Troubleshooting Methods and Tools Deliverables 20110730

Recommend Documents