UMTS Radio Network Optimization Proposal(for New Network)_R1.0

Operator Logo

UMTS Radio Network Optimization Proposal R1.0 (For xxx new network)

UMTS Radio Network Optimization Proposal

LEGAL INFORMATION By accepting this certain document of ZTE CORPORATION you agree to the following terms. If you do not agree to the following terms, please notice that you are not allowed to use this document. Copyright © 2014 ZTE CORPORATION. Any rights not expressly granted herein are reserved. This document contains proprietary information of ZTE CORPORATION. Any reproduction, transfer, distribution, use or disclosure of this document or any portion of this document, in any form by any means, without the prior written consent of ZTE CORPORATION is prohibited. and are registered trademarks of ZTE CORPORATION. ZTE’s company name, logo and product names referenced herein are either trademarks or registered trademarks of ZTE CORPORATION. Other product and company names mentioned herein may be trademarks or trade names of their respective owners. Without the prior written consent of ZTE CORPORATION or the third party owner thereof, anyone’s access to this document should not be construed as granting, by implication, estopped or otherwise, any license or right to use any marks appearing in the document. The design of this product complies with requirements of environmental protection and personal security. This product shall be stored, used or discarded in accordance with product manual, relevant contract or laws and regulations in relevant country (countries). This document is provided “as is” and “as available”. Information contained in this document is subject to continuous update without further notice due to improvement and update of ZTE CORPORATION’s products and technologies.

ZTE CORPORATION Address:

NO. 55 Hi-tech Road South ShenZhen P.R.China 518057

Website:

http://dms.zte.com.cn (Technical Support)

Email:

[email protected]

ZTE Confidential Proprietary

© 2014 ZTE CORPORATION. CORPORATION. All rights reserved.

I


LEGAL INFORMATION By accepting this certain document of ZTE CORPORATION you agree to the following terms. If you do not agree to the following terms, please notice that you are not allowed to use this document. Copyright © 2014 ZTE CORPORATION. Any rights not expressly granted herein are reserved. This document contains proprietary information of ZTE CORPORATION. Any reproduction, transfer, distribution, use or disclosure of this document or any portion of this document, in any form by any means, without the prior written consent of ZTE CORPORATION is prohibited. and are registered trademarks of ZTE CORPORATION. ZTE’s company name, logo and product names referenced herein are either trademarks or registered trademarks of ZTE CORPORATION. Other product and company names mentioned herein may be trademarks or trade names of their respective owners. Without the prior written consent of ZTE CORPORATION or the third party owner thereof, anyone’s access to this document should not be construed as granting, by implication, estopped or otherwise, any license or right to use any marks appearing in the document. The design of this product complies with requirements of environmental protection and personal security. This product shall be stored, used or discarded in accordance with product manual, relevant contract or laws and regulations in relevant country (countries). This document is provided “as is” and “as available”. Information contained in this document is subject to continuous update without further notice due to improvement and update of ZTE CORPORATION’s products and technologies.

ZTE CORPORATION Address:

NO. 55 Hi-tech Road South ShenZhen P.R.China 518057

Website:

http://dms.zte.com.cn (Technical Support)

Email:

[email protected]



I


Revision History Product Version

Document Version

Serial Number

1.0

Reason for Revision First published

Author Document Version

Date 2011-11-30 2011-11-30

1.0


Prepared by

Reviewed by

Approved by

Mawei


II


Applicable to: UMTS network optimization engineers

Proposal: Before reading this document, you had better have the following knowledge and skills. SEQ 1

Knowledge and skills Null

Reference material Null

2 3

Follow-up document: After reading this document, you may need the following information SEQ 1

Reference material Null

Information Null

2 3



III


About This Document Summary Chapter

Description

1

Introduction

Introduction

2

Scope

Scope

3

RF Design

RNC/LAC/RAC/SAC and scrambling code/neighbor cell planning

4

Optimization Process

Optimization Process

5 Optimization definition for different network implementation phases

Optimization definition for different network implementation phases

6

Organization Structure

Organization Structure

7

Optimization Analysis Method


8 Optimization and diagnostics tools


Optimization and diagnostics tools


IV


TABLE OF CONTENTS 1

Introduction..................................................................................................... 1

2

Scope............................................................................................................... 1

3 3.1 3.2 3.3 3.4 3.5

RF Design........................................................................................................ 2 RNC Planning................................................................................................... 2 LAC/RAC planning............................................................................................ 3 SAC Planning ................................................................................................... 3 Scrambling code planning................................................................................. 4 Neighbor Cell Planning ..................................................................................... 6

4

Optimization Process ..................................................................................... 8

5 5.1 5.1.1 5.1.2 5.1.3 5.2 5.3

Optimization definition for different network implementation phases........ 9 Pre-launch Optimization ................................................................................... 9 Single Site Verification (SSV) ........................................................................... 9 Cluster Optimization ....................................................................................... 10 Whole Network Optimization ........................................................................... 13 Soft Launch (Trial-Running Period) Optimization ............................................ 14 Post Launch Optimization ............................................................................... 15

6

Organization Structure ................................................................................. 15

7 7.1 7.2 7.2.1 7.2.2 7.2.3 7.2.4 7.2.5 7.2.6 7.3 7.3.1 7.3.2 7.3.3 7.4 7.4.1 7.4.2 7.4.3 7.4.4 7.5 7.6

Optimization Analysis Method ..................................................................... 16 Optimization Target and Description ............................................................... 16 Optimization Based On Driver Test/Walk Test Data ........................................ 18 RF coverage optimization ............................................................................... 18 Call Failure Analysis ....................................................................................... 19 Call Drop Analysis .......................................................................................... 23 Handover Failure Analysis .............................................................................. 25 High Access Latency ...................................................................................... 25 Low Data Throughput ..................................................................................... 26 Optimization Based On OMC Performance Data ............................................ 27 Overall optimization process ........................................................................... 27 Call Setup Success Rate Optimization............................................................ 30 Call Drop Rate Optimization ........................................................................... 31 Optimization in Network Operational Phase .................................................... 32 Traffic and RTWP (Received Total Wideband Power) monitoring ................... 32 Code resource utilization monitoring ............................................................... 32 Transmission utilization rate and RLC retransmission rate monitoring ............ 33 Neighbour Cell Optimizaiton ........................................................................... 33 Optimization Based On Capacity .................................................................... 33 Optimization for VIP needs ............................................................................. 35

8 8.1 8.2 8.3

Optimization and diagnostics tools............................................................. 35 Drive test and Analyzing tool .......................................................................... 35 ZXPOS-CNO .................................................................................................. 44 Detected set neighbor report analysis ............................................................. 47


© 2014 ZTE CORPORATION. All rights reserved.

V


8.4

Signaling trace ................................................................................................ 49



VI


FIGURES Figure 3-1

Illustration of Neighbor Cell Planning .................................................................. 7

Figure 4-1

Optimization Milestone ........................................................................................ 8

Figure 5-1

Cluster Optimization Work Process ................................................................... 11

Figure 5-2

Cluster definition ............................................................................................... 12

Figure 6-1

Optimization Organization Structure.................................................................. 16

Figure 7-1

RF optimization based on coverage .................................................................. 18

Figure 7-2

Call Failure Analysis Process ............................................................................ 20

Figure 7-3

Paging Problem Troubleshooting Process ........................................................ 21

Figure 7-4 RRC Connection Setup Problem Troubleshooting Process ............................... 22 Figure 7-5

Overall optimization process ............................................................................. 30

Figure 7-6

Call Setup Success Rate Analysis .................................................................... 31

Figure 7-7

Optimization Based On Capacity....................................................................... 34

TABLES Table 3-1

RNC division results ............................................................................................. 2

Table 5-1

Single Site Verification ......................................................................................... 9



VII


1

Introduction The purpose of this document is to present the solution of UMTS radio network optimization of ZTE for the Operator in 20xx, and make certain guidelines for the activities involved in UMTS radio network optimization. Network optimization is to verify whether the network design, installation, integration and the corresponding network configuration have been implemented correctly, and whether the implemented design is consistent with the planning. Through optimization, the network performance will meet the operator ’s requirement. The core problems involved in optimization are input and output definition, resource planning and working schedule as well as work process definition, the tools and methodologies used in the optimization tasks. During each stage of radio network optimization, related project staffs should check this document for information like working scopes, plans, principles, procedures, tools, resources and suggested troubleshooting methods, so that coordinated working results can be expected. Notes: Currently, because this document is still under developing, the following is only for reference. In the project, the engineers should make some content adjustments according to the actual project conditions.

2

Scope From view point of project management and technical problems solving, this document describes how important UMTS RF parameters are designed and how radio network optimization tasks should be performed in different stages of targeted project. Topics are merely concentrated on some important RF parameters planning and optimization activities, coverage and capacity planning related work and acceptance test related details are not included. Also, optimization works not related to RAN part of network are not covered. 

RNC/LAC/RAC/SAC planning



Scrambling code/Neighbor cell planning



Optimization process



Optimization analysis method



Resource for optimization



Optimization and diagnostics tools



1


3

RF Design RNC/LAC/RAC/SAC planning, scrambling code planning and neighbor cell planning will be introduced in this section, but coverage and capacity planning related work are not included. Notes: In the project, the detailed RNC/LAC/RAC/SAC design and scrambling code/ neighbor cell planning should be carried out based on actual network information.

3.1

RNC Planning The principles of RNC planning are as follows: 

The NodeBs managed by one RNC are geographically centralized to avoid geographically overlapping with NodeBs managed by other RNCs, thus reducing the number of unnecessary cross-RNC handovers.



Avoid planning the border area between RNCs in densely populated areas or along highways. When performing RNC border planning, avoid deploying RNC border along traffic arteries including highways and railways so as to prevent ping-pong handover between RNCs due to the movement of UEs. Furthermore, inter-RNC handovers are hard handovers or cross-lur soft handovers with success rate lower than that of intra-RNC soft handovers. Therefore, avoid deploying RNC border in densely populated areas so as to decrease inter-RNC handovers and reduce the signaling load arising thereof.



It is recommended that RNC and the switching office are co-located so as to simplify lu interface connection.



As for the RNC coverage planning, the factors that need to be taken into account include the number of NodeBs and cells supported by RNC as well as the CS traffic, PS traffic and signaling handling capability.

According to above principles and RNC dimensioning analysis, RNC division results of the operator network are as follows: Table 3-1

RNC division results

RNC ID


Location

Number of NodeBs


Number of Cells

2


3.2

LAC/RAC planning The Location Area Code (LAC) and Router Area Code (RAC) are two parameters used to indicate UE paging location. The LAC is a part of the L AI (LAI = MCC + MNC + LAC). The paging request of the CS domain is sent through the LAI. The RAC is a part of the RAI (RAI = PLMN-ID + LAC + RAC). The paging request of the PS domain is sent through the RAI. The numbering rules of RAI show that the RACs are numbered within the LAC. Therefore, an LAC may contain several RACs and an RA cannot span more than one LA. The principles of LAC/RAC planning are as follows: 

The LAC/RAC cannot be too large. When a UE is paged, CN will send a paging request to all UEs in the area in relation to the LAC/RAC through RNC. One LAC/RAC may contain scores and even hundreds of cells, thus resulting in incredibly high paging traffic to RNC. NodeB has to send the paging request to UEs on a limited number of PCHs. Therefore, large LAC/RAC may result in NodeB paging overload and eventually the signaling congestion and loss of paging information.



The LAC/RAC cannot be too small. If on the other hand, the LAC/RAC is too small, there will be a large number of LAC/RAC borders and UEs on these borders are very prone to frequent LA/RA update between LAs/RAs. When there is a paging message incoming to a UE which happens to quickly move to another LA/RA and initiates the LA/RA update procedure, the UE will not receive the paging message sent to the original LA/RA, thus resulting in call connection failure.

3.3



It is recommended to plan the LAC/RAC border in areas with low user mobility and small traffic.



When performing LAC/RAC border planning, avoid deploying LAC/RAC border along traffic arteries including highways and railways so as to prevent ping-pong LAC/RAC update due to the movement of UEs. On the other hand, avoid planning the LAC/RAC border in densely populated areas due to the huge number of location updates on the LAC/RAC border so as to reduce the signaling load arising thereof

SAC Planning 1.

SAC definition for the 3G network



3


SAC is Service Area Code, the only identification used to one cell or more than one cell belonging to the same LAC. SAC’s main purpose is to designate UE to CN. Generally one cell can have one or two SACs. One, called SACPC, is belonging to SAC for both PS and CS domain. The other SAC, called SACBC, is belonging to BC domain. Each cell must have one SACPC for PS and CS domain. 2.

SAC planning principle for the network

Related SAC parameters have SACPC, SACB and SACCPre, Each parameter planning principle is shown below: SACPC: SAC is Serving Area Code, uniquely identifies one cell or several cells in one location area. It is used to locate UE for CN. One cell has one or two SAC, one belongs to CS +PS domain, and another belongs to BC domain. One cell has to own one SAC belongs to CS +PS domain, which is SACPC. The value of cell’s serving area code belongs to CS+PS domain is corresponding to RAC. SACB SACB is Serving Area Code belong to BC domain; its default value is 0. SACBPre: (Service Area Code for BC Domain, current configuration is 0) SACB Configuration Tag represents whether SACB is configured; its present default value is 0.

3.4

Scrambling code planning Downlink scrambling code planning is a process of distributing 512 groups of primary scrambling codes to various cells. Several rules need to be followed during scrambling code planning: 1

Reserve sufficient distances for geographical isolation of PSC reuse sites during the reuse of the same scrambling code. The isolation distance required for scrambling code reuse is relevant to the radio environment, and it is used to ensure that the identical scrambling code signal will not be received at the same place and thus to prevent scrambling code confusion. The longer the scrambling code reuse distance, the smaller the scrambling code confusion probability, but the scrambling codes may be insufficient for distribution. On the contrary, the shorter the scrambling code reuse distance, the greater the scrambling code confusion probability, but the scrambling codes are sufficient for



4


distribution. The scrambling code reuse distance can be determined based on your experience or onsite test result. 2

The same scrambling code shall not be distributed to the primary cell and the neighbor cell of one of the primary cell’s neighbor cells. For instance, assume that A is a primary cell, B is a neighbor cell of A, and C is a neighbor cell of B (C is known as the L2 neighbor cell), then A and C must not be distributed with the same scrambling code; otherwise, two cells with the same scrambling code will appear in the neighbor cell list of B, and the scrambling code will be mixed up during the handover from B to A or C, which will affect the handover success rate and service quality. For another instance, assume that A and B are neighbor cells and they both stay in an active set, C is another neighbor cell of A, and D is another neighbor of B, then C and D must not be distributed with the same scrambling code. When both A and B stay in the same active set, the neighbor cell of A and that of B will be merged to form a new neighbor cell list to be distributed to UEs. In this case, C and D will probably be merged into the same neighbor cell list. If C and D have the same scrambling code, their scrambling code will be mixed up during the handover from A or B to C or D, which will affect the handover success rate and service quality. Here, C is known as the L3 neighbor cell of B, and D is known as the L3 neighbor cell of A. Considering the two cases above, try to void using the same scrambling code in L2 and L3 neighbor cells, especially not in those cells within a certain range (e.g. the L2 and L3 neighbor cells within a radius of 20km from the primary cell) during scrambling code planning; otherwise, the scrambling code may easily get mixed up during handover decision. Note: The detection of scrambling codes of neighbor cells is implemented through the CNO developed by ZTE. At present, ZTE follows the scrambling code planning procedure below:

3

i

Perform scrambling code planning by using CNO. Take into account the reuse distance instead of neighbor cell relationship during planning.

ii

Check planning results by using CNO based on the neighbor cell relationship or the reuse distance. For details, see the description of scrambling optimization function of CNO in Section 8.2.

The cells with different carriers in the same sector can be configured with identical scrambling code in order to simplify the scrambling code planning and optimization.

The UMTS scrambling code planning of the operator is carried out by using CNO practically, a scrambling code planning tool developed by ZTE. For details, see the description of scrambling optimization function of CNO in Section 8.2.



5


3.5

Neighbor Cell Planning 1.

Intra-system neighbor cell planning

Principle of neighbor cell planning mainly includes neighbor cell number, geographic topological structure, coverage interference, etc. Because the maximum number of neighbor cell list within a system is regulated as 31 in protocol, neighbor cell lists will be combined in soft handover status; we need to control neighbor cell number in configuration and try to minimize it as much as possible in the precondition of insuring the successful handover. Whether neighbor cell configuration is reasonable impacts handover between sites; the initial neighbor cell list formed in system design phase is set according to the following mode, then neighbor cell list will be adjusted according to handover times after the system is commissioned. Cells of one same site have to be set as each other ’s neighbor cell; the first layer and the second layer cells can be chosen as the present cell ’s neighbor cell according to present cell’s coverage (shown in the following figure). The second layer cells on the same direction of the present sector are set as its neighbor cells, the first layer cells on the opposite direction of the present sector are also set as its neighbor cells. The following is an example of neighbor cell setting shown in the following figure. The red one is the present cell whose scrambling codes are set as 4, 8 and 12; those cells formed in boldfaced broken lines are present sector ’s neighbor cells. Pink ones are the first layer cells; blue ones are the second layer cells.



6


neighbor cell configuration method in UMTS system, but the situation of neighbor cell traffic congestion shall also be considered at the same time. For the interoperation between 2G and 3G system, besides configuring GSM cell information in UMTS system, UMTS cell information and neighbor cell information should also be configured in GSM system that requires coope ration with GSM equipment vendor. Provided parameters vary with vendors 3.

Inter-RNC Neighbor Cell Planning

Inter-RNC handover is hard handover; besides neighbor cell needs to be configured, corresponding cell information should also be configured. The UMTS neighbor cell planning of the operator is carried out by using CNO practically, a neighbor cell planning tool developed by ZTE. For details, see the description of scrambling optimization function of CNO in Section 8.2.

4

Optimization Process Radio network optimization consists of three major stages generally: Pre-Launch Optimization, Soft Launch Optimization (optional) and Post Launch Optimization. Figure 4-1

Optimization Milestone Post Launch Optimization Network Commercial

Soft Launch Optimization Network Soft Launch Pre-Launch Optimization

Whole Network Optimization Cluster Optimization

Single Site Verification Network Construction Installation & Commissioning & Test Network Design Site Survey & Planning

Start

1.

2.

Commissioning

Network Design

PAC

End

FAC

Pre-Launch: The main objective of Pre-Launch Optimization is to control RF network air interference, assure network hardware functionality work norm ally, and ensure the KPI target of Preliminary Acceptance Test is achieved. Pre-Launch Optimization inludes following three steps: 

Single Site Verification



Cluster Optimization



Whole Network Optimization

Soft Launch (optional): There could be a several months trial period from the date of issuing PAC. The optimization in this period is named as Soft Launch Optimization.



8


The objective of Soft Launch Optim ization is to assure that no Punch List item s exists in the System. The Punch List is the list that consists of all defects identified during the respective Preliminary Acceptance Test, during the period prior to Final Acceptance. When all items on the respective Punch List have been resolved in the System, a Final Acceptance Certificate will be issued. 3.

Post Launch: The optimizaiton after issuing FAC is named as Post Lauch Optimization. The network can be put into commercial servies after FAC. The objective of Post Lauch Optimization is to assure the network performance stabilization when subscribes are increasing. Post Lauch Optimization focus on customer experiences, system load, capcity balance, resource utilization, etc.

5

Optimization definition for different network implementation phases

5.1

Pre-launch Optimization

5.1.1

Single Site Verification (SSV) The goal of SSV is to eliminate potential errors introduced during the site construction and configuration, so as to lay a reasonable basis for the following RF optimization, for it will be time-consuming to find out causes of unexpected results during optimization. Normally, functional requirements are the main concerns during the SSV; service performance of the single site is not strictly required. The check items involved in the SSV can be classified into several categories, such as equipment-related problems, engineering-related problems, and configuration-related problems. Typical problems are presented in the following table. These problems should be solved before the service-related SSV test, which involves the coverage test, voice call, video call, PS R99 download, HSDPA download specifically. Table 5-1

Single Site Verification

Equipment-related


Engineering-related

Configuration-related


9


Abnormal power alarm PA alarm Transmission broken Board-related alarms Internal/external link alarms Antenna VSWR alarm Abnormal RTWP alarm Clock source/GPS alarm Cell/Node B down alarm SW version alarm …

Swap feeder Loose connection of connectors Unreasonable antenna position Signal obstacle by buildings Wrong antenna tilt and azimuth …

Frequency Scrambling code LAC/RAC CPICH power Cell capability Cell status Transmission bandwidth CE configuration …

Above mentioned problems are to be solved by corresponding technical staffs. Most equipment-related problems are to be solved by RNS engineers; engineering- related problems by RF optimization engineers and installation engineers together; configuration-related problems by RF optimization engineers and OMC engineers. After site verification, obvious problems that can make the site incapable of being put on air should be eliminated. The SSV process is mainly based on the stationary check and drive test (DT). The former means performing desktop check on items according to configuration data, or walking around the site using test terminals. Following items are needed for the stationary check: 

Technical Site Survey (TSS) report



Planned Engineering Parameters



Planned Radio Parameters



Site Configuration Parameters.

These materials are also used in the DT verification of the site. The SSV report is the main output of this step. Besides, SSV engineers will propose suggestions for adjusting the site.

5.1.2

Cluster Optimization Cluster optimization mainly involves the coverage optimization, neighbor cell optimization, scrambling code optimization and solutions to service access failure, call drop, and handover failure, etc. Data collected from the DT and stationary test will be analyzed to locate problems, optimize the network and verify the adjusted schema. It is an iterative process to achieve cluster acceptance standards. The cluster optimization work flow is as follows:



10


Figure 5-1

Cluster Optimization Work Process

TSS/SSV Report

System Parameters

Engineering parameters

Digital Map

Preparation

Output Engineering Parameters

Initial Coverage Test

Adjusting Report

Radio Parameters Adjusting Report

Problem Analysis Cluster Optimization Report

Optimization Suggestion No

Acceptable? Execution

Yes

End Verification

No

Are problems solved? Yes

Submit Report

Following preparation needs to be made before the cluster optimization: 1.

Cluster Definition

The main rules of cluster definition are: 

The geographical location



The service distribution



The same RNC/LAC/RAC region information



There shouldn’t be too many sites in a cluster and there should be overlap between clusters.



11


Figure 5-2

2.

Cluster definition

Cluster Optimization Test Schedule

Generally if the ratio of on-air NodeB of one cluster is over 80%, the cluster can be optimized. The clusters can be optimized according to priorities. After the optimization suggestion is adopted, a new test schedule will be made to justify if it is effective. 3.

Testing Route Planning

Before cluster optimization, it is necessary to define the testing route. Continuous coverage is required along the testing route if not all of sites are on air. 4.

Network Parameters Checking

Before the cluster optimization, system parameters, such as NodeB ID, Cell ID, LAC, RAC, Scrambling codes, and Neighbor cell list, should be imported into the OMC. 5.

Optimization Method Definition

There are two optimization methods. One is RF parameters adjustment, such as antenna azimuth, down tilt, and height. The other is radio parameters adjustm ent, such as channel power allocation and handover parameters etc. 6.

Document Preparation

Following documents need to be prepared before the cluster optimization: 

Technical Site Survey report (TSS)



Single Site Verification report (SSV)



12




Site Engineering Parameters Table



OMC Configuration Parameters

7.

Optimization Equipments Preparation

Optimization equipment includes data collection software, post-processing and analysis software, test mobile phone, HSPA data card, Scanner, Test laptop, digital map, GPS, and Test vehicles.

5.1.3

Whole Network Optimization The whole network optimization, based on the cluster optimization, follows the same work flow as the cluster optimization. At this phase, optimization is first carried out in the area including several clusters, then in a RNC area, and at last in the whole network. The goal of whole network optimization is to verify the functionality of the future commercial network operation, such as the cell camping, cell reselection, and handover between different network layers. Main tasks of the whole network optimization include: 

Locating network problems which are not solved at the cluster optimization stage



Solving new problems



Making sure that the network offers continuous service by optimizing the boundary of clusters



Optimizing the hard handover



Ensuring KPI acceptance level

The whole network optimization is related to the system commercial strategy. After optimization, the network should be ready for commercial launch in terms of both function and performance. Every new commissioning site should be integrated into the cluster after optimization without degrading the overall performance. The input required for the whole network optimization is listed below: 

System commercial strategy description



Neighbor list configuration



Radio parameters configuration



TSS report and SSV report



Cluster optimization report



13


Test equipment used for the whole network optimization is the same as that for the cluster optimization.

5.2

Soft Launch (Trial-Running Period) Optimization When network construction and Pre-Launch optimization work are finished, the network can be put into a Soft Launch phase, which means friendly users with special access right can begin to use services provided by network and generate useful feedback for the enhancement of network performance. The goal of Soft Launch optimization is to further optimize the whole network in order to provide a continuous service experience in the majority of desired coverage area and assure that no Punch List items exists in the System. If any problems are detected in the Soft Launch optimization phase, they should be solved and checked thoroughly in the whole network. These problems might cover diverse areas such as the CN, RNC, NodeB, transmission, UE, etc. Improving the related KPIs to be commercial launch ready is the main purpose. The main target in Soft Launch Optimization stage is focused on coverage, neighbors, RRM parameters, and border area of clusters. Neighbours optimization mainly includes missing neighbour, unidirectional neighbour, inter-frequency neighbour, and inter-RAT neighbour. Handover related parameters should be optimized as well. Other RRM parameters such as access control, power configuration, load control, etc., should also be tuned selectively to meet the traffic requirements. The Soft Launch optimization is normally based on both drive test and friendly user feedback. As a supplementary data source, the signaling tracing is needed to help troubleshooting some inner system problems. Passive signaling monitoring equipments like Iu signaling tracing system can be used too, which might also be helpful when drilling down the identified problems. Although there is not much traffic statistics from trial users at this stage, the KPI report generated from the Soft Launched network is still helpful for problem analysis. At the same time, some optimization assistant tools, helpful at the early phase of the commercial network, based on OMC statistics, can also be used. For example, the system’s missing neighbor detecting function can be used to help optimize the wide intra-frequency neighbor list. The output of Soft Launch optimization is the performance optimization report for the whole network, and the complete set of tuned parameters for the forthcoming commercial launch. These parameters, which are to be further optimized in a dynamic process, serve as a good baseline for future improvement of the system performance.

Note:



14


Based on the different launch strategy from different custmer in the future, if immediately, cluster or whole network is launched after PAC, Pre-Launch optimization and Sof t Launch optimization should be combined to finish together during Pre-Launch optimization.

5.3

Post Launch Optimization Post Launch Optimization belongs to operation and maintenance optimization. The target in this phase is both the coverage and system performance from OMC statistics. Normally, after large subscribers register, the optimization goal is straight forward, that is, to keep stable and satisfactory end to en d system performance, and enhance the system KPI. The daily KPIs from OMC statistics should be monitored and optimized to designed level. As the main input for this optimization stage, OMC statistics and customer complaints are given higher priority than the drive test and walk test data, for after the commercial launch, the traffic data of the network is enough for providing detailed statistics on each KPI. The end to end performance monitoring result also matters at this stage. For example, it is conducive to problem drilling down, troubleshooting, KPI comparison, and cell traff ic load. The optimization process at this stage is mainly driven by KPI analysis result. For selected KPIs, daily analysis is made to keep up-to-date view on the dynamically changing network performance. If any problems are identified and classified into specific domains, corresponding teams from different domains are responsible for the troubleshooting work, and make possible testing, adjustments and verifications until the problematic KPIs fall into the acceptable level again. The output at this optimization stage includes daily and weekly KPI reports, as well as monthly performance test reports based on the drive test data. Typical or critical troubleshooting reports generated at this stage are documented, too.

6

Organization Structure The network optimization team will be divided into 3 functional roles: 1.

SSV engineers – The key role of these engineers are to conduct SSV field test according to predefined test schdeles, and analyzing of test files.

2.

Cluster&network optimization test engineers – The key role of these engineers are to conduct cluster&network DT test and perform post processed data analysis of drive test data, identify network optimization solutions such as parameter neighbor list changes, document such changes, monitor cluster improve throughout the optimization process and generate reports.

3.

Experts support engineers – These engineers are mainly ZTE RNC, NodeB R&D staffs. The key responsibility is to assists post processing and analysis engineers to troubleshoot hardware/software related issues, core network and transmission



15


issues. There are also key resources to identify network wide parameter changes required during the optimization process. Figure 6-1

Optimization Organization Structure

Optimization PM

ExPerts Team

P&O Experts Cluster & Network Optimization Team

SSV Team

RNS Experts Analysis Engineer

Analysis Engineer

Analysis Engineer

Analysis Engineer

Analysis Engineer

Analysis Engineer

Field Test Engineer

Field Test Engineer

Field Test Engineer

Drive Test Engineer

Drive Test Engineer

Drive Test Engineer

SSV Team 1

SSV Team 2

SSV Team 3

R&D Experts

Cluster & Network Cluster & Network Cluster & Network Optimization Team 1 Optimization Team 2 Optimization Team 3

Tools Experts

Notes: In the project, the detailed optimization resource and organization structure should be decided based on actual project conditions.

7


7.1

Optimization Target and Description During different phases of network optimization, the emphasized optimization targets might be a little different. Some target items need iterative tuning to become stable, while some are not so frequently tuned. Some adjustments come with costs on time and money, while others are easy to be implemented. Main optimization targets and their descriptions are listed as follows. 1) Network hardware / software / configuration faults When there are some major releases of network hardware or software, and the new version can bring more features or more stable network performance, and then upgrade of network is needed, this is some special kind of network optimization choice. Normally, more common optimization activities related to the software and hardware are configuration faults checking. When there are critical configuration errors, the network performance would be unacceptable, even network down can be seen. Sometimes configuration faults might cause a hidden problem only when the network develops to a certain stage, for example, the configuration of transmission and CE won ’t be bottleneck until the traffic increase to certain degree. This kind of configuration problem needs to be analyzed proactively. This can be achieved by correlating monitoring performance KPIs



16


such as Traffic (Erlang number of Call setup per RAB, Total Data throughput, Total TxPower per cell, and CE usage) with NodeB CE and transmission configurations. 2) Radio parameters Radio parameters cover nearly every aspect of the whole system, and they are commonly used to fine tune the network performance, especially for those RRM related radio parameters. Typical parameters include the channel power allocation parameters, the handover parameters, the access control parameters, the load control/congestion control/load sharing parameters, and so on. Please refer to RAN Feature Description for details on parameters. 3) Default parameter values From a statistical point of view, although there are many parameters can be changed, practically only a small part of them are frequently modified others are regarded as baseline parameters that have been optimized before the release of system hardware or software. But for special purposes and scenarios, they can be modified anyway. These half-fixed parameters are default ones, and their modification needs extensive experiments to prove the validity of the change. Once the change proposal is accepted, it is implemented in the baseline parameter set. 4) Site / Case specific parameter settings such as neighbor cell relationships, HO parameters and power settings These parameters are emphasized during the cluster optimization stage and they are often performed together with RF adjustments. Traditionally these optimization targets are most time-consuming part of the whole process, because it ’s hard to have unified parameter settings for all the sites, especially for the neighbor relationship optimization. With the help of CNO/CNA, some analysis process will be automated, e.g., the detected set neighbor reporting can help to find missing neighbors for each cell. 5) Antenna orientation ( in case of shared antennas, should take into account the effect on other networks sharing the same antenna ) RF related adjustments play important role in the whole optimization work, because they are proved to be the most effective way to complement the deficiency of network planning and implement a good coverage basis for the network. All optimization work related with radio parameters should base on a good coverage, or else the tuning can hardly get a satisfactory result. Here good coverage means not only the signal strength and Ec/Io, but network interference level in terms of pilot pollution, overshooting, etc. Antenna related tuning are key parts of RF optimization, including antenna type, azimuth, down tilt, even antenna height when possible. When the antenna system is shared by multiple operators or radio systems, the influence of tuning to other system should be carefully considered. With antenna types that down tilt can be tuned standalone, the influence is negligible, but for antenna azimuth, the tuning can influent all combined



17


system, so this work should be perform carefully. Normally, several round of drive test is required to confirm the suitability of such changes. 6) Sectorization sectorization may be one of the solutions when system traffic become a limiting factor of network performance, and its related problems include neighbor list re-optimization, RF retuning, power adjustments, radio parameters modification, etc. This should be a progressive process to guarantee the stability of network.

7.2

Optimization Based On Driver Test/Walk Test Data

7.2.1

RF coverage optimization The optimization based on coverage process is shown in the following figure. Figure 7-1

RF optimization based on coverage

Poor Coverage

RF Optimization

N

Meet Coverage Requirement

Y

N

Y

Add Sites and RF Optimization

Meet Coverage Requirement

Finish

Scanner measurement data are mainly used for RF coverage optimization. The purpose of RF coverage optimization is to provide a good coverage and build a solid basis for various services. CPICH RSCP and Ec/Io are the key index for coverage analysis. RF coverage optimization focuses on poor coverage area, pilot pollution, overshooting, interference, etc. Typical solutions for RF adjustments are given below:



18


7.2.2



Tune the antenna azimuth, down tilt, or CPICH power, which are the most frequently used methods to eliminate pilot pollution, overshooting and related intra-frequency interference. When possible, the antenna type and antenna height can be changed too.



Check the radio environment to avoid signal shadowing, canyon effect, water land effect, etc. When no change can be m ade on site selection, adopt RF adjustments or parameter modifications to eliminate the influence of unideal site selection.



Find the weak coverage area and try to optimize by RF adjustments. If the weak coverage area exists in shopping malls, tunnels, underground park ing lands, subway entrance, or high buildings, etc., the coverage can be enhance by adding indoor coverage system.



Check the hardware problems, especially the power output of NodeB, make sure that there is no cell shrink caused by PA problems.

Call Failure Analysis The troubleshooting process for call failure is shown as below.



19


Figure 7-2

Call Failure Analysis Process

Call Fail

Coverage OK?

No

Enhance coverage

Yes

Caller?

No

No

Paging problem

Yes

Yes RRC connection setup failed

Paging received?

Yes

RRC connection setup problem

Yes

Authentication and ciphering problem

No Authentication and ciphering failed No Yes RAB setup failed

RAB setup problem

No Handover caused failure

Yes Handover problem

No Abnormal problem

End

For paging related problem, check the following flow chart.



20


Figure 7-3

Paging Problem Troubleshooting Process

Paging problem

RNC sent paging?

No

Abnormal equipment problem

Yes UE received Paging？

Yes

No Power allocation insufficient

Yes

Chang power allocation for AICH/PICH/PCH

No Cell reselection problem

Yes

Optimize reselection problem

No Abnormal problem

End

For RRC connection setup related problem, check the following figure.



21


Figure 7-4

RRC Connection Setup Problem Troubleshooting Process

RRC setup problem

UE sent RRC request?

No

UE abnormal probl em

No

Adjust PRACH param eters

No

Congestion or other probl em

Yes RNC received RRC request? Yes RNC sent setup message? Yes UE received setup message?

No No

Yes UE sent setup complete message?

Cell reselection

Yes No

Adjust FACH param eters

Optimize cell reselection UE abnormal probl em

Yes RNC received setup complete message?

No

Adjust UL open loop power control param eters

Yes

End

Typical reasons that will cause the call failure are: 

Originate call in weak coverage area that the signaling process cannot complete;



Callee is originating location update that results in the failure of paging;



Cell reselection not quick enough such that the call cannot be originated in the best cell;



Cell radius is not properly configured, and the UE cannot access network.

Corresponding solutions can be given below: 

Make RF optimization to eliminate coverage hole, pilot pollution, overshooting, etc;



Optimize the location area border, so that location update happens less frequently and at the areas with low traffic if possible;



22


7.2.3



Change cell reselection parameters for only problematic cells, so that the UE can select to the best cell quickly;



Change power allocation parameters for PICH/AICH/PCH/PRACH/FACH if needed, and also the random access parameters can be modified;



Modify the cell radius so that it can match the desired coverage radius of the cell.

Call Drop Analysis The call drop rate reflects the system sustainability for different services; it ’s the most important performance indicator that users can directly experience. Wider range call drop rate definition should include both the CN and UTRAN call drop, in this section; radio related call drops are emphasized. From high level point of view, most call drops are caused by the following three types of radio problems, that is, coverage, handover, and interference. They are described below, respectively.

7.2.3.1

Caused by coverage problems According to optimization experience, the following coverage problems may lead to call drops: 

Serving cell has the overshooting problem due to either good propagation environment, or high power settings, or high site. When UE moves to the area covered by both overshooting cells and normal expected cells, UE might handover to the overshooting cells, but after that cannot make any outgoing handover from the overshooting cells due to absence of neighbor cells, which result in the call drop;



Overshooting caused invalid scrambling code reuse;



Canyon effect or water reflection effect leads to the overshooting of serving cell, handover decision maybe affected and result in the call drop;



Due to the isolated island effect, the UE in the isolated cell can not make outgoing handover;



Pilot pollution arise due to no significant dominant pilot can be found, the ping pong handover might be observed during the call and result in the call drop;



Coverage holes exist at the intersected area of two adjacent cells, UE lost coverage during the call;



Shadowing effect caused by high buildings, this leads to weak coverage area or areas with rapid signal fluctuation.

Possible solutions are given below:



23


7.2.3.2



Eliminate the drifting or overshooting signals. Check the down tilt or azimuth of antenna, or reduce the pilot power to optimize the coverage area. Pay attention to special coverage scenarios such as street coverage or water area coverage.



Find the weak coverage area and try to optimize by RF adjustments. If the weak coverage area exists in shopping malls, tunnels, underground park ing lands, subway entrance, or high buildings, etc., the coverage can be enhance by adding indoor coverage system.



Check the hardware problems, especially the power output of serving cell and neighbor cells, make sure that there is no cell shrink caused by PA problems.

Caused by handover problems According to optimization experience, the following handover problems m ay lead to call drops: 

Missing neighbor cells;



Hardware problems result in abnormal handover;



Illegal scrambling code reuse result in handover failure;



Isolated island problem result in handover failure



Handover target cell is overloaded and result in handover failure;



Source cell in seriously interfered in downlink or uplink and result in handover failure;



Handover parameters are not properly set, such as 1a/1b/1c/1f/2d/2b/2c/3c/3a related parameters, which result in handover latency or call drop;



Ping Pong handover failure result in call drop;



Signal sudden drop in corner or shadowed area, which result in handover failure and call drop.

Possible solutions are given below: 

Check neighbor list definition;



Check hardware faults, make sure there’s no hardware alarm;



Check scrambling code planning;



Make RF adjustments to control interference in network;



Control cell load at stable level;



24


7.2.3.3



Modify handover related radio parameters according to radio environments;



Control and eliminate the internal and external interference in network,

Caused by interference problem According to optimization experience, the following interference problems may lead to call drops: 

Intra-frequency multiple access interference;



Interference casued by missing neighbor;



Interference caused by illegal scrambling code reuse;



Pilot pollution;



Other system interference;



Interference caused by microwave, satellite receiver, radar, TV receiver, etc.

Possible solutions are given below:

7.2.4



Control cell load;



Check neighbor list definition to avoid missing neighbors;



Check scrambling code planning;



Make RF adjustments to control the overshooting;



Add space isolation or Tx/Rx filter to reduce interference from other systems.

Handover Failure Analysis For handover failure problem analysis, please refer to the call drop analysis methods described above.

7.2.5

High Access Latency If the service access delay time is abnormal, it is commonly a system level problem, that is to say, not only part of cells suffers from the latency problem, but most cells should be affected. The delay time is related with both RAN and CN. In order to find out the limiting factor, tests can be made to check the time consumed in each major signaling stage, and figure out which contributes most to the access time. The test should be made in fixed spot where the signal Ec and Ec/Io is good and only one cell is dominant cell, in order to



25


avoid the influence of weak coverage or signal fluctuation on the access time measurement. Since the measurement needs end to end signaling timing, the signaling tracing system should cover the whole process of service call. Main configurations that will affect the access delay time are listed as follows: 

Early assignment or later assignment in CN;



Authentication switch in CN;



Maximum FACH power allocation;



PICH/AICH/PCH power allocation;



RRC setup on FACH or DCH 13.6k or DCH 3.4k;



Random access power settings;



RRC retransmission timer and counter T300 and N300;

Using later assignment and turn off authentication from CN can reduce the access time. Increase the values of power related parameters can reduce the access time too, but they need to be fine tuned to avoid interference and influence on capacity. A normally used policy is to adopt DCH 13.6k high speed SRB to reduce the signaling time, and then reconfigure it to 3.4k after service is established.

7.2.6

Low Data Throughput Data throughput for PS services is related with many factors, so its optimization needs to check many possible problematic configurations and find out the major reasons for low data throughput. The troubleshooting process involves nearly all network elements including USIM, UE, NodeB, RNC, HLR, SGSN, GGSN, Transmission, even inter-connected routers and switches or cables. Data collected from other interfaces in addition to Uu interface is often needed, for example, the signaling tracing from RNC, the data log from Node B, the transmission test, etc. According to optimization experience, the following problems may lead to low data throughput, and corresponding suggestions are presented for reference. 

Equipment Alarm. Alarm checking is the first action to be taken when facing with performance problems. For throughput problem, alarms should be checked for Node B, RNC, SGSN, GGSN, LAN switch, router, clock abnormal, transmission error, and other equipment alarms.



Radio environment. Data services have strict requirements on the signal strength, purity and stability. Signal strength is easy to be checked, signal purity is reflected by pilot pollution situation in network. If no dominant pilot can be found, there might be



26


interference or ping pong handover, which in turn will make throughput drops. The stability of signal will also directly map to stability of throughput, without continuous good coverage, there throughput might drop due to dynamic rate control to adapt to the changing of radio conditions. The solution is to make RF optimization to provide good basis for PS services. 

Configuration faults for hardware or software. The following miscellaneous factors might be concerned during the troubleshooting process. 1) USIM rate allocation; 2) Compatibility of UE or data card; 3) Capability of UE or data card; 4) Device drivers for test notebooks and servers; 5) Firewall settings for test notebooks and servers; 6) APN settings for test notebooks; 7) Server performance, TCP/IP settings, or service platform software configurations for PDN or other servers; 8) SGSN/GGSN problems. The correctness of related configurations should be confirmed.



The power allocation and code allocation strategy, along with scheduling strategy in network. Normally for different application scenarios, there are default strategies recommended, but due to concrete optimization requirements, they can be tuned case by case.



The Uu interface block error rate, including both UL BLER, DL BLER and CQI, the solution is to check the coverage of interference, and make RF adjustments.



The Iub interface transmission error, delay jitter, or bandwidth insufficiently configured. These factors should be checked so that Iub is not the bottleneck.



Check the rate of application layer throughput over RLC layer throughput, if the value is considerably lower, there might be h eavy overhead result from unnecessary TCP/IP retransmission. The solution is to modify the TCP receive window or MTU packet size, or try adopting advanced TCP features such as TCP SACK, to improve the performance for single TCP connection. Another alternative way is to use multi-threaded download or upload test tools so that TCP window and other parameters are not the limiting factors during throughput evaluating process.

7.3

Optimization Based On OMC Performance Data

7.3.1

Overall optimization process When the network is commercial launched, the traffic is expected to increase gradually, and statistics from OMC can provide more and more rich support for performance monitoring and optimization. From the KPI report some KPI degradations can be observed and analyzed. The basic idea for analyze the problems is judging from RNC level to cell level, try to exclude unrelated factors. Correlated analysis between different KPIs is important for the troubleshooting. The RNC level KPIs are firstly examined to understand the overall performance. If RNC level KPIs are abnormal, more detailed cell level checks should be



27


made to see whether the problem is common in the whole network or specific to some certain cells. TopN Cell level abnormal KPIs will also be checked. For common problems, analysis shall be made on coverage, capacity, interference, transmission, software, hardware, radio parameters; for specific cells, cell level statistics should be checked in detail. What needs to be noticed is that, when analyzing KPIs that are presented in form of percentage, the absolute counts should be checked at the same time, because the percentage value sometimes may cover some potential problems. One case is that the percentage value is not good, but the denominator and numerator are small, which means the statistics is not based on large number of samples and are not reliable temporarily. The other case is that the failure counter as numerator is large, but the total samples as denominator is much larger, so the percentage value is good and problems are easy to be ignored, this case should be avoided. During the troubleshooting, pure OMC data is not sufficient, other needed data sources include alarm data, drive test data, signaling tracing data, etc. For complicated performance problems, only through combined analysis can the trouble been solved. The following figure shows the troubleshooting process base on OMC statistics. First, the sudden and self recoverable problems due to holiday, weather, and so on are identified by the engineers, and corresponding actions are taken to prevent similar problems happen again in the future. For other problems, equipment alarms are checked in the first step, to make sure that problems are not caused by major alarms. Next, the low KPI are filtered out and presented on the map, at the same time, collect some background information such as transmission configuration table, software and hardware version, radio parameters configuration, etc., check whether those filtered cells have some characteristics in common, if any, focus on the common points to make further analysis. Typical check points are listed below: 1) Whether software or hardware version upgrading is made recently 2) CPU load and link utilization rate; 3) Transmission broken or transmission bit error rate too high 4) Whether uplink interference exists 5) Check radio parameters with stand default configurations 6) Whether problems happens in

certain time only

7) Whether problems happen in certain area only 8) Whether problematic cells belongs to certain logic group 9) Which reason contributes most to the failure 10) Neighbor list checking 11) Whether there is site work conducted recently



28


If no common points can be found for abnormal cells, or after the optimization there are still unqualified cells, then analysis focused on single cell is made. Main KPIs to be analyzed are typically call setup success rate, call drop rate, various handover success rate, etc. The whole analysis process can benefit from CNO, which make the statistics and query of huge volume of performance data easier for engineers



29


Figure 7-5

Overall optimization process Start

Retrieval of performance data

Weather fluctuation, holiday, congregation, sudden broken of transmission, power supply fault, etc.

Abnormal RNC performance index？ Y

Causation analysis and record

Sudden, selfrecoverable abnormal?

Y

N

Equipment alarms exist?

Improvement suggestion

Y

Handling alarms N

RNC performance index recovered?

Y

N

Abnormal cells TopN analysis and visualization on map

Common features exist in abnormal cells?

Transmission, SW/HW configuration, radio parameters settings.

Y

CN/RNC

Transmis sion

HW

SW version

Interfer ence

N

Radio parameters

Time span

……

Analysis of common characteristics N

Problem solved?

Y

N

Abnormal performance index analysis for single cell

Call seup

Call drop

Soft 2/3G PS rate handover interoperation

……

N

Related problems solving

Performance index OK? Y

End

7.3.2

Call Setup Success Rate Optimization The main process for optimization of call setup success rate based on OMC performance data is similar to the process based on drive test data, and the process is given below.



30


Figure 7-6

Call Setup Success Rate Analysis

Traffic analysis

Alarm exist？

Yes

Handle equipment problems

No Low paging success rate?

Yes

Handle paging problems

Yes

Handle RRC problems

Yes

Handle RAB setup problems

No Low RRC setup success rate? No Low RAB setup success rate? No Low RB setup success rate?

Yes No

Handle RB setup problems

No

End

Using CNO the TopN cells with deteriorated certain KPI can be filtered out quickly. At the same time, some other counters and information should be used such as the access attempt counter, the CS traffic or PS throughput to help making a combined judgment. Analysis can focus on each signaling phase one by one, including the RRC setup process, the paging process, the RL setup process, the RB setup process, the transmission setup process, the synchronization process, the admission control process, etc. Importantly, the OMC can generate statistics on each failure reason counts, which provides a very helpful hint for tracing the problem in a correct way. When reasons for problematic cells are found, then actions can be taken for optimization. Analysis methods are the same as optimization based on drive test data.

7.3.3

Call Drop Rate Optimization For call drop rate optimization, firstly TopN cells are identified, and then most serious call drop reasons are figured out for analysis. Except the commonly used check points, other typical check points for optimization of call drop rate include: 

If the failure signaling is UciuError or RL failure, and the failure reason is RLC unrecoverable, then the call drop might due to poor coverage;



31




Analyze the incoming and outgoing handover success rate for the cell, to check whether the handover failure caused the call drop;



Analyze the uplink RTWP during the time span with high call drop rate, if the RTWP level is not accord with cell traffic, there might be uplink interference which caused the call drop.



Check code utilization rate, if the percentage is above threshold, the call drops in the cell might be caused by code resources congestion.

When call drop reasons are roughly identified, drive test in the problematic cell is needed to collect more data for combined analysis. Finally, suggested optimization solutions are implemented to verify the validity of solutions.

7.4

Optimization in Network Operational Phase In the operational phase of network, daily KPI monitoring is needed to keep system performance at a stable level. This is actually a active optimization action intended to find potential performance problems before they affect the user experience. Typical active performance optimization is described below.

7.4.1

Traffic and RTWP (Received Total Wideband Power) monitoring With the correlated analysis function offered by CNO, the relationship curve of call drop rate versus traffic and RTWP versus traffic can be drawn and monitored. If any abnormal relationship or trend is observed, the engineer can use the hint to drill down the pr oblems. For example, if traffic and call drop rate both grows, and they embody proportional relationship, also the traffic grows beyond certain threshold, then capacity expansion might be needed. Another example, if RTWP and traffic both grows, but the RTWP rise is not proportional to the traffic grow, there might be outer interference from other radio systems. When cell is unloaded, the RTWP is among the value of -106dBm and -104dBm. Considering maximum 75% uplink load, which corresponds to 6dB noise rise, the RTWP around -100dBm to -98dBm is regarded as normal. If the maximum or average cell RTWP is not stably increasing, the cell should be checked.

7.4.2

Code resource utilization monitoring The utilization rate of code resources is expected to be at a desired level, too high or too low utilization rate is not good situation. Practically, sites at network borders have low utilization rate, while sites in dense urban or hot spots have a higher rate, which should be monitored with special emphasis, to avoid the network congestion due to spreading code.



32


Normally the code resources are allocated dynamically, if the code resource congestion lasts for too long time, the traffic might be too much for the cell, and capacity expansion should be considered.

7.4.3

Transmission utilization rate and RLC retransmission rate monitoring For PS services, the throughput is main goal of optimization. From OMC point of view, the transmission utilization rate and RLC retransmission rate can both hint the problems related with transmission part of network. If the Iub interface bandwidth is not configured sufficiently, then with the increasing of data traffic, the bandwidth might be used up quickly. By filtering out those cells with high transmission utilization rate, engineer can easily identify which cells should be equipped with more transmission resources. Another problem related with transmission is the bit error rate which can affect the PS performance. From RLC retransmission rate monitoring, the potential transmission problem can be noticed by engineer, and active troubleshooting can be made.

7.4.4

Neighbour Cell Optimizaiton CNO can make neighbor cell optimization in terms of adding, deleting and reordering, all of these decisions are made based on the statistics on handovers and detected set report. After the sorting the existing neighbor cells according to happened handovers within several days, the more frequently used handover target cells are assigned with higher priority, and those neighbor cells with handover occurrence count less than a threshold can be ranked as low priority neighbor cells, when possible, they can be e ven be deleted. Those defined neighbor cells with zero handover occurrences can be safely deleted too. In this way, the work effort needed in neighbor list optimization is cut down, and some potential performance problems can be solved.

7.5

Optimization Based On Capacity The optimization based on capacity process is shown in the following figure.



33


Figure 7-7

Optimization Based On Capacity Monitor CE or Power

CE or Power Constrain

Power Constrain

CE Constrain

Add BPC Boards and RF Optimization

Meet Capacity Requirement

Add Freqency and RF Optimization

Y

Y


N N Cell Split and RF Optimization

RF Optimization

Y


N

N N

Add EBBUB and RF Optimization

Y


Y


N

Add Sites

Y

Meet Capacity Requirement Finish



34


7.6

Optimization for VIP needs 1.

Define a VIP list

We define both VIP sites and VIP customers in order to handle their problem with high priority. 2.

Optimization for VIP SSV

For the SSV of VIP sites, we arrange special engineers take charge for it, and use stricter criterion compared to normal sites when SSV. Unless all the test items have been passed, then the VIP SSV is finished. Meanwhile, we do RF adjustment and other measures to first ensure the coverage for VIP sites, so as to give better service for them. 3.

Optimization for VIP customer complain

Higher priority will be given when a VIP customer’s complain happens and we will deal with it in sooner time. According to the specific complain content, if it is need to implement a work order to change some parameters of sites, the parameter track record for VIP will be checked firstly, Only if the changes are new for the record or current changes will not have adverse effect on recorded changes, the work order can be performed and the parameter track record for VIP should be updated again. Otherwise a compromise or alternative solution needs to be sorted out first.

8

Optimization and diagnostics tools In the following sections, tools to be used to collect, analyze and improve the performance during the various stages of the project are described. These tools are introduced based on the stages they are applied in. For each tool, its features, its applicable problem domains, as well as its usage are described. And, it ’s necessary to check calibration of equipment before use. Notes: In the project, the detailed optimization tools should be decided based on actual project conditions.

8.1

Drive test and Analyzing tool The CNT/CNA will be used to collect process and analyze the drive-test measurements and reports generation. In this document, CNT/CNA will be explicitly described as these are ZTE developed products and hence are deeply understood by ZTE engineers.



35


CNT is professional test software of UMTS radio network. It collects data accurately, including GPS locating data, UE diagnostic test data, as well as data in every layer (RLC, MAC) in the data service, and displays these data in the format of Map, Table, Text, or Graph etc. CNA is professional analysis software of UMTS radio network optimization. It is based on the drive-test data and other auxiliary data. It provides versatile intelligent analysis for radio network, and accordingly performing network optimization quickly and highly efficiently. It provides abundant and useful analysis parameters. For example: Rx Power, Tx Power, Total Aggregate Ec/Io, Best Ec or Ec/Io distribution, Best PSC distribution, specific PSC’s Ec or Ec/Io distribution, pilot pollution, coverage rate, overshooting diagnose, and so on. It provides powerful message analysis functions, supports the thorough decoding of RRC and NAS message, and can assemble system information blocks of segment transmission. Moreover, it supports browse, play, query and filter of messages, the statistics function for basic and user-defined event, as well as versatile delay analysis and KPI analysis function. The CNT software has the following features: 

It provides quick and accurate acquisition and record of data, including GPS location data, UE diagnostic test data, as well as data in each layer (RLC, MAC) in the data service.



It can support both indoor and outdoor test.



It provides real time display of all sorts of test data in the format of Map, Table, Text, or Graph etc.



It supports automatic detection of GPS and scanners in all available ports, and manual configuration of these parameters.



It provides document processing function, including processing of NodeB information, electronic map, dial test route, test data, test plan, and project workspace.



It provides display of scanner related parameters, including scanner related scrambling code, synchronization channel, frequency spectrum analysis, peak value power and channel power.



It provides display of GPS information, including GPS locating information and satellite signal.



It provides elaborate real time acquisition and display of parameters, including view of physical channel parameters of layer 1, transport channel of layer 2, and RLC related parameters.



It provides powerful browse, real time decoding, filtering and sort display of layer 3 messages.



36




It provides powerful geographic display function and binding of data layer and GIS graphic layer, which realizes coverage display and statistics and acquires space distribution property of parameters. Additionally, it has operational functions, including zoom, move, comment, select, information display, and automatic or manual offset, which can avoid overlapping of data graphic layers.



It provides powerful service test function, and supports automatic test plan of voice service and data service, as well as customizing automatic test plan document.



It provides data playback function, including getting the recorded file while testing, and redisplaying test procedure.



It provides powerful function of event alarm analysis, and customized alarm setting of test parameter items.

The CNA software has the following features: 

It provides abundant and useful analysis parameters. For example: Rx Power, Tx Power, Total Aggregate Ec/Io, Best Ec or Ec/Io distribution, Best PSC distribution, specific PSC’s Ec or Ec/Io distribution, pilot pollution, coverage rate, overshooting diagnose, and so on.



Its data analysis is based on dial test data, and is a comprehensive analysis combining station information and GIS information.



It adopts various analysis methods based on geography, graphs and tables, statically and dynamically.



It supports outdoor test data as well as indoor test data. The indoor geography analysis can be performed with indoor maps.



It provides powerful geography analysis functions, and binding of data layer with GIS graphic layer, which realizes coverage display and statistics, and acquires space distribution property of parameters. Additionally, it has operational functions, including zoom, move, comment, select, information display, and automatic or manual offset, which can avoid overlapping of data graphic layers.



It provides powerful message analysis functions, supports the thorough decoding of RRC and NAS message, and can assemble system information blocks of segment transmission. Moreover, it supports browse, play, query and filter of messages, the statistics function for basic and user-defined event, as well as versatile delay analysis and KPI analysis function.



It provides powerful and professional pilot analytic and diagnostic functions, such as coverage render of specific or all PSCs, pilot pollution analysis, neighbor list analysis, and coverage rate analysis etc.



It provides flexible and customizable statistics and report functions.



37




It provides convenient and flexible station information management, and supports establishment, modification and exchange of station information.



It provides powerful query function based on analysis.



It provides versatile customized playback functions, and supports integral data play and single point track play.



It integrates advantages of many main stream network optimization software in the market.



It is compatible with the test data of Agilent WCDMA dial test equipment in all aspects.

The user manuals of CNT and CNA provide the detailed operation guide for both tools. In this document, only typical usage of CNA is covered. Normally, CNA is used to solve RF related performance problems, such as swapped feeder, weak coverage, pilot pollution, overshooting, missing neighbors, call drop, etc. These analysis actions are mainly driven by various KPIs. When some KPIs cannot meet the requirement, such as the coverage KPI, the accessibility KPI, the sustainability KPI, etc., then drive tests are made to collect data, then CNA is used to analyze from air interface side. Sometimes the signaling tracing on RNC side is also needed. 1) Firstly, the coverage of the network is examined. Normally this comes from the scanner test data, both Ec and Ec/Io of distribution of the network can be generated, and coverage rate is defined as (Number of samples with RxPower

 R

and Total

aggr. Ec/I0  S) / (Total number of samples), see the picture below.



38


Generally, good coverage is not solely defined by coverage rate, some other RF problems can also be identified through the coverage plot. See the picture below, three typical coverage problems can be figured out, that is i) Cell 425 and Cell 427 are wrongly configured with swapped feeder, ii) Cell 414 and Cell 415 have overshooting problems; iii) the area near the oval road in the top middle of the picture has pilot pollution problem, where several signals exist together and become the best server alternatively. All these problems are to be solved before service related optimization can start.



39


2) In CNA, the identification of pilot pollution and overshooting can be automatically done. For those pilots that are above the defined pilot pollution threshold but are not candidate of the active set, they are regarded as pilot polluter. See a demonstration picture below. Pilot pollution is an indication of how and what is causing an area having bad quality RF coverage (i.e. low Ec/Io values measure). When an area is identified to be a pilot polluted, it means there are too many measured pilots in the area. As a result, the solution is to reduce the number of pilots in the area by creating a dominant server. The resulting actions are normally antenna tilt changes or increase/decrease of CPICH power to modify cell coverage.



40


For those cells that overshoot to distance greater than the distance threshold, and the percentage of sample points in overshot area is greater than the percentage threshold, then they are considered to be overshooting cells, and the overshooting critical levels can be judged according to the overshooting distance a nd the signal strength of overshooting signal in the affected area. See the picture below. Again, there is no absolute threshold of overshooting. T he analysis is always going to be relative and driven by bad quality coverage area (i.e. low Ec/Io).

3) After the RF rough optimization through scanner based data, the service related tests can be performed. In this stage, scanner may not be used, but UE data along with scanner data is preferred. For example, by combined analysis of these data, the missing neighbour can be identified. The algorithm principle is described as follows. All drive test conducted will consist of active calls UEs and at least 1 scanner. In the drive test, the UE obtains the neighbour list from the Node B and the scanner continuously performs the frequency scan for 512 PSCs. From a data point, when a PSC is not in the neighbour list and the scanner finds that the PSC strength exceeds the threshold X, this PSC shall be judged to be the missing neighbour. Each point with missing neighbour can be shown on the map, see the picture below. The threshold X can be set to -100dB by default. However, again this value is relative to the ex isting coverage environment. It should be defined by individual RF optimization engineer. For example, if the overage coverage of the area in terms of RSCP is -80dBm, it can be defined that signal strength of individual pilot (i.e. RSCP) of <-95dBm will have very little effect to the overall coverage of the area. As a result, individual pilot of RSCP of -95dBm can be set as a threshold.



41


4) With CNA replay function, the problematic points such as access failure, call drop, etc., can be identified on the digital map and engineer can replay the data before these points, with desired time span and replay speed. Each point can be analyzed in details, including the Rx Power, Tx Power, the RSCP and Ec/Io, the active and monitored set, etc. See the following picture.

Call drops

Normally, the diagnostic process involves air interface decoding. CNA can provide the detailed message decoding support including NAS messages. Messages can be filtered, and customized events can be defined based on the specific combination of messages. See the following picture.



42


5) After the drive test, the performance evaluation can be done by various statistics. User can query the data according to user-defined rules, and for items that need maximum/minimum/average statistics along with PDF/CDF statistics, there are also quick and easy ways to present these data in the form of tables and figures. See the picture below. Rx Power Percentile(%) Cumulative Percentile(%) 100

100 100

90

) % ( e l i t n e c r e P

85.93

90

80

80

70

70

60

60 47.55

50 40

20

0

40

38.38 28.98

30

10

50

30 14.07

8.61 9.4

0.79 0.79 0 (-INF, (-100, (-90, (-80, -100] -90] -80] -70] Range 0

20 10 0

(-70, -60]

(-60, +INF)

In CNA, some built-in KPI items are ready to be checked. User can also define the desired KPIs based on related messages and events. See the picture below.



43


6) Automatic report generating function is used to output drive test and performance statistics data according to the preset templates. This helps to save time when performing the post processing of drive test data.

8.2

ZXPOS-CNO ZTE CNO is mainly based on the OMC performance data, it can help to analyze the network performance in many aspects and provide rich data views, so that the whole network optimization process including the data preparation, network evaluation, network performance troubleshooting, network adjusting, performance verification, and the output of optimization results can be fulfilled in an integrated way. Performance analysis is an important part of network operation and maintenance work, and that is why performance analysis is the key feature provided by CNO. Commonly used performance index includes the call drop rate, call congestion rate, handover success rate, call setup success rate, traffic volume, resource availability, etc. During the network operation or optimization, the performance anal ysis can help to identify problems reactively or proactively, so that the network performance can be under the control, and troubleshooting process can be driven by live network statistics more quickly. Main functions supplied by CNO are listed as follows: 

Import various data sources like engineering parameters, radio parameters, performance counters and KPIs, neighbor configurations, etc.



Neighbor cell planning, validity checking and optimization



Primary scrambling code planning and optimization



Radio parameters validity checking



44




Cell filtering query



Performance index query and quick query



Customization and view, including the raw counter view, KPI customization, cell group customization, query template customization, cell filtering template customization, import and export of templates

The objects to be analyzed with CNO include the whole network, the local region network, RNC, NodeB, cell and cell group. Cell group can be defined as the combination of multiple cells from different RNC/NodeB/local network. The counters and KPIs analysis can support hourly, daily, weekly granularity, and user can select to make summary on input time. The data from certain t ime span can be presented as curve chart, bar chart, or pie chart, and the cell level performance data can be rendered on digital map using different gradient colors. The analysis result can be saved as Excel files for further external usage. Normally there are two ways to use CNO for network performance monitoring. The one is searching certain cells based on KPI input; the other is searching certain KPIs based on cell input. Typical scenarios for these two methods are TopN cell analysis and VIP cell performance guarantee. For example, after daily performance check, if some KPIs are not satisfactory, the engineer might want to find out which cells are the worst ones, then the troubleshooting work will be targeted at these cells, that is so called TopN cell analysis method. See the following picture. Pre-defined performance measurements based on performance counters are listed in CNO. By selecting various performance counters, for a cell or specific groups of cells, within any defined periods, all performance measurements can be queried with a click of buttons. This data can be extracted in to excel spreadsheet for further analysis.



45


On the other hand, for some certain important cells, such as VIP cells, recently tracking problematic cells, etc., their performance KPIs should be checked every day for performance guarantee. For convenience, these two types of quer y can both be saved as templates which can be reused repeatedly. See the following picture.



46


Based on the different queries, some special purpose analysis can be made to help troubleshooting the performance problems. One typical scenario is to use RTW P query to help identify network interference. The idea is simple, first, all cells in the whole network or in certain RNCs which have average RTWP above the defined threshold are filtered out, then engineer can choose to present these cells on the digital map, and check whether the distribution of these cells are in certain small regions. If so, there might be interference source exists in the center of these interfered cells.

8.3

Detected set neighbor report analysis Detected set neighbor analysis is a helpful function for optimization engineer to quickly complete the neighbor list definition. Normally neighbor list optimization is time consuming, and the miss neighbors will often lead to call drops, which is critical performance problem in commercial network. So the neighbor list optimization based on OMC data can help a lot with this problem. Before the detected set neighbor can be reported, cell detected set measurement switch (The parameter is used to control whether UE report cell detected set or not, the missing neighbor checking could rely on the statistic of detected set, The step to turn on the switch is as following: login in Citrix —RAN EMS—view—performance management— Performance—Measurement task management —Create task: in task basic information sheet, select cell detective set statistic for the object type; meanwhile in task measurement object sheet, select the target cells, then confirm it and active this task.) should be turned on for targeted cells, and measurement tasks should be configured to UEs. When the event 1a is triggered for a detected set neighbor, a so called virtual handover is recorded. RAN should make statistics on all such virtual handovers, and report to the background processing system, i.e., the detected set neighbor analysis tool. With the analysis tool, the handover data between various cell pairs can be analyzed and presented, so that engineer can easily find the missing neighbors and the redundant configured neighbors, and make corresponding adjustments. The detected set neighbor analysis can be made on the whole network basis first, so as to help engineer to quickly locate the problematic cells, then for each target cell, the detailed missing neighbor analysis can be performed, such as how many missing pilots are found, how the pilot PSC is mapped to geographical adjacent cell or cells, where are these missing cells are located, etc. An example is given in the following. From the following picture, cell 10233 and 10232 are found to have virtual soft handovers, they are then suspected to have missing neighbors. Next, cell 10232 is picked for detailed analysis to check how many neighbor cells are missing for this cell.



47


In the single cell handover statistics table, different colors for each record line have different meaning: 1) Black means the handover is performed through well-defined neighbors 2) Blue means there is missing neighbor reported, and from the reported PSC, only one matched cell is found. 3) Green means there is missing neighbor reported, and from the reported PSC, more than one matched cells are found, and the geographically nearest matched cell is sorted at the first position. 4) Red means there is missing neighbor reported, but no matched cell is found that has the reported PSC. This situation is not common in the network, but it might happen. Checking the following picture, the PSCs 190/342/125/511 are found for cell 10232, and then in a defined limited nearby area of cell 10232, the matched cells are found for PSCs 190/342/125, but PSC 511 has no matching cell found. PSC 190 has only one matching cell ID 12113, which indicates cell 10232 should add cell 12113 as neighbor cell with great probability. For PSC 342 and 125, two matching cell ID can be found, but cell 10311 and 10331 has shorter distance than cell 10312 and 12043, respectively, so cell 10232 should add cell 10311 and 10331 as neighbor from practical point of view. In additional, it is necessary to verify the missing neighbor judging from the coverage.



48


For a visual presentation, the suggested missing neighbors can be rendered on the digital map, and the user can then modify the current neighbor definition manually together with important paths information that was collected through drive test and VIP complaints, and export to the OMC finally to implement the modifications. See the following picture.

8.4

Signaling trace In some cases, troubleshooting process should be supported by many data sources. T he drive test data can embody the downlink behavior and performance in Uu interface, but the behaviors of other interfaces such as Iub, Iu cannot be observed. For complicated problems, the combined analysis is indispensable. Signaling trace in RNC is then mandatory for engineers.



49

UMTS Radio Network Optimization Proposal(for New Network)_R1.0

Recommend Documents