RG CDMA Network Planning and Network Optimization Cases AnalysisWireless Products Course Development Room

HUAWEI

Introduction Concern current network running status and quality of service (QoS). Concern current network structure and rational configuration. Deepen the understanding about the system according to network faults and their solutions. Improve the algorithm according to the solutions to network faults.

HUAWEI

Objectives After the course, you will be able toDeepen the understanding about system planning.Enrich the knowledge about optimization procedure and methods.Grasp difficulties and emphasis in the work.

HUAWEI

Contents Chapter 1 Planning Cases

Chapter 2 Optimization Cases

Chapter 3 Terminal Cases

HUAWEI

Chapter 1 Planning Cases Planning contentsDifficulties and emphasesCase review

HUAWEI

Content of Project PlanningProject planning plays an important role in the project implementation, and has a great impact on the network quality and optimization after project completion. The content is as follows:Project parameters: Latitude and longitude of BTS, mounted height, azimuth, PN, adjacent cell and BTS power.Region parameters: location area code (LAC), REGZONE, signaling points and their coordination.Network performance parameters: performance parameters The parameters in this page are not enumerated completely. For the specific parameters, refer to network planning parameters list.

HUAWEI

Difficulties and Fallibilities of Project Planning LAC and REGZONELAC, signaling points and BSS FrameAdjacent cell dataSearch window configurePower setting (Focus on mixed networking)Now that we have got some information about the system, think about what will happen in following cases.

HUAWEI

Case 1 As shown in the figure, network in X city is topographically composed of 2 LACs and 2 OPCs. And only one REGZONE is available at the moment. Due to the large number of BTSs, they are distributed into 4 frames. To balance the traffic between frames, the project engineers arrange several urban BTSs with heavy traffic into rural frames. Then, what will happen? How to handle it?

HUAWEI

Problems: Due to 2 LACs and 1 REGZONE, terminal will not originate location registration when performing handoff between LACs, which leads to terminal paging failure before originating calls.Two OPCs in every LAC will necessarily bring about a successful and a failed response to the paging originated by MSC, which makes a low paging success ratio at BSC side. Urban BTSs are distributed among frames and between the 2 signaling points, which leads to a mass of soft handoffs between frames or between signaling points. Large numbers of call drop will occur when CPU is overloaded.Solutions:Add REGZONE corresponding to each LAC.Switch cross-regional BTSs to the same frame in the next cutover.

HUAWEI

Think about: What should we concern about when planning location area and signaling points during project planning?The planning of BTSs must conform to the following rule: geographically adjacent BTSs, especially the BTSs with frequent handoff, must be installed in the same module.New office must reserve port numbers for BTS expansion. The reserved numbers are decided by data setting center. The port numbers are not reserved currently. They are added into new modules during each expansion.One signaling point can be connected with one or more modules. One signaling point can be connected with 15 modules at the most. (Each module with one FMR)Each signaling point should correspond to one LAC. One signaling point may have several LACs. But avoid the case that one LAC have several signaling points. It is not allowed that BTSs in the same LAC are distributed in different modules and several LACs exist in the same module in such configuration. Each signaling point should correspond to one LAC, and handoff area must be in the sparsely populated area.

HUAWEI

Case 2 Case: Two 3601 i-sites of V100R001B02D002 are added in an office expansion. The call drop rate is considerable since the 2 i-sites are activated, which keeps fluctuating at 5%. The call drops in one i-site even reaches 120 in a single day.Most of call drops in the 2 i-sites are caused by radio link according to traffic measurement result. All the indexes are normal except the considerable call drops and high call drop rate. The drive test shows that indexes such as EC/IO, RX and FFER are normal except TX of transmit power. TX in some area even reaches +20 dBm, indicating a very poor reverse coverage.Setting search window closely relates to the coverage capability of BTS. During on-site implementation, we should set the search window parameters according to practical planning and avoid network failure due to incorrect parameter setting.

HUAWEI

Causes 2Coverage causes: Forward coverage is good, while the reverse one is poor.PN planning causes: It is excluded by check.Hard handoff causes: It is excluded since it is not involved.Soft handoff causes: Since no call drop upon handoff is captured in on-site drive test, it can be excluded.Data setting causes: We find that the reverse coverage of i-site is poor according to the analysis of drive test and traffic measurement. Then modify the parameters to improve the effect of the reverse coverage of i-site.

HUAWEI

SolutionsIncrease the coverage radius of BTS from 20 km to 39 km and enlarge search window from 1 to 5.Change the dynamic adjustment of search window from ENABLE into DISABLE.Result:Call drop rate falls obviously, which decreases from about 5% to less than 1%.Sum-up: Setting search window closely relates to the coverage capability of BTS. During on-site implementation, we should set the search window parameters according to practical planning and avoid network failure due to incorrect parameter setting.

HUAWEI

Case Review 3 Case: The equipment around site A is manufactured by M. The customer expects the areas covered by Site A are distributed asymmetrically. There are Blind Zones with short distance or long distance from site A in some directions. There are also weak signal areas in some directions. MS access is difficult and call drop occurs frequently (34 out of 10 call drops occur) before activating Huawei omni-directional i-site in site A. And MS access is still difficult and call drop occurs frequently (78 out of 10 call drops occur) after activating Huawei omni-directional i-site in site A.

HUAWEI

Cause Analysis There are 4 pilot signals with similar strength in Site A before activating Huawei omni-directional i-site. The 4 pilot signals will create pilot pollution, which causes difficult MS access and frequent call drops. (This is confirmed by the test to network M)Since the area where problem occurs is 3 km away from site A, and the signals transmitted by the omni antenna are not strong, the activation of Huawei omni-directional i-site equals to the addition of a new pilot signal with similar strength. Therefore, the interference increases, aggravating the problem.

HUAWEI

Solutions Replace omni antenna of Isitec3601 with directional antenna to make that the main pilot to cover the area. Move the handoff band to the region without residents, and the call drops in 90% of the coverage areas decrease.Since the M network structure cannot be modified, we can modify the system parameters of network M only. Faults in 95% of the area disappear, but they remains in some small spot interference area, which can be removed by adjusting network structure and increasing the tilt angle of BTS generating the interference.

HUAWEI

Suggestion and Sum-up Since the equipment around the site is from a third-party manufacturer, it is not suitable to install Huawei equipment in the blind zone. Since it is not possible to shift handoff band to the area without residents, and even it is, some small interference areas in the coverage area cannot be controlled.Since Huawei i-site is suitable for blind zone in belt shape or standalone site, the sites committed by market department should be monitored during network planning. If it is compulsory to install Huaweis BTS in similar environment, select the directional antenna for adjustment instead of omni-directional antenna.

HUAWEI

cases summarizer We should concern the following in planning according to the above three cases:BTS address (Avoid mixed networking)Antenna System (Including azimuth, tilt angle and antenna selection)Network parameters (Search window, carrier power and common channel power)Region parameters (Location area parameters and signaling points)

HUAWEI

ContentsChapter 1 Planning Cases

Chapter 2 Optimization Cases

Chapter 3 Terminal Cases

HUAWEI

Chapter 2 Optimization CasesOptimization ProceduresDifficulties and EmphasesCase Review

HUAWEI

Optimization Procedures The general procedures for project optimization are shown in figure on the left. Remember to conform to the general procedures to avoid careless omission when handling certain problem on site.

HUAWEI

The optimization procedures are as follows Preparation: Preparation of equipment, data and telephone directory of personnel for optimization.Obtaining basic network information on site: Network structure, network data, version information, user number and distribution.Collecting traffic measurement and drive test data: The traffic measurement data is the major criterion for operator evaluation. The traffic measurement and drive test data are important for analysis.Data analysis: Analyze the collected data with various tools.Adjusting parameters: Adjust corresponding network parameters according to analysis result.Network optimization report: network optimization deliverables and work achievements. The report should be acknowledged.

HUAWEI

Difficulties and Emphases The advantage of CDMA over 2G lies in the application of power control and soft handoff, which is the emphasis of the optimization. Besides, the optimization of data service is also important, which should be focused on.Optimizing power control parameter Optimizing access parameters Discover the problems of algorithm and software.Optimizing load controlOptimizing data service

HUAWEI

Case Review 1 Case: call setup success rate fluctuates at about 91.5% after the cutover (R02B03) of project Z. 2% of the failures are caused by MSC rejection and call early release. And 4.5% of the call setup failures are caused by capture failure of traffic channel preamble and MS response expiration. Cause analysis:Call setup failure is generally caused by capture failure of traffic channel preamble and MS response expiration, which exists in almost every cell, and it is especially serious in some cells. There is obvious interference in most of the BTSs with low call setup success rate according to the analysis of reverse RSSI. However, the call drop rate of these BTSs is not high, which is less than 5%. That indicates there are other causes of call setup failure besides interference.

HUAWEI

Case Review 1The signaling procedures of call origination and paging, together with action point of open and close loop power control are shown in the following figure:

HUAWEI

Case Review 1 Analyzing from MS: MS enters close loop power control mode after sending reverse service traffic channel preamble. The MS transmit power of reverse common channel and traffic channel is calculated as follows: Transmit power on access channelAverage output power (dBm) = Average input power (dBm)+ offset power+ interference correction factor+ NOM_PWRs - 16NOM_PWR_EXTs+ INIT_PWRs+ PWR_LVLPWR_STEPsTransmit power of reverse traffic channel before MS receives power control bitAverage output power (dBm)= Average input power (dBm)+ offset power+ interference correction factor+ ACC_CORRECTIONS + RLGAIN_ADJIt is possible thatDifferent parameters of RC power control (Power offset) brings different results

HUAWEI

Case Review 1 The offset power of both varies according to different radio configurations. The access channel in the case is IS95 access channel and the offset value is 73, while the traffic channel is IS2000 traffic channel and the offset value is 81.5, therefore, 8.5 dB is reduced in transmit power when MS accesses traffic channel. The difference can be compensated by RLGAIN_ADJ. The correction rang is 015, which corresponds to 87dB. The value obtained through query of DB value is 8, that is, 0dB is corrected. Solution Case 2: call setup success rate rises when RLGIAN_ADJ is modified to 15, which corresponds to 87dB.

HUAWEI

Case Review 1From the viewpoint of BTS, most MSs in current network are in IS95A mode, so their forward power control mode conforms to slow power control mode instead of fast power control that we are familiar with. In the slow forward power control list, the maximum transmit power of forward channel is 215 and the initial value for transmit power of forward channel is 199, which is much less than the pilot channel power (227). Since the loaded script after upgrade does not involve the modification of slow forward power control, the problem may be caused by the default value set in the version. Solution: Set the value of maximum transmit power of forward channel to 227 and the initial value to 227 through dynamic data configuration.After modification, the forward load is less than 70%, which has little effect on system,

HUAWEI

Case Review 2 Case: There is always way of optimization of CDMA MS access speed. Users of certain commercial office complain that it takes a long time for all MSs in the network to be paged when they are called. The problem occurs in CDMA MSs of many models. Analysis:When BTS originates the first paging, MS does not receive the signal from BTS. This indicates the BTS transmit power is so low that MS does not respond with access probe.When BTS originates first paging, MS receives the signal from BTS, but BTS does not receive the response message from MS. MS will not stop sending access probe until MS accesses the system or call drop occurs. The key to locate the problem is to acknowledge whether MS sends the access probe when BTS sends first paging.

HUAWEI

Case Review 2ProceduresUse the MS of other brands (or MS of China Mobile or China Unicom or PSTN of China Telecom) to originate a call to the Samsung SCH130 MSClick [View/statistics/access and paging statistics] in CAIT menu. Observe the access change when the MS is called in the menu.MS has to send 6 probes consecutively before accessing system during each paging. The cause is that the initial reverse transmit power is low.pwr_step=4, that is MS has to increase transmit power by 46=24dB before accessing the system on the basis of open loop power control. The best access is generated when the second or the third probe accesses the system, which increases access speed without affecting the capacity, so we need to increase the transmit power by 4(62)=16dB. All MSs can originate one paging to access system after parameter modification.

HUAWEI

Case Review 3 We need to optimize the timer parameter while call setup success rate remains low, even if the access speed is optimized. Case: Based on on-site analysis of network optimization, the two causes affecting the call setup success rate are as follows;Traffic channel preamble failure during BTS waiting for MS serviceOn-site statistic result of the failure during BTS waiting for response from MS. Further statistics show that the first cause accounts for 66%, and the last one accounts for 34% of call setup failures.

HUAWEI

Case Review 3Cause analysis: the two causes for call setup failure are that BTS fails to capture reverse traffic channel preamble and timer expires during BTS waiting for response from MS (Ms Ack Order). The following two timers implement control: CCM_T_WT_TCH_preamble and CCM_T_WT_MS_ACK_ORD. The value of CCM_T_WT_TCH_preamble protocol ranges from 1 to 4s, and so is the value range of theCCM_T_WT_MS_ACK_ORD protocol.The values of the timers are 3s and 2s. We can increase response probability by increasing time span of the waiting timer to raise call setup success rate. Procedures: Change values of the two timers into 4s and 3s. After parameter adjustment, the indexes of failures (BTS fails to capture reverse traffic channel preamble, and failure of BTS waiting for response from MS) drop from 360 and 150 to 320 and 130 every day respectively, which decrease by 17%. It is effective to adjust timer according to traffic measurement analysis.

HUAWEI

Case Review 3

HUAWEI

Access optimization Sum-up Sum-up: We should be familiar with each stage of MS calling procedures so that we can match the stage with corresponding formula immediately.Access process should be always optimized during system optimization. Soft capacity of CDMA enables system capacity to increase and access status to vary with it.

HUAWEI

Case review 4 Call drop during soft handoff remains a trouble for us. There are lots of causes. Take the following case for an example.Case: The call drop in area A of certain office is related to PN multiplexing processing. Incorrect BSC adjacent cell judgment: BSC takes No. 17 BTS 400 km away for the current BTS serving the MSs in area A. When MS detects the incorrect adjacent cell, it reports the message to BSC by PSMM. If there is PN multiplexing in the system, BSC is unable to detect correct BTS from the PN directly. There are two solutions:Configure correct adjacent cell for BTS. BSC can detect correct BTS from PN according to adjacent cell relation.BSC can detect PN by longitude and latitude of BTS provided by system. The default detection precision requires a PN multiplexing distance of over 10 km at least.

HUAWEI

Case Review 5 We cannot neglect any details in network optimization due to lots of handoff failure causes. Case: During link test in the ETS450D network, call drop in handoff remains in some area. The area is the handoff area for MS switching from cascading BTS A (level 1 and 3) to BTS B. The handoff from BTS A to BTS B can be completed, but FER reaches 100% after handoff. RX, TX and Eclo are normal during the handoff. The conventional causes for call drop are as follows:Adjacent cell: Excluded.Hardware: ExcludedClock: Excluded Frame combination error: Analyzed through debugging consoleHO ArerA BTSB BTS

HUAWEI

Case Review 5Solutions: Change the transfer mode of cascaded BTS from IMA mode into UNI mode so as to reduce delay caused by IMA. Then the problem is solved. Refer to the attachment for the transmit mode of cascaded BTS before changing transfer mode. When cascaded BTS is used, only IMA transmit mode can be adopted in BSC. Since the UP link at BTS side always senses the transfer mode in the upper link, the transfer mode of UP link depends on that of upper link. For example, if the Down link of level-1 cascaded BTS is using IMA transfer mode, the UP link of level-2 cascades BTS should also use it. After transfer mode of DOWN links of leve-1 and level-2 cascaded BTS changes into UNI mode, the call drop disappears.

HUAWEI

Case Review 6 Some problems are impossible for us to solve except aforementioned cases solved by parameter modification. However, we cannot neglect them, and we can take mitigation measures. We can take the data service and voice service for an example to illustrate it.Case: A CDMA 800M commercial office W receives complaints that users are disconnected from network some time (several minutes later) after they originate data service call through wireless network adapter And then they have to redial to access network. Users express their great unsatisfaction with this. The type of network adapter is Putian YM-1031K PCMCIA TYPE II/ FreeWing speed up.

HUAWEI

Case Review 6 Analysis: Recur the problemThe status transfer from Active to Dormant is normal In the Dormant state, the terminal originates a request for data service. Network side sends a paging request to terminal and it responds with paging response. Then MSC send a clear command.Check the value of service option is 0 in paging response.

Since the network adapter of the terminal is not consistent with the requirement of protocol, the MSC releases the call directly. Though we require manufacturer of the network card to make modification, it is urgent to solve the problem immediately for the customer and reduce complaints. Change the duration of deactivating the timer from 20s to 180s to reduce the times that user enters the dormant state.

HUAWEI

Case Review 7 Some problems can be mitigated during optimization of voice service. Case: An office ranks low in the network quality among the offices in the province because of low call drop rate at radio side. Large number of erasure frames results in call drop, which can reach 310 in the busy hour every day. This keeps call drop rate of whole BSC under 140. Therefore we should focus on how to reduce the call drop caused by excessive erasure frames.Analysis: We should analyze the problem through csl to locate all the possible problems

HUAWEI

Case Review 7 Processing Procedure:The alarm is not generated in the cell with high call drop rate and its adjacent cell.Check reverse RSSI through telnet, the status is normal.Start CSL print to analyze the call drop caused by ERASURE frames. The value of threshold is 300 frames.Set the threshold of module/frame erasure frame timer for urban BTS to 500 frames and observe it for one day.Reset threshold of all module timers

HUAWEI

Case Review 7Sum-up:Sometimes we cannot find the key to the problems in the network within a short time, so we have resort to help from Huawei. But we are pressed to take corresponding mitigation measures on site. There is no substantial change about the optimization of parameter adjustment in this case. The time before call drop changes from 6s to 10s actually, which makes users hang up due to impatience. Then it is recorded as a normal end in measurement result. Viewed from the measurement result, the situation i

HUAWEI

Case Review 8Load control is a field much researched by Huawei. But there is still a long way to go. The forward load of a BTS sector often reaches 100% in a 15-minute traffic measurement. The customer has a great doubt about the network load ability due to the measurement result.

Chart1

10089

9573

10094

10088

100100

10085

10092

10082

8281

8886

10078

10089

10089

9889

10078

8279

9297

8786

10089

100100

10090

10076

9179

10092

8790

10095

9299

93100

9169

9193

10093

9589

9085

10098

10067

9487

8692

9889

10088

93100

9175

8894

8684

8974

9668

10091

8991

10075

9380

8387

8076

8979

9492

8273

8689

9377

0

2

/15

The forward load of a BTS

Sheet1

02

6/17/03 20:00-11(BSC->-19->-0)100-11(BSC->-19->-2)89

6/17/03 20:15-11(BSC->-19->-0)95-11(BSC->-19->-2)73

6/17/03 20:30-11(BSC->-19->-0)100-11(BSC->-19->-2)94

6/17/03 20:45-11(BSC->-19->-0)100-11(BSC->-19->-2)88

6/17/03 21:00-11(BSC->-19->-0)100-11(BSC->-19->-2)100

6/17/03 21:15-11(BSC->-19->-0)100-11(BSC->-19->-2)85

6/17/03 21:30-11(BSC->-19->-0)100-11(BSC->-19->-2)92

6/17/03 21:45-11(BSC->-19->-0)100-11(BSC->-19->-2)82

6/18/03 20:00-11(BSC->-19->-0)82-11(BSC->-19->-2)81

6/18/03 20:15-11(BSC->-19->-0)88-11(BSC->-19->-2)86

6/18/03 20:30-11(BSC->-19->-0)100-11(BSC->-19->-2)78

6/18/03 20:45-11(BSC->-19->-0)100-11(BSC->-19->-2)89

6/18/03 21:00-11(BSC->-19->-0)100-11(BSC->-19->-2)89

6/18/03 21:15-11(BSC->-19->-0)98-11(BSC->-19->-2)89

6/18/03 21:30-11(BSC->-19->-0)100-11(BSC->-19->-2)78

6/18/03 21:45-11(BSC->-19->-0)82-11(BSC->-19->-2)79

6/19/03 20:00-11(BSC->-19->-0)92-11(BSC->-19->-2)97

6/19/03 20:15-11(BSC->-19->-0)87-11(BSC->-19->-2)86

6/19/03 20:30-11(BSC->-19->-0)100-11(BSC->-19->-2)89

6/19/03 20:45-11(BSC->-19->-0)100-11(BSC->-19->-2)100

6/19/03 21:00-11(BSC->-19->-0)100-11(BSC->-19->-2)90

6/19/03 21:15-11(BSC->-19->-0)100-11(BSC->-19->-2)76

6/19/03 21:30-11(BSC->-19->-0)91-11(BSC->-19->-2)79

6/19/03 21:45-11(BSC->-19->-0)100-11(BSC->-19->-2)92

6/20/03 20:00-11(BSC->-19->-0)87-11(BSC->-19->-2)90

6/20/03 20:15-11(BSC->-19->-0)100-11(BSC->-19->-2)95

6/20/03 20:30-11(BSC->-19->-0)92-11(BSC->-19->-2)99

6/20/03 20:45-11(BSC->-19->-0)93-11(BSC->-19->-2)100

6/20/03 21:00-11(BSC->-19->-0)91-11(BSC->-19->-2)69

6/20/03 21:15-11(BSC->-19->-0)91-11(BSC->-19->-2)93

6/20/03 21:30-11(BSC->-19->-0)100-11(BSC->-19->-2)93

6/20/03 21:45-11(BSC->-19->-0)95-11(BSC->-19->-2)89

6/21/03 20:00-11(BSC->-19->-0)90-11(BSC->-19->-2)85

6/21/03 20:15-11(BSC->-19->-0)100-11(BSC->-19->-2)98

6/21/03 20:30-11(BSC->-19->-0)100-11(BSC->-19->-2)67

6/21/03 20:45-11(BSC->-19->-0)94-11(BSC->-19->-2)87

6/21/03 21:00-11(BSC->-19->-0)86-11(BSC->-19->-2)92

6/21/03 21:15-11(BSC->-19->-0)98-11(BSC->-19->-2)89

6/21/03 21:30-11(BSC->-19->-0)100-11(BSC->-19->-2)88

6/21/03 21:45-11(BSC->-19->-0)93-11(BSC->-19->-2)100

6/22/03 20:00-11(BSC->-19->-0)91-11(BSC->-19->-2)75

6/22/03 20:15-11(BSC->-19->-0)88-11(BSC->-19->-2)94

6/22/03 20:30-11(BSC->-19->-0)86-11(BSC->-19->-2)84

6/22/03 20:45-11(BSC->-19->-0)89-11(BSC->-19->-2)74

6/22/03 21:00-11(BSC->-19->-0)96-11(BSC->-19->-2)68

6/22/03 21:15-11(BSC->-19->-0)100-11(BSC->-19->-2)91

6/22/03 21:30-11(BSC->-19->-0)89-11(BSC->-19->-2)91

6/22/03 21:45-11(BSC->-19->-0)100-11(BSC->-19->-2)75

6/24/03 20:00-11(BSC->-19->-0)93-11(BSC->-19->-2)80

6/24/03 20:15-11(BSC->-19->-0)83-11(BSC->-19->-2)87

6/24/03 20:30-11(BSC->-19->-0)80-11(BSC->-19->-2)76

6/24/03 20:45-11(BSC->-19->-0)89-11(BSC->-19->-2)79

6/24/03 21:00-11(BSC->-19->-0)94-11(BSC->-19->-2)92

6/24/03 21:15-11(BSC->-19->-0)82-11(BSC->-19->-2)73

6/24/03 21:30-11(BSC->-19->-0)86-11(BSC->-19->-2)89

6/24/03 21:45-11(BSC->-19->-0)93-11(BSC->-19->-2)77

Sheet1

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

00

0

2

/15

Sheet2

Sheet3

HUAWEI

Case Review 8Analysis of Alarm and its CauseThe over excited alarm of carrier power occurs frequently.The forward power is 43dBm at the end of a 15-minute period according to traffic measurement. The power amplifier may transmit at the maximum power within a short time. Since users are distributed on the cell border, the sector has to consume more power. As a result, the load of the sector is too high. Since the traffic of adjacent BTS N is comparatively small and the traffic can be distributed through load balancing. We can decrease the power of common channel in the sector according to analysis of drive test, which will not affect system performance, especially the coverage performance.

HUAWEI

Case Review 8We do not give much research on data service optimization due to site problem, so the problems found in this aspect are not too much. The following is a case of data service optimization. Case: The mean download rate is too low and transmission gap is comparatively large during test of data service optimization in area J. The details are as follows: the mean download rate of serial port cable in forward direction is 6 kbyte and the transmission gap is obvious.

HUAWEI

Case Review 9We will introduce RLPTCP before analyzing the case.The RLP detects forward data in buffer every 20ms and requires retransmission frames.If forward SCH does not exist, the data size in buffer exceeds the SCH_LOCK_ THRESHOLD or retransmission frames required by RLP exceeds REXMIT_FRAME_THRESHOLD, RLP reports MSG _SDU _RRM_SCH_APPLY_IND to request for SCH.If forward SCH exists, the data size in buffer is less than the SCH_LOCK_ THRESHOLD or retransmission frames required by RLP are less than REXMIT_FRAME_THRESHOLD within certain time, RLP reports MSG_SDU_RRM_SCH_STOP_IND to inform RRM of terminating SCH request.To avoid frequent requests for SCH by RLP, set the interval between 2 SCH requests to 10 frames

HUAWEI

Case Review 9We will introduce RLPTCP before analyzing the case.The RLP detects forward data in buffer every 20ms and requires retransmission frames.If forward SCH does not exist, the data size in buffer exceeds the SCH_LOCK_ THRESHOLD or retransmission frames required by RLP exceeds REXMIT_FRAME_THRESHOLD, RLP reports MSG _SDU _RRM_SCH_APPLY_IND to request for SCH.If forward SCH exists, the data size in buffer is less than the SCH_LOCK_ THRESHOLD or retransmission frames required by RLP are less than REXMIT_FRAME_THRESHOLD within certain time, RLP reports MSG_SDU_RRM_SCH_STOP_IND to inform RRM of terminating SCH request.To avoid frequent requests for SCH by RLP, set the interval between 2 SCH requests to 10 frames

HUAWEI

Case Review 9 Therefore, 2 possible defects in RLP request mode:Incoming Packet triggers the SCH request of RLP. If no packet is received, RLP cannot request for SCH due to data existence in buffer.PPU sends a data packet to RLP every 10 frames. RLP checks for SCH request every 10 frames. This may cause the delay between 2 SCHs requests: 10 frames + signaling delay= 20 frames, affecting the transmit efficiency.The problem may be caused by long delay of requesting SCH according to transmission figure. The SCH can be reestablished to transmit data some time later after disconnection.

HUAWEI

Case Review 9Solution:The long interval of requesting SCH by RLP causes the large SCH gap. The retry interval of requesting SCH is 200 frames (4s). It is recommended to set the parameter to 20 frames as shown below. Then the problem is solved.

HUAWEI

Case Review 9 Sum-up: In the data service optimization, we should:Master signaling procedures of data service and know the meaning of every parameter.Utilize capture signaling.Analyzing the signaling correctly.Then we can find corresponding solutions to problems immediately.

HUAWEI

ContentsChapter 1 Planning Cases

Chapter 2 Optimization Cases

Chapter 3 Terminal Cases

HUAWEI

Chapter 3 Terminal CasesDifficulties and emphasesCase review

HUAWEI

Difficulties and Emphases The tool of plan optimization and final service object, the terminal plays an important role in our work. Therefore, the terminal failure will bring great inconvenience to our work. Generally, the problems of terminal are as follows: Incorrect configuration of terminal.Incorrect configuration of terminal causes failure of connection with software.Terminal defects

HUAWEI

Case Review 1 Case: Since there are lots of parameters to be configured in MS, incorrect setting of one of parameters will lead to service provision failure of MS. In normal cases, we should pay attention to following parameter configurations: (Take SAMSUNG MS for example):NAM 1 LKout affects conversation.NMSID affects conversation.FOR_SID_REG and FOR_NID_REG affect roaming. The above parameter configuration is mentioned in user manual of terminal configuration. The problem can be avoided if we understand the meaning of every parameter.

HUAWEI

Case Review 2 Case: The incorrect parameter configuration will lead to communication failure of debugging software or drive test software. See the following:The incorrect port rate setting of Kyocera 2235 will lead to frequency setting failure through Mbtest. (To set the port rate to 115.2 kbps) Incorrect rate will lead to failure of online drive test between SAMSUNG X250 and PC (The rate of Ds Baud and Diag Baud must be set to 115200).

HUAWEI

Case Review 3 Case: Since the problem caused by terminal defects will lead to serious fault, we should be careful during problem location The networking structure of Huawei ETS450D in X city is 1 S111 BTS (Frequency 210), and 1 S222 BTS (Frequency 210, Frequency 260). The auxiliary hard handoff is used between frequency 210 and frequency 260. After the handoff parameter is correctly configured, the 2 out of 3 types of MSs can perform handoff normally in network application. The handoff of the last type of MS fails in any event.Perform user interface tracing during handoff test according to IMSI of the MS. Although there is HDM message during handoff, MS does not respond correctly. Since the cause value of CLEAR COMMAND in interface A is Equipment fail, the problem is caused by handoff failure of MS.

HUAWEI

Case Review 3 Conclusion: The BUG exists in parameters of ETS668 terminal software related to hard handoff after the confirmation by the supplier. The parameters are designed for 800M instead of 450M, so the problem occurs.

HUAWEI

Sum-up From the above cases, we can find hundreds of problems in the network and their causes vary during optimization. We should master the methods to locate problems instead of remembering their symptoms. That is, we should utilize tracing tools to analyze the large number of tracing data and compare them with that in the normal procedures to locate the fault.We should work out network planning under the guide of optimization experience so that we can foresee the optimization result from network planning. Then we can solve the problem occurring frequently at planning stage.

HUAWEI

How to handle such a problem?Discovering faults: Generally, the customer discovers the faults when some network problem occurs, such as subscribers paging failure and rising call drop complaints. The methods for locating the causes are as follows:1. Check the cell with serious call drop ratio according to the traffic measurement. Focus on those cells if such a cell is found.2. Perform signaling tracing and debugging console tracing.3. Make large number of dialing tests to capture the complaints. Locate the causes with signaling tracing and debugging console tracing.Reverse search windows of BTSMS can be switched from one BTS to another by adjusting forward search window. Sometimes we need to adjust the acquiring window of traffic channels the BTS to enlarge the influence area. The principle of the parameter is the same as that of MS except that this parameter is considered from the viewpoint of BTS. The difference of searching algorithm between MS and BTS lies in the following: all common references of BTS are based on GPS clock, which means that the search window center of BTS is always 0 and width is defined by search window.There are two parameters involved: one is the BTS radius, which is used to detect the information of MS-originated calls. The other is the size of traffic channel obtaining window used by the BTS to detect MS handoff. BTS will not detect the request originated by MS beyond the radius of BTS. The BTS also cannot detect MS handoff originated outside the traffic channel of TchAcqWinSz. The BTS window size is defined by round trip delay (RTD). The RTD in Huawei is used for hard handoff.Since the causes of problem on site are indirect, we should exclude possible causes. We should conform to the optimization procedures when analyzing causes for all problems.

The collected data includes not only traffic measurement and drive test data, but also signaling tracing data, printed information of debugging console and data of interference test. We have to master the methods for obtaining and analyzing data to cope with different situations. The offset power of both varies according to different radio configurations. The access channel in the case is IS95 access channel and the offset value is 73, while the traffic channel is IS2000 traffic channel and the offset value is 81.5, therefore, 8.5dB is reduced in transmit power when MS accesses traffic channel. The difference can be compensated by RLGAIN_ADJ. The correction range is 015, which corresponds to 87dB. The value obtained after query of DB is 8, that is, 0dB is corrected.The figure shows that the timer functions after it is added and this is a negative case. Since there are various methods for adjustment with negative effects, we should balance advantages and disadvantages of the methods to be selected.Take the calling MS for an example to describe the time when two timers function.Since the longitude and latitude of BTS are not configured in the system, the problem occurs. The adjacent cell distance is set to 10 km in the system. System can add adjacent cell automatically. System can select correct adjacent cell among precise configuration of the longitude and latitude of each BTS. 1. Since it is call drop upon handoff, the problem may exist in neighbor list configuration. The call drop occurs after normal handoff from BTS A to BTS B is completed and Eclo does not decrease markedly, so the problem of neighbor list configuration can be excluded. 2. To judge whether the problem caused by certain BTS, keep MS in the handoff area and block BTSs A and B respectively. The call drop does not occur when BTS A or B serves MS alone. Therefore, the problem caused by BTS can be excluded. 3. Judge whether the problem is caused by clock. Since the call drop is mainly caused because two BTSs do not synchronize, keep the two MSs in the BTS A and BTS B respectively, restart the MSs, and then compare the time of MSs. Since the time of MSs is synchronous, the synchronization problem can be excluded. (The time of H100 MS is not precise, it is difficult to judge whether they are synchronous. Another simple but effective method is to originate a Markov call in BTSs that may be asynchronous. The call drop occurs in 40s and debugging console tracing generates lots of Tch_Error. 4. Since BTS A is level-3 cascaded BTS using IMA transfer mode, the problem may be caused by short duration of timer in reverse frame combination of FMR board. The D302 version is used and default duration is 18ms, so the problem caused by reverse frame combination timer can be excluded. 5. The Tch_Error message reported by FMR to CCM causes call drop according to the print message of FMR tracing. The cause value is 04, which results in many idle frames. During processing of multiple soft handoff branches, FMR activates a combination timer when receiving the first signal sent to the branch in 20 ms. FMR combines the multiple branch signals received before the combination timer expires. (Under normal circumstance, these signals are the frames with the length of 20s) Then, FMR sends the best frame to upper layer. Since the transfer delay of multiple branches may differ with each other, (some may be longer than the duration of combination timer.) FMR cannot combine multiple signals correctly and idle frame is thought the cause for call drop. Since we exclude the problem of reverse frame combination timer, the cause of call drop should be in level-3 cascaded BTS A, using IMA transfer mode. The delay of IMA is not fixed, which may fluctuate between 3ms20ms leading to the call drop.The problem in transfer mode is hard to be located unless we are familiar with product and project during network optimization.This is a software problem and should be mitigated. (The problem of network adapter in data service.) Improvement: 1. Trace the network access process of the network adapter through M2000 user interface tracing. The status of the network adapter transfers from Active to Dormant normally if it is idle for some time.2. Then the terminal sends a request for data service. Network side sends a Paging Request to terminal and it returns a Paging Response while MSC delivers a clear command.3. Check the Paging Reponse sent by terminal. Then we find the field carried in Service Option is 0 (For the signaling field, refer to attachment)Understand meaning of Service Option: If paging response message is required to carry service option, fill the value of service option designated by MS. If the service option field is not required, fill the default value 000000000000001. If MS does not support the service option in the paging message, it selects other option according to descriptions in protocol. If MS does not support the service option in Paging message and no other option is selected, fill it with 000000000000000, which means that MS rejects service option carried in paging message.Therefore, if the value of service option is 0, it means that the terminal rejects the service option delivered by MSC and other service option is not selected. Then MSC releases link directly and users switch to the Idle state, so the redialing is required.4. We can conclude that the network adapter is not consistent with the protocol requirement. We can explain this to the customer and coordinate with manufacturer of the network adapter to urge the modification.5. It takes some time for us to coordinate with manufacturer of the network adapter. But the customer requires an immediate solution to the problem. To reduce customer complaints, we check the setting network deactivation timer, and find that the timer duration is set to 20, that is, the terminal switches to the Dormant state if no data is transmitted within 20s. We should set the duration of the timer to 180s as required by customers so as to mitigate the probability of the problem temporarily. Meanwhile, the manufacturer of the network adapter should upgrade the network adapter immediately. [Cause Analysis] Generally, there is some time during which terminal dose not download any data (for example, the user is browsing the web page) when user accesses network. If the time exceeds the duration set for deactivation timer, the user switches to the Dormant state. Then the system will disconnect air interface and reserve PPP connection. System establishes the air interface connection when user originates a service request to enter the active state. The user in idle state may switch to the Idle state instead of active state when entering the Dormant state. Then network connection must be established again. Note:The 4956 in the horizontal ordinate indicates a 15-minute measurement data of the forward load of BTS at busy hour after adjustment.Carrier power over excited alarm often occurs [Processing procedure] The specific adjustment plan is as follows:1. Increase tilt angle of the sector to reduce coverage area.2. Decrease the power of pilot channel; synchronous channel and paging channel and the maximum transmit power by 1dB3. Reduce the tilt angle of sector adjacent to that of BTS N to widen the coverage area. (For the specific figure, refer to attachment.)[Cause Analysis]: Forward load denotes the consumption of forward power. The consumption of forward power is closely related to traffic and users distribution. There is a stark difference between the forward transmit power consumption near the BTS and on the cell coverage border for the traffic with the same Ireland. The forward power is 43dBm at end of a 15-minute period according to traffic measurement. This does not mean that the average output power within 15 minutes is 43 dBm, instead, the power amplifier only transmits at maximum power within a short time. Huawei output power has a protection over the 2dB. The carrier can still function and hardware will not be damaged even if the load reaches 100%.The traffic of BTS N adjacent to the sector is not very heavy and traffic can be distributed through load balancing. We can decrease the power of common channel in the sector by 1dB according to analysis of drive test, which will not affect system performance, especially the coverage performance.Huawei output power has a protection over 2dB. The carrier can still function and hardware will not be damaged even if the load reaches 100%.Max Throughput=12639.846 bytes/s=12639.8468/1000=101.12kbpsAverage Throughput=4682.733bytes/s=4682.7338/1000=37.46kbpsMaximum interval of SCH request: 20.88sAverage interval of SCH request: 7.35sMaximum signaling delay = 90 ms (4.5 frames) ( From the time when RRM receives SCH request to the time when SCH preparation completes at network side)Average signaling delay=70 ms (3.5 frames)The SCH requesting mode in current version is as follows:PPU checks whether data packet is transmitted to RLP every 10 frames.RLP checks whether SCH request is needed upon receiving PPU packet.The requirement on request: Buffer data size > 1KB and Last SCH request delay is over 30 framesWhen RRM receives an SCH request, it will perform allocation if SCH is not exist. Otherwise it clears request.Two defects in above mode:Incoming Packet triggers the SCH request of RLP. If no packet is received, RLP cannot apply for SCH due to data existence in buffer.PPU sends a data packet to RLP every 10 frames. RLP checks for SCH request every 10 frames. That may cause the delay between 2 SCHs requests: 10 frames + signaling delay= 20 frames, affecting the transmit efficiency.Parameters:RLP stops SCH persistence test at every 25 frames. Currently RLP perform check at every frame if the data in persistence test in 25 consecutive frames is lower than RLP Disables SCH Threshold, the RLP requests for release.RLP Disables SCH Threshold: 500 byte RLP Requests SCH Threshold: 1kByte. When the data size in buffer of RLP is over the threshold, RLP sends an SCH request in unit of byte.Interval between 2 SCH requests sent by RLP: 10 frames (the interval between last request and the next request) Avoid requesting SCH frequently. The unit is frame.Check for retransmission frames required by RLP: RLP sends request to check data frames in retransmission queue. RLP can still send request when data frames is over threshold.Time for conversion from active to DORMENT state: the unit is second.Buffer size for each user: 50kBytes.