TEMS-Optimization and Log File Analysis in GSM
TEMS-Optimization and Log File Analysis in GSM
TEMS-Optimization and Log File Analysis in GSM
and
GSM
by Somer GOKSEL
Contents
1 INTRODUCTION ........................................................................................................................................04
5 CONCLUSION...............................................................................................................................................94
4
1. INTRODUCTION
Optimization will be more effective and successful if you are aware of what you are
doing. The point is that you should now where to start, what to do and how to do.
The optimization is to intend providing the best network quality using available
spectrum as efficiently as possible. The scope will consist all below;
• Finding and correcting any existing problems after site implementation and
integration.
• Optimization can not reduce the performance of the rest of the network.
Analyzing performance retrieve tool reports and statistics for the worst
performing BSCs and/or Sites
Disabled GPRS
Coverage holes
Drop Calls
Capacity Problems
Missing Neighbors
One–way neighbors
Ping–Pong Handovers
Equipment Performance
Faulty Installations
7
Plotting RX Level and Quality Information for overall picture of the driven
area
1.2.6. Recommendations
Proposing new sites or sector additions with Before & After coverage plots
BSIC changes
Attenuation Adds/Removals
MHA/TMA adds
8
1.2.7. Tracking
Verifying handovers
Verifying Sectorizations
Verifying coverage
Benchmarking
9
This document was prepared with TEMS screen shots from live examples of previous
experiences to guide RF Engineers on how to define/analyze problems or cases and optimize
network. After each case/problem demonstration, specific step to be taken will be defined and
appropriate recommendation will be given.
The document will be focusing on Drive Testing part of the Optimization Process and give
definitions on basic GSM principals, features and parameters when needed.
The readers of this document are considered to have basic knowledge of cell planning and TEMS
Investigation usage. Only little information will be given just to remember TEMS interface.
10
2. DRIVE TESTING
Drive testing is the most common and maybe the best way to analyze Network performance by
means of coverage evaluation, system availability, network capacity, network retainibility and
call quality. Although it gives idea only on downlink side of the process, it provides huge
perspective to the service provider about what’s happening with a subscriber point of view.
The drive testing is basically collecting measurement data with a TEMS phone, but the main
concern is the analysis and evaluation part that is done after completition of the test. Remember
that you are always asked to perform a drive test for not only showing the problems, but also
explaining them and providing useful recommendations to correct them. Please note that a
successful analysis should be supported by handling of network statistics from a statistics tool
(Metrica/NetDoc–NMS/SRP–OSS, etc..) and careful evaluation of coverage predictions from a
cell planning tool (Planet, DB–Planner, TEMs Cell Planner, etc..). Please see Figure 1 for a usual
view from TEMS.
Figure 1– TEMs Screen: TEMs gives great presentation options to the user like displaying
multiple windows of different indicators on the map. Theme properties will make you understand
easier by showing the serving cell on the map.
11
The information provided by TEMS is displayed in status windows. This information includes
cell identity, base station identity code, BCCH carrier ARFCN, mobile country code, mobile
network code and the location area code of the serving cell.
There is also information about RxLev, BSIC and ARFCN for up to six neighboring cells;
channel number(s), timeslot number, channel type and TDMA offset; channel mode, sub channel
number, hopping channel indication, mobile allocation index offset and hopping sequence
number of the dedicated channel; and RxLev, RxQual, FER, DTX down link, TEMS Speech
Quality Index (SQI), timing advance (TA), TX Power, radio link timeout counter and C/A
parameters for the radio environment.
The signal strength, RxQual, C/A, TA, TX Power, TEMS SQI and FER of the serving cell and
signal strength for two of the neighboring cells can also be displayed graphically in a window.
Layer 2 and 3 messages and SMS cell broadcast messages are displayed in separate scrollable
windows as can be seen in Figure 2. If desired, specific Layer 3 messages can be displayed.
By connecting an additional TEMS phone to a vacant serial port of the PC, data from two
networks can be monitored and logged at the same time. In this case, the data from the second
mobile phone is serving cell and neighboring cell data and radio environment parameters.
TEMS Investigation also can perform frequency scanning of all significant carrier frequencies.
The information presented is ARFCN, RxLev and, if successfully decoded, BSIC.
13
2.2.1. Accessibility
2.2.2. Retainability
TCH Call Drop rate is calculated by dividing total number of drop calls to
number of total TCH seizures and attempts. Total number of drop calls
contains all types of TCH drops including any radio related, user activated,
network activated, ABIS fail, A interface, LAPD, BTS failure or BSCU
reset drops. Please note that any TCH re–establishment should be
subtracted from TCH call drop rate as call is somehow able to continue.
Total number of TCH attempts and seizures will include any TCH seizures
for new calls and TCH to TCH attempts during Handover and number of
intracell handovers as well.
Access failures are the total number of unsuccessful TCH attempts which
is calculated by subtracting number of assigned TCH seizures from
number of TCH attempts – including the ones during handovers.
14
A powered on mobile station (MS) that does not have a dedicated channel allocated is defined as
being in idle mode (see Figure 3). While in idle mode it is important that the mobile is both able
to access and be reached by the system. The idle mode behavior is managed by the MS. It can be
controlled by parameters which the MS receives from the base station on the Broadcast Control
Channel (BCCH). All of the main controlling parameters for idle mode behavior are transmitted
on the BCCH carrier in each cell. These parameters can be controlled on a per cell basis.
Moreover, to be able to access the system from anywhere in the network, regardless of where the
MS was powered on/off, it has to be able to select a specific GSM base station, tune to its
frequency and listen to the system information messages transmitted in that cell. It must also be
able to register its current location to the network so that the network knows where to route
incoming calls.
Figure 3– Idle Mode Behavior: Cell Re–selection in Idle mode corresponds to handover in
Dedicated Mode. When a new call is set up on the MS, MS goes to Dedicated Mode.
The PLMN selection mechanism, the cell selection and reselection algorithms in addition to the
location updating procedure are the core of the idle mode behavior. The purpose is to always
ensure that the mobile is camped on the cell where it has the highest probability of successful
communication. In idle mode the MS will notify the network when it changes location area by the
location updating procedure. Thus, the network will be kept updated concerning which location
area the MS is presently in. When the system receives an incoming call it knows in which
location area it should page the MS, and does not need to page the MS throughout the whole
MSC service area. This reduces the load on the system. If the MS does not respond to the first
paging message, then the network can send a second paging message.
Sometimes MS does not camp on the best cell and needs to perform a cell re–selection process
before initializing the call (see Figure 4). This could be related to wrong Cell Reselection
parameters like CRO – Cell Reselect Offset, Cell Reselect Hysteresis, Temporary Offset or
Penalty Time used in C1–C2 criteria calculation.
The MS listens to the system information, compares the Location Area Identity (LAI) to the one
stored in the MS and detects whether it has entered a new location area or is still in the same
location area. If the broadcast LAI differs from the one stored in the MS, the MS must perform a
location update, type normal. The MS sends a channel request message including the reason for
the access. Reasons other than location updating can be for example, answering a page or
emergency call.
The message received by the BTS is forwarded to the BSC. The BSC allocates a signaling
channel (SDCCH), if there is one idle, and tells the BTS to activate it. The MS is now told to tune
to the SDCCH. The outcome of the procedure is that a radio resource connection is dedicated to
the MS. The procedure is therefore called RR connection establishment.
The MS sends a location updating request message which contains the identity of the MS, the
identity of the old location area and the type of updating. The authentication parameter is sent to
MS. In this case the MS is already registered in this MSC/VLR and the authentication parameter
used is stored in the VLR. If the MS is not already registered in this MSC/VLR the appropriate
HLR or the previously used MSC/VLR must be contacted to retrieve MS subscriber data and
authentication parameters. MS sends an answer calculated using the received authentication
parameter.
If the authentication is successful, the VLR is updated. If needed, the old HLR and old VLR are
also updated. The MS receives an acceptance of the location updating. The BTS is told to release
the SDCCH. The MS is told to release the SDCCH and switches to idle mode.
If the MS is moving in a border area between location areas, it might repeatedly change between
cells of different location areas. Each change of location area would require a location updating
to be performed, which would cause a heavy signaling load and thereby also increasing the risk
of paging messages being lost.
Cells bordering a different location area may have lots of location updating, and cells on a
highway probably have many handovers. In order to calculate the need for SDCCHs the number
of attempts for every procedure that uses the SDCCH as well as the time that each procedure
holds the SDCCH must be taken into account. The procedures are; location updating, periodic
registrations, IMSI attach/detach, call setup, SMS, facsimile and supplementary services.
Next step will be analyzing Call Set–up process. Being the start point and direct factor to
accessibility of the network, call set–up has great importance in GSM performance. Some basic
information on the procedure will be given. As Layer 3 messages will be our reference point
when defining problems during log files analysis, they will also be explained with their
appearance order during and after call set–up.
17
The cell selection algorithm tries to find the most suitable cell of the selected PLMN according to
various requirements. If no suitable cell is found and all available and permitted PLMNs have
been tried, the MS will try to camp on a cell irrespective of PLMN identity and enter a limited
service state. In this state the MS will be able to make emergency calls only. If the MS loses
coverage it will return to the PLMN selection state and select another PLMN.
After a cell has been successfully selected, the MS will start the cell reselection tasks. It will
continuously make measurements on its neighboring cells to initiate cell reselection if necessary.
For multiband MSs the strongest non–serving carriers may belong to different frequency bands.
The MS continuously monitors all neighboring BCCH carriers, as indicated by the BA list, in
addition to the BCCH carrier of the serving cell, to detect if it is more suitable to camp on another
cell. At least five received signal level measurement samples are required for each defined
neighboring cell. A running average of the received signal level will be maintained for each
carrier in the BA list. Provided that the MS is listening to the system information in the cell and
that it is registered in the MSC/VLR handling this cell, the MS can attempt to make a call.
First, radio connection between MS and network is established. Then MS indicates that it wants
to set up a call. The identity of the MS, IMSI, is analyzed and the MS is marked as busy in the
VLR. Authentication is performed as described for location updating. Then ciphering is initiated.
MSC receives a setup message from the MS. This information includes what kind of service the
MS wants and the number (called the B number) dialed by the mobile subscriber. MSC checks
that the MS does not have services like barring of outgoing calls activated. Barring can be
activated either by the subscriber or by the operator. If the MS is not barred, the setup of the call
proceeds. Between the MSC and the BSC a link is established and a PCM TS is seized. The MSC
sends a request to the BSC to assign a traffic channel (TCH). The BSC checks if there is an idle
TCH, assigns it to the call and tells the BTS to activate the channel. The BTS sends an
acknowledgment when the activation is complete and then the BSC orders the MS to transfer to
the TCH. The BSC informs the MSC when the assignment is complete. The traffic control
subsystem analyzes the digits and sets up the connection to the called subscriber. The call is
connected through in the group switch. An alert message is sent to the MS indicating that a
ringing tone has been generated on the other side. The ringing tone generated in the exchange on
the B subscriber side is sent to the MS via the group switch in MSC. The ringing tone is sent over
the air on the traffic channel.
When the B subscriber answers, the network sends a connect message to the MS indicating that
the call is accepted. The MS returns a connect acknowledgment, which completes the call set–up.
Figure 5– Call Set–up process: Please pay attention to the C/I appearances of hopping and non–
hopping cells on the chart. C/I for every hopping channel are displayed separately. This explains
how hopping deals better with interference, every other frequency in hopping list has different
effects from the interferer and this optimizes the overall speech quality minimizing fading dips
and reduces interference effect.
Call Set–up procedure starts with Channel Request Command and MS passes to Dedicated Mode
with this command. This channel request message is sent in random mode on the RACH
( Random access channel –Uplink only, used to request allocation of a SDCCH ) and the most
important part of the message is Establishment cause. The cause for channel request could be;
–Answer to paging
–Emergency call
–Call re–establishment
–Other services requested by the mobile user (originating call, supplementary service short
message)
–All Other cases
Below window dump in Figure 6 showing a channel request from TEMS is an example to an
originating call.
The Channel request command is followed by Paging request message (Figure 7) which is sent
on the CCCH (Common Control Channel) by the network to up to two mobile stations to trigger
channel access by these.
System Information Type 13 (Figure 8) message is sent to determine GPRS options of the cell
with the given ARFCN.
Immediate Assignment message in Figure 9 is sent on the CCCH by the network to the mobile
station in idle mode to change the channel configuration to a dedicated configuration while
staying in the same cell.
CM Service Request message in Figure 10 is sent by the mobile station to the network to request
a service for the connection management sub layer entities, e.g. circuit switched connection
establishment, supplementary services activation, short message transfer.
System Information Type 5 (Figure 11) is sent by the network to mobile stations within the cell
giving information on the BCCH allocation in the neighbor cells. When received, this
information must be used as the list of neighboring cells to be reported on. Any change in the
neighbor cells description must overwrite any old data held by the MS. The MS must analyze all
correctly received system information type 5 messages.
Class mark Change message in Figure 12 is sent on the main DCCH by the mobile station to the
network to indicate a class mark change.
Ciphering Mode Command message (Figure 14) is sent on the main DCCH from the network to
the mobile station to indicate that the network has started deciphering and that enciphering and
deciphering shall be started in the mobile station, or to indicate that ciphering will not be
performed. This message is followed by a Ciphering Mode complete message (Figure 15).
Call Set–up message (Figure 16) is sent, from either the mobile station or the network, to initiate
call establishment. It consists of below information elements;
–Protocol discriminator
–Transaction identifier
–Message type
–Repeat indicator: The repeat indicator information element is included immediately before the
first bearer capability information element when the in–call modification procedure is used.
–Bearer capabilities: In the mobile station to network direction, at least one bearer capability
information element must always be present. In the network to mobile station direction, the
bearer capability information element may be omitted in the case where the mobile subscriber is
allocated only one directory number for all services.
–Mobile identity: May be included by the calling mobile station to identify the calling mobile
station.
–Progress indicator: Included in the event of interworking or in connection with the provision of
in–band information/patterns.
–Signal: Included if the network optionally provides additional information describing tones.
–Calling party BCD number: May be included by the network to identify the calling user.
–Calling party sub–address: Included in the Mobile Station–to–network direction when the
calling user wants to indicate its sub address to the called user. Included in the network–to–
Mobile Station direction if the calling user includes a calling party sub–address information
element in the SETUP message.
–Called party BCD number: The called party BCD number information element is included by
the network when called party number information is conveyed to the mobile station. The called
party BCD number shall always be included in the mobile station to network direction.
–Called party sub– address: Included in the Mobile Station–to–Network direction when the
calling user wants to indicate the called party sub address. Included in the Network–to Mobile
Station direction if the calling user includes a called party sub address information element in the
SETUP message.
–Repeat indicator: The repeat indicator information element is included when the in–call
modification procedure is used and two low layer compatibility information elements are
included in the message.
28
–Low layer compatibility: Included in the MS–to–network direction when the calling MS wants
to pass low layer compatibility information to the called user. Included in the network–to–mobile
station direction if the calling user included a low layer compatibility information element in the
SETUP message.
–Repeat indicator: The repeat indicator information element is included when the in–call
modification procedure is used and two high layer compatibility information elements are
included in the message. The repeat indicator information element is not included when the
optional high layer compatibility information elements are omitted.
–High layer compatibility: Included in the MS–to–network direction when the calling MS wants
to pass high layer compatibility information to the called user. Included in the network–to–
mobile station direction if the calling user included a high layer compatibility information
element in the SETUP message. This information element may be repeated if the in–call
modification procedure is used. Bearer capability, low layer compatibility, and high layer
compatibility information elements may be used to describe a CCITT telecommunication service,
if appropriate.
–User–user: Included in the calling mobile station to network direction when the calling mobile
station wants to pass user information to the called remote user. Included in the network to called
mobile station direction when the calling remote user included a user–user information element in
the SETUP message.
Figure16– Set Up
29
System Information Type 6 (Figure 17) is sent on the SACCH by the network to mobile stations
within the cell giving information of location area identification, of cell identity and various other
information.
FACCH –Fast Associated Control Channel is used to transmit Handover commands, last
messages of call setup and call clearing messages. These messages are sent on TCH by using the
TCH in signaling mode. No speech or data is transmitted while the TCH is used as FACCH.
Measurement Report message (Figure 18) is sent on the SACCH by the mobile station to the
network to report measurement results about the dedicated channel and about neighbor cells. This
message is repeated for every new measurement report to generate neighbor lists and is the basis
for handover command.
Call Proceeding message (Figure 19) is sent by the network to the calling mobile station to
indicate that the requested call establishment information has been received, and no more
call establishment information will be accepted.
Assignment Command message (Figure 20) is sent on the main DCCH by the network to the
mobile station to change the channel configuration to another independent dedicated channel
configuration. Below are some definitions of information given in this message.
–Channel mode information element appears if the channel mode is changed for the channel
defined in the mandatory part of the message.
–Channel description information element appears in the case of a so–called intracell handover or
an assignment occurring after a call reestablishment if the MS carries two connections (on two
dedicated channels).The connection using the channel previously defined in the mandatory part
of an ASSIGNMENT COMMAND or HANDOVER COMMAND message shall use the channel
defined in the mandatory part of the ASSIGNMENT COMMAND message defining the new
configuration.The first indicated channel carries the main DCCH. The SACCH used is the one
associated with that channel.
–Channel mode 2 information element appears if the channel mode is changed for the channel
defined in the optional channel description information element.
–Mobile allocation information element appears in the case of frequency hopping. It applies to all
assigned channels.
Low signal level is one of the biggest problems in a Network. The coverage that a network
operator can offer to customers mostly depends on efficiency of network design and investment
plans. This problem usually pops up when building a new Network or as the number of
subscribers increases by the time resulting in new coverage demands.
Low signal level can result in unwanted situations that could directly lower the network
performance. Poor coverage problems are such problems that are really hard to solve, because it
is impossible to increase coverage by optimizing network parameters. Any hardware
configuration changes might improve the coverage a little.
Let’s have a look at some different cases to poor coverage related problems.
34
Figure 22– Low Signal Level: In areas where there are few sites and too many different types of
terrain structures like hills or obstacles those stopping the line of sight to the broadcasting
signal, there might be a lot of coverage holes or places with insufficient signal level. Pay
attention to the significant oscillation on the C/I affected by the drop of signal level.
35
Figure 23– Lack of Dominant Server: Signals of more than one cell can be reaching a spot with
low level causing ping pong handovers. This might happen because the MS is located on the cell
borders and there is no any best server to keep the call.
Figure 24– Sudden Appearance of Neighbors – Terrain Effect: Due to terrain or obstacles,
neighbors may pop up with high levels causing the BSC to give wrong handover decisions. In this
case, there won’t be a stable server, but the call will be handed to the neighbors for very short
period.
Figure 25– Sudden Appearance of Neighbors – Fast Moving Mobile Effect: When MS moves
very fast, the tester will see a lot of handovers and sudden changes on signal levels. This case
might happen when the MS user is driving fast on the highway. The serving time of the cell will
depend on the cell size and vary with hierarchical cell structure of the network. There seem to
happen too many handovers but this is due to fast moving mobile.
Figure 26– Sudden Decrease on Signal Level – Tunnel Effect: Tester may notice sudden
decrease on signal level when analyzing the log files. This will result in excessive number of
handovers. Before suspecting anything else, check if the test was performed on a highway and
that particular area was a tunnel or not. Signal level on the chart will make a curve rather than
unstable changes. Tunnel effect will most likely result in ping pong handovers.
Figure 27– Stable Behavior – The same Cell serving for a long Time: Looking to above view,
tester may think the serving cell’s coverage is very good and it’s serving through a long period.
Sometimes this may not be correct and this stable look may result in misunderstanding. Check if
the drive test vehicle was waiting for a red light or any traffic jam causing the vehicle to wait at
the same spot for a long time.
Figure 28– Oscillation on Hopping Channels Become more Significant with Low Level: Poor
coverage brings low quality and is a very significant sign of future drop calls. Low level on down
link signal strength can mostly occur because of low number of sites in the network, high
attenuation from the obstacles like buildings or hills, or high path loss caused by Rayleigh
Fading.
41
Figure 29– The same cell always popping up as the second strongest neighbor in the list
through a large area might show an overshooting cell. This kind of situations will
42
Figure 30– Ss levels of 3–4 cells are too closed to each other – This might point overlapping
cells: Other cells else than the one that suppose to serve at that particular area should be
coverage reduced by power reductions, downtilts or other configuration changes.
3.1.10. RX Levels of Many Cells are Almost the Same (Figure 31)
Figure 31– Signal Strength of All the neighbors are almost the same with each other. This
shows the network needs big optimization work because there are too many cells having
overlapping coverage. This will cause quality problems because of frequency reuse and
immediate action to optimize cell coverage should be taken.
Figure 32– Both Signal Strength and SQI are changing fast due to far away server being
blocked by obstacles from the terrain. The other way, signal from the server looses line of site
to the mobile because of a hill or something.
Figure 33– TEMS on Resume. Don’t worry, everything is fin: Although this looks weird, the
straight look in the chart is just because tester resumed recording log file to take a break during
the test. You will see a straight line for the period of time test was resumed.
46
Figure 34– Drop Call due to Bad Coverage: Call is dropped because of poor coverage. The
signal level goes down below the minimum signal level that system could carry on. Remember
this minimum level is much lower than RX Access Minimum Level to prevent on–going call from
dropping.
47
Figure 35– Access Failures After a Drop Call: Access failures can happen because of low level
below ACCMIN, bad quality or blocking in the target cell, or hardware failures. If you get a
blocked call message during call set–up, it is because the signal leveling the cell you are trying to
make call set–up is below ACCMIN which prevents MS to access the cell. ACCMIN is generally
set to –104dBm depending on sensitivity level of equipments and is referred during call set–up. A
low value of ACCMIN means that the coverage in idle mode is improved at the expense of the
risk of having an increased number of call set–up failures.
The best thing to do in case of low signal strength could be recommending new site additions. A
prediction tool with correct and detailed height and clutter data supported with a reasonable
propagation model could be used to identify the best locations to put new sites. If client is not
eager to put new sites because of high costs to the budget or finds it unnecessary because of low
demand on traffic, then appropriate repeaters could be used to repeat signals and improve the
coverage. Adding repeaters always needs extra attention because they can bring extra interference
load to the network. The received level in the repeater should be above –80dBm (or desired
limits) so that it can be amplified and transmitted again. The mobile should not receive both the
original and the repeated signals at the same area, cause signal from the repeater is always
delayed and it will interfere with the original signal. A repeater should not amplify frequencies
outside the wanted band.
If none of the above recommendations are accepted by the client, then cheaper and easier ways
should be followed. First things to be checked would be possible attenuation on the cells. Faulty
feeders–jumpers–connectors or other faulty equipment, high combiner loss, reduced EIRP,
decreased output power, the orientations and types of antennas, unnecessary downtilts, existence
of diversity and height of the site should be deeply investigated. Putting higher gain antennas,
increasing output power, removing attenuations, changing antenna orientations towards desired
area, reducing downtilts, replacing faulty equipment or usage of diversity gain could improve the
coverage.
Please note, amplifiers (TMA or MHA) could be used to improve uplink or compensate the loss
caused by long feeder. Be careful, because they will also amplify interfering signals and they will
be received at higher level.
49
Indicators collected from the network which give information about the speech quality are:
Let’s remember BER– Bit error Rate and FER– Frame Erasure Rate
expressions:
The speech quality is degraded by high BER for the air interface. The BER
and frame erasure ratio (FER) are dependent on a number of factors such
as fading and interference. Therefore a good cell planning is needed to
avoid co–channel interference, adjacent–channel interference, time
dispersion and other types of radio interference. The BER and FER
caused by the radio network is the most important speech quality degradation
factor. The degradation can be minimized by using the radio network
features DTX, Power control and Frequency hopping. The handovers while
moving from cell to cell will also introduce a speech quality disturbance.
50
3.2.2. Bad Quality due to Signal Strength – FER is Bad (Figure 36)
Figure 36– Bad Quality due to Signal Strength – Bad FER: As the signal strength drops down,
the quality of the call becomes worse being effected by interference and/ or fading. Consequently
the system becomes weaker to handle the interference. Notice that not only Rx Quality is bad, but
also FER is high. SQI is still within acceptable limits. That’s why we check all RX Quality, FER
and SQI when analyzing interference problems. System will face bad RX Quality, drop calls and
ping pong handovers in such environments.
Bad RxLev
Bad RXQual
Bad FER
51
Figure 37– Bad Quality due to Signal Strength – FER is OK: The difference of this case from
the previous is only the difference in FER. Signal strength is also bad in this, but FER is still fine
which means there is no obvious interference in the area. The area in this case should most
probably be a flat area without any obstacles to create reflection and the site density should not
be dense or re–use of frequencies is good to prevent any co–channel interference.
RX Quality is bad
FER is fine
52
3.2.4. SQI
The need for speech quality estimates in cellular networks have been
recognized already in the GSM standard, and the RxQual measure was
designed to give an indication of the quality.
Figure 38– Collusion of MA list causing low C/I: The collusion of frequencies with neighboring
cells MAIO list frequencies become more significant with dropping signal level. To prevent this
kind of interference, MAIO lists of neighboring cells should be properly planned or MAIO step
could be used.
54
One must remember that interference does not only appear on the down–
link, but also on the uplink. If interference on the downlink is experienced
in one cell, there is a risk that you would have this problem on the uplink as
well. Not in the same cell, but in the interfering cell. However, downlink
interference is normally a larger problem than uplink interference.
The GSM specification states that the signal strength ratio, C/A, between
the carrier and the adjacent frequency interferer, A, must be larger than –9
dB. However, adjacent channel interference also degrades the sensitivity as
well as the C/I performance. During cell planning the aim should be to
have C/A higher than 3 dB.
C/A > 3 dB
55
Figure 39– C/A Interference: This is a good example of adjacent channel interference. Although
our first and prioritized concern in frequency planning will be Co–Channel interference, some
cases where you use adjacent channels in neighboring cells might bring you quality problems
and even handover failures.
By using all or most of the received power, instead of only the direct
signal, there is a larger probability to decode the information. This may be
considered as a type of time diversity.
There are couples of things to remember when dealing with time dispersion
problems:
• Not all reflections are harmful, only low attenuated reflections that are
delayed more than the equalizer can handle.
• The further away the objects are, the weaker the reflections are. Hence,
objects just outside the ellipse are the ones most likely to cause the
majority of problems.
• For problems to occur, there will, in most cases, be a line of sight from
both mobile and base station to the reflector. Small cells in an urban
environment are not likely to encounter any time dispersion problems.
The distances (time delays) are short, and the buildings will shield
from, and also scatter the reflections.
Reflections are a function of the location of the site and mobile. Therefore
none of the radio network features are of much help in combating
reflections. Frequency hopping does not improve on BER that is due to
reflections (C/R).
The most radical solution is to move a site. A site could be placed near the
reflecting object to prevent time dispersion. Easier and sometimes just as
efficient is to modify the antenna arrangement, either in azimuth
(horizontally) or by tilt (vertically). Those measures may also be used to
improve C/I. For down–tilt to be efficient, antennas with narrow vertical
lobes would be needed to avoid illuminating the reflector.
57
Figure 40– Bad Quality due to Time Dispersion: This is a good example to reflection from a
source of reflection whether a hill or a tall building. In this case, the reflected signal R is even
stronger than the original signal C. Please pay attention to Time Advance value although MS is
less than 1 Mile far to the cell.
The first and most simple solution is to use more power at the
transmitter(s), thus providing a fading margin. Another way to reduce the
harm done by Rayleigh fading is to use space diversity, which reduces the
number of deep fading dips.
59
3.3. Handover
Mobiles in communication with the network will continuously perform measurements on serving
and neighboring cells. The measurement results are sent to the BSC and used in the locating
procedure to make decisions about handover. There are different types of handovers:
· Intra BSC handover: The new and old cells both belong to the same BSC. The BSC can handle
the handover on its own.
· Inter BSC handover: The new and old cells belong to different BSC but the same MSC/VLR. In
this case the MSC/VLR must help the BSC to carry out the handover.
· Inter MSC handover: The new and old cells belong to different MSC/VLR. The serving
MSC/VLR must get help from the new MSC/VLR to carry out the handover.
· Intra cell handover: No change of cell but of connection within the cell.
During a call, the serving BSC decides that a handover is necessary. The handover procedure
happens in this way:
• The serving BSC sends Handover Required, including the identity of the target cell, to the
MSC.
• The new MSC allocates a handover number (ordinary telephone number) in order to reroute
the call. A handover request is sent to the new BSC.
• The new BSC, in cases where there is an idle TCH in the target cell, tells the new BTS to
activate a TCH.
• The new MSC receives the information about the new TCH and handover reference.
• The TCH description and handover reference is passed on to the old MSC together with the
handover number.
• The old MSC is informed (via, the new BSC and the new MSC) about the detection of HO
bursts. The new path through the group switch in the old MSC is set–up.
• A handover complete message is sent from the MS. The new BSC and MSC inform the old
MSC. The old MSC informs the old BSC and the old TCH is released.
The originating MSC retains the main control of the call until it is cleared. This MSC is called the
anchor MSC. Because the call entered a new LA the MS is required to perform a location
updating when the call is released. During the location updating, the HLR is updated and sends a
Cancel Location message to the old VLR telling it to delete all stored
information about the subscriber.
– RXQUAL
– RXLEV
– DISTANCE
– PBGT
61
Figure 41– Handover in Layer 3 Messages: Before ho takes place, system needs to decide the
best candidate. First it repeats consecutive measurements to rank the cells according to HO
algorithm. Please note that HO algorithm in different vendors systems or even in operators using
the same equipment could be different. Some systems might rank the cells looking to their signal
strength or some can rank them looking to their Path Loss or some can use both.
Let’s have a look at Layer 3 messages to understand better about the Handover process.
Synchronization Channel Information (Figure 42) message is sent on the SCH, which is one of
the broadcast channels. Its purpose is to support the synchronization of a MS to a BSS and is
repeated for many times during the call.
System Information Type 5 message (Figure 43) is sent on the SACCH by the network to mobile
stations within the cell giving information on the BCCH allocation in the neighbor cells. This
downlink information will later form neighbor lists. When received this information must be used
as the list of neighboring cells to be reported on. Any change in the neighbor cells description
must overwrite any old data held by the MS. The MS must analyze all correctly received system
information type 5 messages.
63
Handover decision process continues with System Information Type 6 message (Figure 45)
which is sent on the SACCH by the network to mobile stations within the cell giving information
of location area identification, of cell identity and various other information.
Handover command message (Figure 46) is sent on the main DCCH by the network to the
mobile station to change the dedicated channel configuration. The message contains Cell
description, Channel description, Handover reference, Power command, Synchronization
indication, Cell channel description, Channel mode, Channel description, Channel mode 2,
Frequency channel sequence, Mobile allocation and starting time information.
Cell channel description information element appears if frequency hopping is used on the new
cell. Channel mode element appears if the channel mode is changed for the channel defined in the
mandatory part of the message. Channel description information element appears if the MS
carries two connections (on two dedicated channels). The connection using the channel
previously defined in the mandatory part of an assignment command or handover command
message shall use the channel defined in the mandatory part of the handover command message
defining the new configuration.The first indicated channel (i.e. in the mandatory part) carries the
main DCCH. The SACCH used is the one associated with that channel.
Channel mode 2 element appears if the channel mode is changed for the channel defined in the
optional channel description information element. Frequency channel sequence element is a
combination of mobile allocation element and cell channel description element. It is designed to
allow the sending of the handover command in one signaling block for systems using frequency
hopping. If this element is present, then the cell channel description and mobile allocation
information elements are not required. Mobile allocation information element appears if
frequency hopping is used on the new cell. If it appears, it applies to all assigned channels. This
65
information element cannot appear if the cell channel description information element is not
present. Starting time information element appears if a frequency change is in progress. It refers
to the new cell time.
Handover Access message (Figure 47) is sent in random mode on the main DCCH during a
handover procedure.
Handover Complete message (Figure 48) is sent on the main DCCH from the mobile station to
the network to indicate that the mobile station has established the main signaling link
successfully.
If handover is not successful for some reason, then comes a Handover Failure message. This
message is sent on the main DCCH on the old channel from the mobile station to the network to
indicate that the mobile station has failed to seize the new channel.
67
When signal strength difference between serving cell and neighbor cell
exceeds Power Budget Handover margin which is set in Handover
parameters, the call is handed over to the neighboring cell. This margin is
usually set to 3 to 6 dB. Power Budget HO feature should be enabled for
this type of Handover.
In case of ping–pong handovers between the same two cells, power budget
handover margin between the two could be increased to reduce number of
handovers. Margin should be decreased if faster handover decision is
wanted.
Please keep in mind that adjusting power budget handover margin between
two neighbors is something you have to pay extra attention. If it is not set
correctly, there is high risk of interference. The strongest cell will not serve
in the cell border area resulting C/I (Carrier to Interference ratio) to get
bad.
Another risk will be stepping stone effect. Let’s investigate this effect with
an example:
A to B HO Margin PBGT = 0
A to C HO Margin PBGT = +10
B to C HO Margin PBGT = 0
In an area where signal strength of cell A is the weakest and signal strength
of cell C is strongest, the HO attempt might happen in such an order that
cell A will first hand the call to B and cell B will immediately hand it to
cell C. In this case, cell B will be used as a stepping stone to make a
handover to the strong cell C. You will observe ping–pong or unnecessary
handovers.
Figure 50– Quality Handover: See a handover was performed to a better quality cell just after
experiencing quality problems.
70
Any quality problem when downlink signal level is higher then HO Thr
Interference DL causes a handover. This level is generally set to –80dBm.
Any quality problem when uplink signal level is higher then HO Thr
Interference UL causes a handover. This level is usually set to –80dBm.
Macro site which is defined as umbrella will shift all the TCH traffic to
small sites until signal level of small site becomes worse then HO Level
Umbrella RX Level parameter. This parameter could be set from –80 to –
90dBm. HO Margin PBGT parameter should be set to 63 maximum to
prevent any power budget handover.
If an overshooting site is needed to hand its traffic after some distance from
its origin, MS Range Max value could be adjusted to limit the serving area
of the site.
Intra–cell handover decision is given when serving cell is the best cell and
quality is worse then 4 and signal strength is lower then –85dBm.
Figure 51– Intracell Handover, Changing Rate from Full Rate to Half Rate
72
Figure 52– Intracell HO Based on Quality: This is an example of intracell handover after
experiencing quality problems.
73
There is still a way to save the call and hand it to one definite neighbor that
is known to be the best handover candidate in such cases. This is done by
handing the call to a chained cell whenever a threshold is reached in Uplink
direction. The handover candidate has to be defined as Chained Adjacent
Cell to the target cell to take the call regardless of any other neighbors in
the area. This threshold is set with the parameter HO Thr Rapid Lev UL. Its
value is –110dBm by default.
Figure 53– Short Neighbor List: Sometimes TEMs will help you notice possible problems like a
short neighbor list. If you see a couple of cell is the neighbor list of a cell when analyzing your
log files, check if that cell really does not need.
75
Always keep in mind that all power related parameters need to be correctly
set. Otherwise the handover (HO) attempts will be done in a wrong place.
There will be such cases that you will notice handover process taking place
a little late. There could be couple of reasons to that. First thing to check
will be handover margins between the neighbors. If margins for level,
quality or power budget handovers are not set correctly, handover will not
take place at the right time. If margins are too much, handover will happen
late, vice versa.
Figure 54– Late Handover: Could be because of incorrect setting of handover margins or
hierarchy between the cells or fast moving mobile.
77
Sometimes you will see a good handover candidate in the neighbor list but
handover will not take place and call will drop. Although that neighboring
cell with a very good signal level appears to be a neighbor, in reality it is
not. Just because the serving cell has another neighbor with the same
BCCH, this cell appears in the list.
There will be cases you will notice handover is not taking place although
two cells are defined as neighbors to each other. If neighbors BCCHNO is
missing from serving cells MBCCHNO list, MS will not monitor or report
the neighbors BCCH frequency. This will prevent HO attempt to the
neighbor.
During calls the MS reports the BCCH, BSIC and Signal strength of the six
strongest neighbors to the BSC. BSC has to map the neighbors
BCCHNO&BSIC combination to neighbors CGI (Cell Global Identity –
combination of MCC–MNC–LAC and CI). Therefore all neighbors of
same cell shall have a unique BCCHNO&BSIC combination.
Otherwise the BSC will start a HO attempt always to the first neighbor will
matching BCCHNO&BSIC combination even though the MS may have
reported another neighbor having the same combination.
83
If cell B has a lake of strong coverage inside cell A and cell B is not a
neighbor and cell C is a neighbor and the BCCHNO&BSIC combination is
the same in cells B and C, then BSC will start a HO attempt to cell C, when
the Ms has monitored/reported the cell B.
Please remember there will be rare cases where planner will need one–way
neighbor relations.
Figure 60– Handover not Happening and Dropped Call: The call is not handed to the other cell
although the level of the neighboring cell seems to be good enough. This example looks like fake
neighbor case. After making sure all handover parameters between these two cells are correctly
set, you should look for a neighboring cell to the serving cell with the same BSIC and BCCH
combination. The cell that appears in the neighbor list is not a neighbor to the serving cell in
reality. Just because there is another neighboring cell with the same BCCH, TEMs measures this
BCCH and lists as a neighbor. Handover will not happen because those two are not neighbors in
reality.
85
Figure 61– Handover Failure: Handover attempt was failed and the call returned back to its all
channel.
86
If the radio link fails after the mobile sends the Service Connect Complete Message then it is
considered a dropped call. Dropped call analysis can consume a considerable amount of time.
Using good post–processing analysis tools, the root cause of some of the drops can be determined
from mobile data alone. However, there will be cases where the cause cannot be reliably
confirmed unless system data is also used.
Calls often drop when strong neighbors suddenly appear. When the mobile is suddenly
confronted with a strong new signal, or when the signal it is using takes a sudden deep fade, it
will have poor C /I o and high forward FER. The call will drop unless it gets help quickly.
Using a post– processing tool, display a map of the locations of dropped calls that exhibit
symptoms of poor coverage. Verify this type of drop is not occurring in good– coverage areas. If
so, suspect and investigate hardware at the serving site.
Use the prediction tool to help identify other strong signals reaching the drop areas.
Another technique is to examine the dropped call message files and identify the BTS from which
the sync channel message is received immediately after each drop (this will be the cleanest pilot
the handset sees at that time). This could be achieved by analyzing Layer 3 messages in log files
or running traces from NMS/OSS.
• TCH radio drops are the drops that occurred due to summation of radio and ABIS reasons.
• TCH non–radio drops are the drops that occurred due to summation of network management,
BSCU reset, BTS fail, LAPD failure, user and A interface.
• TCH handover drops are the drops that occurred in handover phase while the call tries back to
old serving channel but fails and drops. These drops may occur due to RF, ABIS and A
interface reasons.
88
Radio Failures
Transcoder Failures.
Lapd Failures
BTS failures
Figure 63– Dropping Call due to MS Stuck on Overshooting Cell: In this example, MS is stuck
on an overshooting cell which is far away. You may check Time advance value in the Radio
Parameters window. MS can not handover to another cell because this overshooting cell was not
defined as neighbor to the cells nearby as expected. The only way to leave that cell and restart
call on a nearby cell will be possible after the call is dropped.
90
4. REPORTS of ANALYSIS
Below are some example recommendation reports those were provided to the clients during
previous optimization projects.
A Drive Test was performed in Downtown Seattle on 05/21/02 by WFI team. The selected route
starts with AT&T Eastlake Office, includes most of the major and primary roads in the
downtown area and ends with China town in the south. Two MSs were used during the test. One
was acting like an ordinary Mobile phone on dedicated mode and the other was scanning the
1900MHz frequency band. RXQual & RXLev Plots are attached for your reference. The
following recommendations are being given after analyzing the drive test data:
4– Northlake 122B is overshooting and causing interference on Blanchard 6th 374B. They have
the same BCCH (608). Increasing the downtilt on Northlake is recommended.
Woodlandpark 116A is overshooting in the northern downtown as well. It has the same BCCH
(595) with Alaskan Way 211B and causing low speech quality. Decreasing EIRP on
Woodlandpark 116A by 3dB will prevent it overshooting. Alaskan Way 211B is also interfered
by Spokane Street 178C.
University District 137B is overshooting near Dennyway and interfering with Blanchard 6th 374A
(594). Retune recommended.
5– 4th & Pike 172B is being interfered by Harbor Island 175A (591). Uptilt on Harbor Island
175A is recommended.
Union#2 210A is being interfered by White Center 184A (593). White Center should be
sectorized and the antenna facing to downtown should have reduced power.
6– Proposed site Spring&3rd will bring a best server to 3rd & Marion Street and prevent
Waterfront 186B serving there.
7– Other proposal First&Columbia will serve on 2nd Ave and prevent Safeco Field serving out
there. It will as well prevent Ping–Pong handovers between International District 375C / Safeco
Field NE 213A and Key Tower 195B.
8– Increasing EIRP for Pier70 204A by 6dB is recommended. This will pop it up on the area
where 2nd & 3rd Ave starts. It will also prevent Ping–Pong handovers between Harrison 173A and
Pier70 204A. Harrison 173A is interfered by Interbay 188A (600). Retune recommended.
9– Safeco Field and International District are causing many interBSC handovers in the area. If
BSC border could be shifted to south by moving these two sites from BSC5 to BSC2, HO
performance will be improved.
92
A Drive Test was performed on 06/10/02 by WFI team. The selected route covers all around
Lake Sammamish and some part of Bellevue on the west specifically near site Crossroads. Also a
trouble ticket of a customer complaint was taken with and that problematic area was driven. Two
MSs were used during the test. One was on dedicated mode and the other was scanning the
1900MHz frequency band. RXQual & RXLev Plots are attached for your reference. The
following recommendations including site configuration changes proved by before and after plots
are being given after analyzing the drive test data:
1– Sunset Village should be sectorized to have better sector gain and improve coverage. The
proposed sectorization is attached in a plot.
2– Site Issaquah is overshooting and overlapping with West Issaquah and South Sammamish
mostly. Rotating sector A from 50 degrees to 30 degrees with keeping the downtilt but
exchanging antenna type with 7250_05_6deg_1900 that has 6 degrees electrical downtilt. The
possible coverage loss after this modification on southeast cost of Lake Sammamish will be
compensated after appropriate modifications to West Issaquah and South Sammamish sectors
described below.
3– Rotating the antennas in South Sammamish Sector C from 340 to 310 degrees and changing
antenna type with 7250_02_1900 for higher gain is recommended. This will help the coverage on
Southwest coast of Lake Sammamish.
Rotating Sector A antennas from 115 to 80 degrees and exchanging the antenna type to a higher
gain antenna is recommended. This will reduce Ping–Pong handovers between South
Sammamish, Issaquah and West Issaquah and improve the coverage in the area.
4– Rotating West Issaquah Sector C from 355 to 10 degrees will help to have a better coverage
on southeast of the lake.
5– Eastgate should be sectorized to have better coverage on East Side of the Lake Sammamish
where a customer complaint has been reported. In this area drop calls due to poor coverage was
observed. The proposal for the sectorization is given in a plot with the following site data:
*Sector A facing to 20 degrees with 0 degrees downtilt on a 7250_05_6deg_1900 type of
antenna having 50 feet’s height and 50dBm EIRP.
*Sector B facing to 100 degrees with 3 degrees downtilt on a 7250_02_1900 type of
antenna having 54.5 feet’s height and 55dBm EIRP.
*Sector C facing to 210 degrees with 0 degrees downtilt on a 7250_05_6deg_1900 type of
antenna having 50 feet’s height and 50dBm EIRP.
6– We were told not to make any changes on Site Gene Beal, because it is a VIP site. If we could
only be able to increase EIRP on both of the sectors, than we could have a better coverage in the
area. If it could be done with the appropriate hardware configuration, adjacencies of this site
93
should be rechecked. Currently the level is so poor in the area and even Downtown Redmond is
serving there.
7– Overlake Sector A does not handover to Sector B even with level difference of 7 to 10dB.
This is the same case with Factoria Sector B and C.
8– Lack of coverage and poor quality with Ping–Pong handovers on North–Up Way on East of
156th Ave were observed. Changing the antenna type to type 3 for Overlake Sector A and reduce
its downtilt from 5 to 2 degrees is recommended. Also attenuation reduction to 0 is recommended
on Crossroads (All Sectors).
10– Drop call on NE 8th street near 156th Ave was observed. South Sammamish 428C is serving
here and its heavily interfered by Crossroads 483B (BCCH 602). It is strongly recommended to
add Crossroads 483A as neighbor to South Sammamish 428C.
11– Drop call on NE 4th Street on west of 156th Ave NE is observed. It is recommended to add
Crossroads 483B as neighbor to West Issaquah 251C.
12– Lack of coverage and poor quality with Ping–Pong handovers near intersection of 148th Ave
NE and NE 8th Street is observed. It is recommended that Lake Hills which is an Omni site to be
sectorized with the high gain antennas facing this intersection.
94
5. CONCLUSION
Optimization is a process that never ends. You can always try improving the quality of the
Network by making site configuration changes or changing parameters or try to understand what
is happening in the network by looking into stats. Everybody knows that there is a lot that could
be done by means of optimization but it is always hard to start. People sometimes get lost in stats,
sometimes struggle with parameters or log files.
Below is a chart (Figure 64) that shows how you could start doing optimization and the steps to
be followed. I am sure you will find vital information on the total process. Hope this works..