08 WO_NAST3014_E01_1 UMTS Network Optimization Case-70.ppt

UMTS Network Optimization Case

ZTE University

Content

Coverage Case Antenna Adjusting Case Pilot Pollution Case Neighbour-Cell Case Cell Reselection Case Handover Case GSM/UMTS Inter-operation Case Call Drop Case Voice Quality Case

Coverage Case

Coverage Case

Problem analysis Spot A is about 2.7km from Sousse2 site. A is the entrance of a

uptown highway and has a turn of about 90 degrees. Signals of cell 228 of Erriadh TT site become weak suddenly because the cell is sheltered.

Spot B is about 2km from CTT Skanes site. The seaside road that B located is at a lower sea level than the CTT Skanes site. Signals of cell 332 of CTT Skanes site can be received by the mobile phone after penetrating several 2~3-layer buildings. At around spot B, the pilot signal strength is reduced to be below -100dBm.

The NodeB in Sahaling is quite restricted by the environment. The site height is only 25m; there is little space for increasing the height.

Coverage Case

Solution Adjust the transmit power of common channels Increase the pilot transmit power

Effect assessment The coverage effect and the call-drop rate is optimized. There is

almost no dropped call along the express way.

Channel Before the Adjustment After the Adjustment

CPICH 10% 15%

BCH -3dB -3dB

FACH 0dB 0dB

PCH -3dB -3dB

PSCH -4dB -4dB

SSCH -4dB -4dB

PICH -7dB -7dB

AICH -7dB -7dB

Content


Antenna adjusting Case 1

Problem During the coverage optimization DT along Zhongshan Road No. 1 and

Donghu Road, it is found that the receiving power of the UE one Donghu Road between the Donghu base station and Shuqian Road base station is weak and less than -85dBm. In addition, the pilot signal quality Ec/Io is also poor and less than -13dB in this area.

Signal distribution in the Donghu Road area before the optimization

Problem analysis: Through the review of the DT data with optimization analysis

software ZXPOS CNA and the survey on the site, it is found that in front of Sector 2 (with the scramble 437) of the Shuqian Road base station, there are dense buildings which form a serious barrier and influences in the coverage of the sector. Besides, the areas within scores of meters in front of Sector 1 (with the scramble 439) of Donghu base station is also completely blocked by a row of high residential buildings, which makes Sector 1 unable to cover that area.

Solution Change the direction angle of Sector 2 in the Shuqian Road base

station from 240o to 230o to enhance the coverage of that area of Donghu Road.


Signal distribution in Donghu Road after the optimization

Effect after optimization From the analysis of DT data, it can be seen that in this part of the

Donghu Road, the UE receiving power is >-85dBm and the pilot Ec/Io>-13dB, which meets the coverage requirement.


Signal distribution of Baishi Road before the optimization


Problem Through the analysis of the DT data of Baishi Road, it is found that

pilot strength received in the middle part of road is less than -95dBm, as shown in Area A in the figure below:

Analysis: It is found that the coverage of this area is provided by Sector 2 of

Shenzhen University base station. The direction angle of Sector 2 is 110° and the downward tilt angle is 4°. Both shall be adjusted to enhance the coverage of Baishi Road.

Solution Adjust the antenna direction angle of Sector from 110°to 120° and

the downward tilt angle from 4°to 12°.


Pilot coverage of Baishi Road after the optimization

Effect after optimization Conduct DT on the Baishi Road after the optimization. From the DT

result below it can be seen that the pilot strength is improved to more than 90dBm.


Content


Pilot pollution Case

Flower hall site is located on the Gaoxun Tower beside the Quzhuang cloverleaf junction. Its is at a height of 70m. After line testing, it is found that the 425 (scramble) cell of the site provides cross-cell coverage. Cell signals are still strong in the First Zhongshan Road, which is far from the Flower hall site. As the 425 cell is not configured as the Neighbor-Cell of cell 436 in the first sector of the Shuqianlu site located on the First Zhongshan Road, calls are easily dropped in this area.

The above figure shows the pilot Ec/Io route testing result on the First Zhongshan Road (affected by signals from the Flower hall site, Ec/Io in area A is very poor; call-drop rate in the area is high; however, the pilot strength of the area is good.)

Analysis of the call-drop reason As there is shadow fading, the occurrence of the following events can be detected

from the active set upgrading report. Cell2 is the best service area; Cell1 is deleted from the activation cell; Cell3 is not in the Neighbor-Cell list of Cell2; strong signals from Cell3 result in poor

Ec/Io; Poor Ec/Io results in call-drops.


Solution Add Cell3 into Cell2’s Neighbor-Cell list As Cell3 is in a far distance, it is not

expected to be a member of the active set in the problematic area;

Reduce the transmit power of Cell3 and increase its tilt angle in order to control its signal coverage range. At the same time, take into consideration the coverage range to be provided by Cell3.

Execute solution: Add the mechanical tilt angle of the antenna of Huachang site 425

cell; Add Huachang site 425 cell into the Neighbour-Cell list of

Shuqianlu site; Reduce the maximum transmit power, public channel power and

pilot channel power of Flower hall site 425 cell by 3dB.

Effect after optimization: After optimization, the pilot Ec/Io of area A is obviously improved. After optimization, there is no call-drop.



There is no strict definition for the high site. It is a relative concept.

It is not necessarily wrong to put the UMTS base station on the top of the hill.

The high site can easily receive uplink interference generated by other users.

The bigger the loads in the high site coverage area, the more possible the problem might occur.

If the network is vacant or lightly loaded, the effect of the high site is not obvious. But it still cause cross-cell coverage, pilot pollution and call-drop.


Suggestion In urban areas, buildings are densely located and the penetration loss is big; the

radio transmission environment is complicated and the NodeB coverage distance is small. Hence the antenna should not be put too high. According to the present building density and average height, the antenna height can be about 35m; it should be 10~15m higher than the average height of surrounding buildings. Ofcourse, the specific height of the antenna should be determined according to the local radio transmission environment.

In rural areas, population is relatively small and buildings are not densely located; distances between base stations are big. Hence the antenna should be high; in general, the antenna height in rural areas is around 50m and should be 15m higher than the average height of its surrounding.

In the sea, the radio transmission model is similar as the transmission model for free spaces. The radio transmission environment is good; radio electric waves can be transmitted to a far distance. The site can be located on a high hill (higher than 100m) in order to expand its coverage.

In deserts and Gobi areas, signals are transmitted to a farer distance than in ordinary plains. The antenna height is usually 60m or higher in order to expand the signal coverage area.

Content


Neighboring cell Case

The Neighbour-Cell list is a cell list that might be added into the active set;

Cells in the Neighbour-Cell list will be measured as whether they meet the requirement for soft handover or softer handover with the main service cell;

The number of cells in the Neighbour-Cell list is up to 32; Avoid missing Neighbour-Cells with best signals in the

Neighbour-Cell list.


The network planning tool can use proper algorithm to automatically plan the Neighbour-Cell list; such planning is always based on the interference among cells;

If the pilot signals of one cell is very strong but the cell is not added in the active set, signals of the cell will become strong interference;

Either single-directional configuration or bi-directional configuration might be adopted between Neighbour-Cells;

In setting the Neighbour-Cell list, take into first considerations about the cell interference and the cell’s possibility of becoming a main service cell of the UE;

The method of automatically creating the Neighbour-Cell list via the network planning tool can be regarded as an initial reference of the Neighbour-Cell list. Manual adjustment is needed. The Neighbour-Cell list should finally be optimized by using the route testing data.


According to repeated route tests, it is found that calls are usually dropped during the handover in the direction from the Flower hall site to the Yunshan Hotel site; in the opposite direction from the Yunshan Hotel site to the Flower hall site, no call-drop occurs.


Problem analysis According to testing data analysis, the section 20m from the call-drop venue is

mainly covered by signals from the third sector (scramble 426) of the Flower hall site instead of signals from the first sector (scramble 424) of the Flower hall site. The reason might be the third sector (scramble 426) of the Flower hall site is sheltered by a tall building in front of it; signals of this sector are reflected to the road segment of 20m between the Flower hall site and the Yunshan Hotel site. Check the Neighbour-Cell list; it is found that the third sector (scramble 414) of the Yunshan Hotel site has configured the third sector of the Flower hall site as an Neighbour-Cell, while the third sector (scramble 426) of the Flower hall site does not configure the third sector (scramble 414) of the Yunshan Hotel site as an Neighbour-Cell. This has caused a failure in single-directional handover and resulted in call-drop.

Solution Configure the third sector (scramble 414) of the Yunshan Hotel site as an

Neighbour-Cell of the third sector (scramble 426) of the Flower hall site. Effect after optimization

After the Neighbour-Cell is configured, route tests are made on the road segment between the Flower hall site and the Yunshan Hotel site. No call-drop occurs.


Summary In the network planning phase, the Neighbour-Cell list can be

automatically generated via the network planning tool. Optimization of the Neighbour-Cell list can be executed via route

tests and statistics analysis of the route testing data. The Neighbour-Cell list optimized via route test data statistics

analysis is a short Neighbour-Cell list. And if necessary, the preference sequence in the Neighbour-Cell list can be very clear.

By analyzing the route test data, Neighbour-Cells not configured in the Neighbour-Cell list via planning tool can be found.

Content


Cell reselection Case

Description In drive test, pilot Ec/Io value was normal in continuous call test.


But the pilot Ec/Io in cycling call test was poor. Between cycling voice calls, the UE was in idle mode. The reason of Poor Ec/Io was that cell reselection did not happen on time, as shown in the figure below.


Troubleshooting Procedure Firstly, checked parameter SIntraSearch and found it was set as "NO",

which meant it was invalid in intra-frequency reselection. SIntraSearch indicates the intra-frequency measurement threshold of cell reselection.

If Sx > SIntraSearch, UE will not perform intra-frequency measurements.

If Sx <= SIntraSearch, UE performs intra-frequency measurements.

Normally, Sx = pilot Ec/Io - Qqualmin

The smaller SIntraSearch, the easier intra-frequency measurement is triggered. On the contrary, larger one will make it more difficult to trigger the measurement of intra-frequency cells.


Troubleshooting Procedure Secondly, checked Treselections and found the value was 1s. To make

the reselection happen earlier, changed its value to 0. Do the drive test again. It showed that the cell reselection

happened more quickly at the fault location.

Content


Handover failure Case 1

Description UMTS external cells and neighbor cell relation were correctly

configured. The setting of reselection and handover parameters were suitable, and reselection of GSM or UMTS network was also normal. However, for CS service, handover from UMTS to GSM failed.


Cause Analysis In this situation, we traced signaling message of the subscriber

according to IMSI to find out the handover failure reason.


Cause Analysis According to signaling messages of the subscriber, when call drop

happened, the 2d event had already been triggered. Besides, the 3a event also reported normally.

Call failure occurred when the it was preparing for relocation. The message showed that failure cause was that encryption or integrity protection algorithm from the CN side was not supported.


Troubleshooting procedure Contacting with Core Network engineers and confirmed that the

MGW sent the RNC integrity algorithm to the MGW of the GSM network with transparent transmission. But GSM did not use encrypted algorithm, which caused relocation failure and handover failure.

After changing the CIPHER option to FLAG in the LAICGI table, the sent encrypted algorithm was removed. Further tests showed that handover was normal.


Description When the UE moves from the

coverage area on Shuqian Road site (PSC: 436) to that of Meihuacun Hotel site (PSC:434), signals on Shuqian Road site (PSC: 436) deteriorate due to the blocking of the dual-deck viaduct. However, the Meihuacun Hotel site (PSC:434) enters the active set slowly for the high threshold. Therefore, the handover success rate is low.


Adjust the handover threshold and Time to Trigger parameters of Event 1A and Event 1B: reduce the handover threshold and Time to Trigger parameters of Event 1A, so that cells with better signal quality can enter the active set as soon as possible; raise the handover threshold and Time to Trigger parameters of Event 1B, so that cells within the active set would be removed for sudden fading of signals.

Effect after the optimization: After the optimization, cell 434 on Meihuacun Hotel site can speedily enter

the active set and cell 436 on Shuqian Road site would be removed from the active set due to the sudden fading of signals. Drive test after the parameter adjustment shows that the success rate of handovers between Shuqian Road site and Meihuacun Hotel site is greatly improved.

Event Parameter Setti ng Before Opti mi zati on Setti ng Af ter Opti mi zati onHandover threshol d 2dB 4dB

Ti me to Tri gger 640ms 200msHandover threshol d 5dB 7dB

Ti me to Tri gger 640ms 1280ms

Event 1A

Event 1B

Content


UMTS to GSM handover failure Case

Description When voice calls were made in UMTS covered area in

one building, call drop happened very frequently when UE moving towards GSM covered area. In 10 call, 9 calls dropped.


Cause Analysis Firstly, check neighbor cell configuration. Because miss

configuring neighbor cell relation is one of common reasons to cause GSM/UMTS handover failure.

But it was confirmed that the GSM cell is in the UMTS cell’s neighbor cell list.

inside room

window


Cause Analysis Secondly, check coverage. It was found that the building had no

UMTS indoor distribution system, and was covered by outdoor UMTS NodeB. When UE moved inside, UMTS signal penetrated two iron doors to the UE, causing fast fading.


Case Analysis The call drop was probably caused by handover delay

because of UMTS signal fast fading. One of the solutions is to adjust handover parameters to make handover happen earlier and execute handover process more quickly.


Optimization measures Modified Cell Independent Offset (CIO) of the neighbor

GSM cell from 0 to 5 dB. The handover happened easier, but call drop still existed.

Modified 2d RSCP threshold from -95 dBm to -85 dBm, and then -75 dBm. The measurement of the GSM started earlier, but call drop still existed.

Modified GSM RSSI threshold from -90 dBm to -95 dBm. Handover to the GSM cell was easier, but call drop still existed.


Optimization measures Modified event 2d Time-to-Trigger from 640 to 320 ms,

and then to 0 ms. The measurement of GSM started easier, but call drop still existed. We changed it back to 640 ms.

Changed the measurement quantity from RSCP to Ec/Io, and changed event 2d Ec/Io Threshold from -24 dB to -10 dB. But call drop still existed.

Changed event 3a Time-to-Trigger from 5000 to 2000 ms. Handover happened more quickly, and the call drop problem relieved.

Furthermore, changed event 3a Time-to-Trigger from 2000 ms to 1000 ms. Handover happened more quickly, and the call drop problem was solved.

UMTS to GSM call drop Case

Description Voice call drop happened during Handover from UMTS to GSM.

Checking the signaling in drive test, it was found that inter-RAT measurement was started.


Case Analysis From previous signaling, inter-RAT measurement was

started and event 3a event was reported, but downlink handover message "handover from UTRAN command" was not received by UE.


Troubleshooting Troubleshooting level by level from BSC to the Core

Network. checked the BSC of another vendor, and found that the

BSC rejected the handover command from the Core network. And then check BSC data. The result was that the inter-RAT service handover function on the BSC of another vendor was not activated.

Content


Call drop Case 1

Problem It is found that the call-drop rate is very high on the seaside

express way from TRI002 to TRI004. According to the testing data analysis, the coverage distance of 404 is very short at the call-drop venue.

Call drop Case 1

Handling Idea To take a bird’s-eye view from the sky, it is found that

there are several tall buildings in front of the 404 cell.

Call drop Case 1

Problem analysis As the handover region is short and the call-drop venue

on the seaside road is close to the TRI002 site (only 400m), signals might be strong at first but disappear quickly. This can cause slow speed of strong signals of the adjacent 404 cell in adding the active set. It can also cause a lot of ping-pang handover and result in call-drop.

Call drop Case 1

Solution Optimize the handover

parameter: Adjust 1A and 1B event handover parameters so that adding events can easily occur and deleting events occur slowly and difficultly. The values of handover parameters 1C and 1D events are adjusted. Replacement threshold with strongest pilot is reduced; replacement observation duration is increased. The advantage of such adjustment is to enable high percentage of the user’s using strongest and stable scramble.

Event Setting before optimization

Setting after optimization

1A event

Reporting Range Constant

3 5

Hysteresis 3.5dB 2dB

Time to trigger 200ms 200ms

1B event

Reporting Range Constant

7 6

Hysteresis 3.5dB 4dB


1C event

Hysteresis 6dB 4dB


1D event

Hysteresis 6dB 4dB


Call drop Case 1

Effect after optimization According to the route testing after handover parameter

adjustment, the handover success rate on this section is greatly improved; the call-drop rate is reduced.

Call drop Case 2

Description In this case, the UE would move in the directions marked by the

red arrow in the following figure. If call drop happens, the two sites nearby would be marked as BKC0044U and BKCOO74U.

Call drop Case 2

The main serving cell of the UE is the third cell (SC53) of site BKC0074U, its Ec/Io is -9.83dB.

Call drop Case 2

As the UE moves on, the main serving cell changes to the third cell (SC48) of site BKC0044U, its Ec/Io is -10.31dB. Cell SC53 of site BKC0074U is removed from the active set and enters the monitoring set.

Call drop Case 2

After 1s, the signal quality of cell SC53 of site BKC0074U is stronger than cell SC48 of site BKC0044U, and the Ec/Io of SC48 reaches -2.39dB. Cell SC48 reports to Event 1A and tries to enter the active set again. At this moment, the pilot quality of the cell SC48 of site BKC0044U is very bad, with its Ec/Io down to -21.05dB. The UE reports to Event 1A, but cannot receive the handover command, then the call drops.

Call drop Case 2

Optimization Solution To avoid the condition that cell SC53 of site BKC0074U cannot

enter the active set after being removed, the value of CellIndivOffset(utranCell) of cell SC53 of site BKC0074U is changed from 0dB to 3dB to prevent the cell from being removed from the active set.

Then the signal quality of cell SC53 of site BKC0074U declines, with Ec/Io down to -13.23dB, which is worse than that of cell SC48 of site BKC0044U. Then, the main serving cell of the UE changes to cell SC48, but cell SC53 remains in the active set.

Call drop Case 2

Verification of Optimization Effect The UE moves in the arrow

direction in a call-hold mode, and its main serving cell is cell SC53 of site BKC0074U.

Call drop Case 2

In the end, the main serving cell of the UE changes back to cell SC53， as shown in the following figure, and no call-drop happens.

Content


Voice quality Case

Description When using iPhone to dial 181 for time inquiry service

at every location in the network, the subscribers sometimes can only hear noise. The problem did not happen frequently, usually one out of 200 calls.

After test, we found that both the pilot strength and quality of the serving cell were good. Besides, UE transmitted power was also normal, and SIR was stable.

Voice quality Case

Discovery and solution of voice quality problem Abnormality was found when checking BLER. In normal cases, the

number of received blocks should remain steadily around 100 (this value has been normalized). However, the figure showed that when the voice quality problem occurred, the number of received blocks was smaller than normal and kept fluctuating. Once this number stopped fluctuating and returns to 100, voice quality returned normal. Meanwhile, the number of wrong blocks remained 0, which meant that block error rate was 0. Thus, the possibility of downlink interference was ruled out. We checked RTWP and found it was at the normal level.

Then we opened the subscriber signaling tracing interface. After analysis, we found that packet loss occurred at the IUB interface, and many time adjustment frames were received at the IUB interface.

Voice quality Case

We all know that UMTS uses the receiving window to synchronize transport channels. If the transport channel synchronization frames sent by the RNC is within the receiving window, then other data should also be within the receiving window. In this case, the transmission in the transport channels of RNC and NodeB is synchronous. If the said frames are outside the receiving window, then the transport channels are not synchronous. In this case, NodeB needs to calculate the offset value and notify the RNC to adjust the sending times of data frames through the uplink synchronization frames on the transport channel in order to re-synchronize data. When establishing the transport channel, the RNC gives the starting point TOAWS and the ending point TOAWE of the receiving window.

Voice quality Case

The time frame adjustment means that the NodeB finds that the delay of some packets are outside the receiving window while synchronizing with the RNC, so the NodeB needs to repeatedly send time adjustment frames to the RNC for the RNC to change the frame sending time, so that NodeB can catch the desired data that is sent at more accurate times. Data that is still outside the receiving window outside adjustment is probably dropped.

Voice quality Case

We first attempted to modify the receiving window of NodeB, which did not solve the problem. Voice quality is related to call traffic heaviness, which illuminated us that the busiest service in the carrier's building is HSPA service. We changed the receiving window back to 16/8 and then reduced the number of HSDPA channels from ten to five. Thus, we forcibly lowered the traffic on the IUB interface. As a result, the voice quality returned normal.

Voice quality Case

We observed that the configured traffic for OMCR was 15 Mbps but the actual traffic was 5 Mbps. The traffic of two NodeBs together would no doubt exceed 10 Mbps. Therefore, the sent data exceeded the bearer capacity and large amounts of voice data were queuing in the transmission equipment. Some data was dropped when the waiting timed out, and some data was dropped by NodeB because it was outside the receiving window when reaching NodeB due to the long queue time. This explains why the number of received blocks decreased, i.e. voice packet loss occurred. Delay jittering explains the severity of jamming within the transport equipment. The more severe the jamming, the longer the queue and the longer the delay. Delay jitters as data traffic varies.

Voice quality Case

Improving call dropping or voice quality degradation through parameter modification

Parameters involved: MAXDLDPCHPWR, BLERTARGET When signal strength or quality reaches a threshold, call dropping

or severe voice quality problems may occur. If this phenomenon occurs in coverage holes and cannot be improved by RF adjustment, we can modify some parameters to make some improvements.

Call drop and voice quality are related with AMR channel power and downlink BLER. AMR channel power and downlink BLER are directly correlated with signal strength and quality. So when signal strength and quality cannot be improved, we can increase DLDPCH power and set a higher BLERTarget.

08 WO_NAST3014_E01_1 UMTS Network Optimization Case-70.ppt

Documents

Transcript of 08 WO_NAST3014_E01_1 UMTS Network Optimization Case-70.ppt