Alcatel Optimization

© Alcatel University - 8AS902001485 VH ZZA Ed.02 Page 1.1

© Alcatel University - 8AS 90200 1485VH ZZA Ed.021.1

Introduction to Radio Fine Tuning

BSS release B8

B8

TYPICAL MODULE STRUCTURE objective(s)

theoretical presentationtraining exercises and/or cases study + feedback


1.2

PROGRAM

1 TYPICAL RADIO PROBLEMS

2 ALGORITHMS AND ASSOCIATED PARAMETERS

3 OTHER ALGORITHMS

4 ALGORITHMS DYNAMIC BEHAVIOR

5 CASE STUDIES

B8





1.4

1 TYPICAL RADIO PROBLEMS Session presentation

▼Objective: to be able to characterize typical radio problems in order to trigger an intervention of the appropriate team

▼Program:

1.1Theoretical presentation

1.2 Coverage problem

1.3 Interference problem

1.4 Unbalanced power budget problem

1.5 TCH Congestion problem

1.6 Deducing the right team for intervention

1.7 Exercises

S1: TYPICAL RADIO PROBLEMSS2: ALGORITHMS AND ASSOCIATED PARAMETERS

S3: OMC-R RADIO PARAMETERS

S4: ALGORITHMS DYNAMIC BEHAVIOR

S5: CASE STUDIES




1.1 Theoretical presentation

Theoretical presentation

Coverage problem

Interference problem

Unbalanced power budget problem

TCH Congestion problem

Deducing the right team for intervention

Exercises


1.6

▼ Several sources of information can alert RFTM team:

�QoS indicators�Customers complaints�Drive tests�Other teams information (NSS statistics)

▼ As many symptoms are common to several causes, it can be necessary to:

�Consolidate standard sources of information�Carryout specific examinations�Deduce the appropriate team for intervention

1.1 Theoretical presentation Justification




1.2 Coverage problem


Coverage problem





Exercises


1.8

▼ Definition: Bad coverage

� A network or cell facing coverage problems presents a bad RxLev and RxQual in the same time on some areas.

▼ Symptoms:

� Customers complain about dropped calls or/and “no network”

� OMC QoS indicators

� TCH failure rate� Call drop rate� Low proportion of better cell HO� High rate of DL quality HO

� A interface indicators

� High rate of Clear Request messages, cause radio interface failure

1.2 Coverage problem Definition and symptoms

▼ No information is available on non-covered parts of the network, as there are non-mobiles making calls over there!▼ Nevertheless, cells in border of non-covered zones do have a particular behavior:

▼ Cell A will mainly perform Better Cell handovers towards its neighbors, whereas cell B, bordering the non-coverage area, will perform emergency handovers for MSs exiting the network.

� For these MS, mainly DL Quality HO will be triggered:� DL because MS antenna is less efficient than BTS one,� Quality rather than Level since Qual has a greater priority in Alcatel HO causes.

AB


1.9

▼Depending on the information sources you have:� Radio Measurement Statistics (RMS) –

� (RxLevel , RxQuality) matrix�Radio Link Counter S vector�Number of calls with DL/UL bad coverage (bad RxLev,

bad RxQual) � Abis interface (for example with COMPASS)

� bad quality > 5%� bad level RxLev < - 95 dBm and RxQual > 4

� OMC-R or A interface� unexpected high traffic, induced by call repetition

� Billing information�High recall rate detected

1.2 Coverage problem Examination

▼ RMS: new PM type in B7� Provides statistics from any area in the network which are available at any time.� Cost-effective.� Easier and cheaper to perform than Drive test or Abis Trace.� The operator can tune 54 parameters (based on RxLev, BFI, C/I, Radio Link Counter S, Path Balance, etc.) to

define up to 16 templates (depending on cell type – rural, urban, etc. – for example).� Trigger from the OMC-R.

� NPA can save up to 15 days of RMS for the complete network.� Templates can be designed in RNO.� Result reports are available in RNO and NPA.


1.10

▼ If the actual coverage is not the one predicted by RNP tools

� check antenna system� increase or decrease antenna down-tilt� check BS_TXPWR_MAX

� to be increased if value different to RNP power budget

▼ If the actual coverage is OK compared to the predicted ones

� indoor traffic, to be handled by specific means� if black spot close to cell border, ease outgoing HO

1.2 Coverage problem Typical causes


1.11

▼Example of an Abis trace analysis

Frequency RxLev_UL RxLev_DL RxQual_UL RxQual_DL Path_loss_UL Path_loss_DL delta_Path_loss delta_quality AV_MS_PW R nb_of samples119 -89.29 -84.67 0.42 0.43 123.82 123.67 0.15 -0.01 34.53 307492 -89.77 -89.09 0.41 0.38 124.87 128.09 -3.21 0.03 35.11 10253

111 -83.15 -79.15 0.17 0.33 116.05 121.22 -5.16 -0.16 32.9 5339

DISTRIBUTION OF UPLINK QUALITYFrequency Qual0 Qual1 Qual2 Qual3 Qual4 Qual5 Qual6 Qual7 Bad_Quality

119 86.50% 3.19% 2.50% 2.57% 1.92% 2.08% 0.98% 0.26% 3.32%92 88.11% 1.82% 1.91% 2.51% 2.14% 2.17% 1.15% 0.19% 3.51%

111 77.70% 4.30% 4.30% 4.36% 3.56% 3.56% 1.70% 0.17% 5.43%

DISTRIBUTION OF DOW NLINK QUALITYFrequency Qual0 Qual1 Qual2 Qual3 Qual4 Qual5 Qual6 Qual7 Bad_Quality

119 88.29% 1.82% 2.05% 2.37% 1.30% 1.46% 1.76% 0.94% 4.16%92 87.50% 2.98% 2.60% 2.43% 2.11% 1.14% 0.74% 0.50% 2.38%

111 71.30% 3.82% 4.02% 4.89% 4.16% 4.30% 4.23% 3.16% 11.73%

1.2 Coverage problem Investigation with Abis trace (1/2)

▼ It could have been coverage problems if this trace was made for 3 mono-TRX cells. In this case, the 3 lines are uncorrelated. Anyway, delta path loss of frequency 111 is greater than 5dB, showing a problem on this TRX.

▼ If this is a 3-TRX cell, it cannot be a coverage problem as the three TRXs are not impacted. It will be either interference or malfunction of one TRE.

▼ If the trace is done on 3 mono-TRX cells, in that case, it could be a coverage problem. Be careful when interpreting this result table: even if average levels in the UL and the DL are high and a lot of Quality problems are seen, nobody can say that samples with bad quality have a good level ! The level seen is just an average…

▼ One should have a look to the next slide…


1.12


5 6 -8 8 .0 0 3

7 3 -9 5 .3 3 3

1 1 3 -7 1 .0 0 1

1 6 -8 0 .0 0 1

1 2 3 -8 0 .0 0 1

B C _ D L : 1 1 5 3 .7 4 % < R x L e v _ S e r v in g > = -1 0 2 .1 7 d B m

N e ig h _ C e l l_ N b B S I C < L e v > S a m p le s

0 2 -1 0 0 .5 3 5 7

1 0 2 -9 8 .7 1 4 5

5 6 -9 8 .0 3 3 4

7 3 -9 8 .6 1 3 3

F r e q u e n c y :9 2

N u m b e r _ U L :1 0 2 5 3

N u m b e r _ D L :1 0 2 5 3

I n t_ U L : 2 0 .0 2 %

B C _ U L : 3 5 8 3 .4 9 %

I n t_ D L : 0 %

B C _ D L : 2 4 4 2 .3 8 % < R x L e v _ S e r v in g > = -1 0 6 .1 7 d B m


0 2 -1 0 4 .6 4 6 7

1 5 -1 0 7 .5 0 4 8

F r e q u e n c y :1 1 1

N u m b e r _ U L :5 3 3 9

N u m b e r _ D L :5 3 3 9

I n t_ U L : 0 0 .0 0 %

B C _ U L : 2 9 0 5 .4 3 %

I n t_ D L : 0 %

B C _ D L : 6 2 6 1 1 .7 3 % < R x L e v _ S e r v in g > = -1 0 6 .5 6 d B m


1 0 2 -1 0 1 .5 4 6 3

Thresholds

❏ Bad Coverage

– RxLev ≤ -95

– RxQual > 4

❏ Interference

– RxLev > -95

– RxQual > 4

1.2 Coverage problem Investigation with Abis trace (2/2)

▼ All samples are Bad Coverage samples (BC). None is interference, showing that this cell is not facing any interference problem.

▼ By the way, if the cell is mono-TRX, this is a coverage problem.▼ If the cell is 3 TRXs, this is a malfunction of the TRE (shown also by the high value of delta_path_loss).


1.13

▼Suspecting a cell coverage problem

� Distribution of samples per RxQual value and RxLev band

� Distribution of samples per RxLev band

Not acceptablecoverage limit:Too low level Too bad quality

1.2 Coverage problem Investigation with RMS (1/2)

▼ A coverage problem is observed when a significant amount of the traffic of a cell is suffering from both low level and bad quality (RxQual).

▼ To confirm, distribution of samples per RXLEV band should be also considered to know the proportion of calls which are experiencing a low signal level.

▼ If a lot of samples of low level and bad quality are observed for only a sub-part of the TRXs (can be one only) then a BTS hardware problem or a problem on the antennae should be suspected.

▼ If all the TRXs are experiencing a lot of samples of low level and bad quality then a coverage problem must be suspected.

▼ These RMS indicators are provided on RNO tool per TRX, per Cell:� Matrix of Number of Measurement Results per DL RxQual value and per DL RxLev band

RMQLDSAM = RMS_DL_RxQuality_RxLevel_sample � Vector of Percentage of Samples per DL RxLev band

RMQLDLVDV = RMS_DL_RxLevel_distrib� Vector of Percentage of Samples per DL RxQual band

RMQLDQUDV = RMS_DL_RxQuality_distrib


1.14

▼Suspecting a cell coverage problem

� Average TA values per RxQual value and RxLev band

Not acceptablecoverage limit:Too low level Too bad quality

Acceptable coverage limit:Sufficient level and good quality

% of TA value over TA threshold has also to be considered

1.2 Coverage problem Investigation with RMS (2/2)

▼ In order to know if the coverage problem is due to a big amount of traffic at the cell border or rather to indoor calls, the average TA value per RXQUAL value and RXLEV band as well as the Percentage of TA values over TA threshold should be observed.

� Matrix of Average TA per UL RxQual value and per UL RxLev bandRMQLUTAM = RMS_UL_RxQuality_RxLevel_TimingAdvance

� Rate of Measurements Results whose TA is greater than the TA thresholdRMTAGTR = RMS_TimingAdvance_greater_threshold_rate

� Maximum TA value of all values reported in Measurement Results RMTAMXN = RMS_TimingAdvance_max




1.3 Interference problem


Coverage problem





Exercises


1.16

▼ Definition: Interference

� A network facing interference problems presents good RxLev and bad RxQual in the same time on some areas.

▼ Symptoms

� Customers complain about bad speech quality (noisy calls) and/orcall drops


� SDCCH/TCH Drop� Low proportion of better cell HO� High rate of DL/UL quality HO and interference HO� Low HO success rate



1.3 Interference problem Definition and symptoms

▼ DL/UL depends on the way on which the interference is present.

▼ Mainly, interferences are in the DL, due to bad frequency planning introducing interferences in the network. And this problem will not change till the frequency plan is not returned…

▼ Sometimes, interference can be in the UL in very dense area (for example, microcell area), since MSs are very close.

▼ Finally, sometimes interferences are not coming from BS or MS but from another radio equipment, either in the UL or the DL.


1.17

▼Radio Measurement Statistics (RMS) –

�RxQual/RxLev matrix �CFE/RxLev matrix�C/I vectors for neighbors�C/I vectors for MAFA frequencies

� MAFA is a new standardized GSM feature for mobiles� MAFA mobiles can provide C/I measurements from

non-neighbor cells

�Number of calls with DL/UL interference (good RxLev, bad RxQual)

�Number of noisy calls (bad RxQual) with bad voice quality (bad FER)

1.3 Interference problem Examination with RMS (1/3)

▼ The feature Radio Measurement Statistics (RMS) is designed to make far easier the work for planning and optimization of the network by providing the operator with useful statistics on reported radio measurements.

▼ In fact these statistics give directly the real cell characteristics by taking into account the MS distribution.▼ Thanks to this feature, the operator is able to:

� detect interfered frequencies.� assess the quality of the cell coverage.� detect and quantify cell unexpected propagation.� assess the traffic distribution in the cell from statistics on reported neighboring cells.� evaluate the voice quality in the cell.� etc.

▼ In regards to the “RTCH Measurements Observation” (measurement type 11), the Radio Measurement Statistics (RMS) bring the following advantages:

� smaller report files.� the report files always have the same maximum length whatever the measurement duration is.� every measurement is taken into account (no sampling).� no more need for measurement post-processing tools for statistics. Directly available with RNO or NPA.


1.18

▼Suspecting a cell interference problem

� Number of samples per RxQual value and RxLev band

Average RxQual value per RxLevband has also to be considered

Average DL RxQuality = 2.81

Quality problems are obvious at any level of RMS data

➨ Interference highlighted

➨ Network fine tuning needed



1.19

▼Suspecting a Voice Quality problem

� Number of samples per BFI band and RxLev band

RxQual to be compared with CFE

since

Bad RxQual does not always mean bad VQ

It is directly linked to Voice Quality

Consecutive Frame Erasure (BFI) is a measurement based on

loss of consecutive speech frames over one SACCH mf


▼ These RMS indicators are provided on RNO tool per TRX, per Cell:� Matrix of Number of Measurements Results per CFE band (or BFI band) and per UL RxLev band

RMFEM = RMS_UL_ConsecutiveFrameErasure_RxLevel_sample� Vector of Average number of Consecutive Frame Erasure per UL RxLev band

RMFEBFAV = RMS_UL_ConsecutiveFrameErasure_avg_per_RxLevel� Vector of Average UL RxQual per RxLev band

RMQLUQUAV = RMS_UL_RxQuality_avg_per_RxLevel


1.20

▼GSM interference

� co-channel� adjacent

▼Non GSM interference

� other Mobile Networks� other RF sources

1.3 Interference problem Typical causes


1.21

▼Adjacent channel interference

� +6 dB are sufficient to interfere (9 dB according GSM)

Frequency

Level

F(BTS2)

F(BTS1)=F(BTS2)+1

6 dB

F(BTS1)

1.3 Interference problem GSM interference: adjacent channel (1/2)


1.22

▼Adjacent channel interference: usually 20% of GSM interferences

� Symptom� Usually downlink interference� High rate of quality HO, call drop (due to HO but mainly due to

radio) and TCH assignment failure

� Examination

� Neighbor cells in Abis trace (only for BCCH)� Non-neighbor cells in RMS (MAFA frequencies)� Frequency planning C/(I adjacent) < -6 dB

� Correction

� Downtilt increase of interferer, or even change of antenna orientation

� Reduction of BS power if necessary, Change of frequency (best solution)

� Concentric cell implementation (1 extra TRX needed if traffic cannot be supported by Outer+Inner configuration)

1.3 Interference problem GSM interference: adjacent channel (2/2)


1.23

▼GSM Interference

� Co-Channel interference

� -12 dB are sufficient (-9 dB according GSM)

Frequency

Level

F(BTS1)=F(BTS2)

-12 dB

1.3 Interference problem GSM interference: co-channel (1/2)


1.24

▼ Co-channel interference (usually 80% of GSM interferences)

� Symptom

� Usually downlink interference� High rate of quality HO, call drop and call failure

� Examination

� Neighbor cells in Abis trace (only for BCCH)� Non-neighbor cells in RMS (MAFA frequencies)� Frequency planning C/I < 12 dB

� Correction

� Downtilt increase of interferer, or even change of antenna orientation

� Reduction of BS power, Change of frequency� Concentric cell implementation (1 extra TRX needed if traffic

cannot be supported by Outer+Inner configuration)

1.3 Interference problem GSM interference: co-channel (2/2)


1.25

BTS1

BTS2

MS1

MS2

1

2

3

▼ GSM interference: µcellular� BTS1: ARFCN 5

� BTS2: ARFCN 6

� MS1 indoor

� RxLev_UL: - 90 dBm

� MS2 outdoor, connected to BTS2

� 1: no level on BTS1 (BTS 1 under-roof)

� 2: - 80 dBm on BTS1: interferer UL/DL

� 3: no level on BTS1� µcell algo prevents BTS2->BTS1

HO

1.3 Interference problem GSM interference: µcellular

▼ When interferences are created by frequency plannig, it’s not so hard to detect them. But frequency planning tools mainly consider DL C/I and coverage.

▼ Some problems are more difficult to predict. For example, let’s consider a microcell layer:

� A and B are 2 microcells with the coverage described before in dense urban environment. � Even if both cells A & B are using adjacent frequencies (5 and 6), the overlapping area is far from cell A

antenna. Thus, in this area C/I is lower than 6 dB.

� A “red” MS is connected on cell A. When the MS starts its call, it transmits full power and a PC algorithm quickly reduces MS power as the received level is very good (microcell coverage). When MS A enters the building, it faces a loss of signal of 20 dB. Then, MS power increases to MS_TXPWR_MAX.

� A second mobile “B” is connected to cell B and moves down in the coverage area of cell B. MS power of B decreases quickly down to MS_TXPWR_MIN as the MS is close to the antenna. But when MS B arrives outside the building where A is sitting, A and B are close and transmitting on adjacent frequencies… Then B has to increase its power to avoid dropping its call. By the way, global level of freq B is increased in all cell B… creating interference in the UL.

A

B


1.26

▼GSM Interference: Forced Directed Retry� The MS should connect to

cell2, but no TCH available� The MS connects to cell 1

with forced directed retry� The MS is emitting at high

level (far from BTS1)�UL interference for

BTS 3� BTS 1 is emitting at high

level�DL interference at

BTS 3

cell 2 : 45

cell 3 : 23

cell 1 : 24

1.3 Interference problem GSM interference: Forced Directed Retry

▼ Another more difficult case of interference: FDR� When examining the preceding situation of planning tool: no problem of C/I. No risk of interference.� The FDR algorithm allows an MS connected on an SDDCH on a cell without any free TCH to make an SDCCH-

TCH handover (cause 20) so that it takes a TCH on its neighbor. As seen from the user, this is not a handover (call establishment phase, no impact on speech quality), and this algorithm is very efficient to avoid cell congestion cases.

� This algorithm is mainly based on neighbor level compared to parameter L_RXLEV_NCELL_DR (n). If the level greater than this threshold, the TCH is to be seized on neighbor.

� FDR is mandatory for dual layer or dual band networks (and very easy to configure in this case), since we have capture handovers. Capture handovers send traffic to lower or preferred band cells. In case these cells are congested, calls may not be established, even if upper or non-preferred band cells are free (due to MS idle mode selection, advantaging microcell for example). With the FDR algorithm, the MS takes an SDCCH in the preferred cell, and FDR is used to take a TCH on the non-preferred cell in case of congestion. This situation highlights a good network behavior, since the MS is at the same time in the coverage area of both cells (preferred and not preferred).

▼ The situation described on the slide corresponds to the usage of FDR in a single layer network. This is in that case a heavy-to-tune algorithm presenting of lot of interference and bad quality call risks, since the mobile will be connected to a cell when being not in its service area.

umbrella

microcellFDRcapture


1.27

▼Other mobile networks: TACS/AMPS/NMT900

� Inter-modulation with GSM BS/MS receiver� spurious RACH for AMPS (AMPS Tx bands close to

GSM uplink band)� examination

� TASC: coverage hole with 600 m from TASC BTS�AMPS => 50% reduction of range if AMPS/GSM BTS

collocated

▼Other RF interferers (Radar, shop anti-theft mechanisms, medical device ...)

1.3 Interference problem Non-GSM interference

▼ Other RF interferers:� medical devices: GSM equipments disturb them more than the opposite !� anti-theft mechanisms. � Example:

� The Microcell is showing a very high call drop rate. On one frequency, very small call duration.� No problem seen in the frequency plannig. No potential interferer.� Abis trace:

� The Spectrum analyzer connected on the antenna feeder highlights a peak on GSM freq 6 in the UL…� Anti-theft mechanism turned off: no more problem…

shop

Microcellantenna

Qual

Level

Qual

Level

DL UL

interference




1.4 Unbalanced power budget problem


Coverage problem





Exercises


1.29

▼ Definition: Unbalanced power budget

� A cell facing unbalanced power budget problems presents a too high path-loss difference between UL and DL (often DL>UL)

� Rule: try to have delta as small as possible to avoid access network possible only in 1 direction (usually BTS->MS: OK and MS->BTS: NOK)

▼ Symptoms:


� High rate of Uplink quality Handover causes� Low incoming HO success rate (no HO Access triggered on the uplink)� Degradation of TCH failures and OC call drop indicators



� O&M Alarms

� Voltage Standing Wave Ratio BTS Alarm (VSWR)� TMA Alarm (in case of G2 BTS or Evolium BTS with high power TRE)

1.4 Unbalanced power budget problem Definition and symptoms

▼ UL Quality HO is triggered:� UL since the problem is in the UL.� Quality as Quality has greater priority than level.


1.30

▼Examination:

�RMS –� Path Balance vector per TRX� Number of calls with abnormal bad FER (good RxQual & bad FER)

�Abis monitoring:

� |delta path-loss| > 5dB)� check if problem is occurring for 1 TRX or all

1.4 Unbalanced power budget problem Examination

▼ Problem on 1 TRX: FU/CU or TRE problem or ANY problem or cables connected to this equipment.▼ All TRXs: problem on antenna, feeder, jumper or common equipment (ex: ANX, ANC).


1.31

Fre

quen

cy

RxL

ev_U

L

RxL

ev_D

L

RxQ

ual_

UL

RxQ

ual_

DL

Pat

h_lo

ss_U

L

Pat

h_lo

ss_D

L

delta

_Pat

h_lo

ss

delta

_qua

lity

AV

_MS

_PW

R

nb_o

f sam

ples

106 -94 .52 -87 .19 0.43 0 .25 127.55 130 .19 -2 .64 0.18 33.03 206689 -84 .29 -75 .17 0.65 0 .44 115.32 118 .17 -2 .85 0.21 31.03 2001

118 -90 .75 -83 .36 0.46 0 .41 123.22 126 .36 -3 .14 0.04 32.46 3193124 -88 .89 -85 .30 0.29 0 .67 120.48 128 .30 -7 .82 -0 .37 31.59 2931

D IS TR IB U TIO N O F U P LIN K Q U A LITYF requenc y Q ual0 Q ual1 Q ua l2 Q ua l3 Q ual4 Q ual5 Q ual6 Q ua l7 B ad_Q uality

106 84.75 % 4.07 % 3.68 % 3.19 % 1.36 % 1.50 % 0.92 % 0.53 % 2.95 %89 81.41 % 1.70 % 2.95 % 3.65 % 6.35 % 2.55 % 1.30 % 0.10 % 3.95 %

118 83.62 % 4.23 % 4.23 % 3.35 % 1.57 % 1.79 % 0.97 % 0.25 % 3.01 %124 90.79 % 1.06 % 2.18 % 2.35 % 1.77 % 1.30 % 0.48 % 0.07 % 1.84 %

D IS TR IB U TIO N O F D O W N LIN K Q U A LITYF requenc y Q ual0 Q ual1 Q ua l2 Q ua l3 Q ual4 Q ual5 Q ual6 Q ua l7 B ad_Q uality

106 90.27 % 3.44 % 2.08 % 1.55 % 0.92 % 1.36 % 0.34 % 0.05 % 1.74 %89 80.16 % 6.45 % 7.00 % 3.85 % 1.50 % 0.50 % 0.45 % 0.10 % 1.05 %

118 86.78 % 2.72 % 3.95 % 1.82 % 1.41 % 1.13 % 1.19 % 1.00 % 3.32 %124 77.14 % 4.37 % 5.87 % 5.94 % 3.48 % 1.36 % 0.82 % 1.02 % 3.21 %


1.4 Unbalanced power budget problem Abis trace

Example of Computation of delta path loss based on Abis measurements

BTS transmitted power 45,4 MS transmitted power 33

combiner loss -4,4measured received DL level -93 measured received UL level -98DL Path loss 134 UL path loss 131

delta path loss computed on Abis -3 dBm


1.32

▼Suspecting a TRX hardware problem

� Average Path Balance

� Fair average Path Balance at Cell level can hide a bad value for one TRX

Average Cell Path Balance= -0. 9 dB

1.4 Unbalanced power budget problem RMS data

▼ These RMS indicators are provided on RNO tool per TRX, per Cell:� Vector of the Number of Measurement Results per Path Balance band

RMPBV = RMS_PathBalance_sample� Average Path Balance value

RMPBAN = RMS_PathBalance_avg


1.33

▼Antennae or common RF components, TMA (pb common to all

TRXs of the BTS)

▼TRX RF cables/LNA ... if problem located on only 1 FU

1.4 Unbalanced power budget problem Typical causes

▼ Every BTS has its proper architecture and the diagnosis must be adapted.




1.5 TCH Congestion problem


Coverage problem





Exercises


1.35

1.5 TCH Congestion problem Definition and symptoms

▼ Definition: TCH Congestion

� TCH Congestion rate (TCH Assignment Phase) is too high (more than 2%)

� Rule: try to meet the offered traffic (asked by users) by providing the right number of resources (TRX extension)

▼ Symptoms:

� Customers complain about ‘Network busy’


� High “TCH Congestion rate”� Low “incoming Intra/Inter BSC HO success rate” (no TCH available)� High “Directed Retry rate” if activated

� A interface indicator: “BSS Congestion failure in OC”

� High rate of Assignment Failure messages, No radio resource available


1.36

1.5 TCH Congestion problem Examination and typical causes

▼ Examination: TCH Congestion

� On a per cell basis examination, check the evolution of the TCH Congestion rate.

▼ Typical causes:

� Special events:

� Foreseeable: football match, important meeting � Activate some TRXs already installed (and use Synthesized

FH)� Add special moving BTSs

� Not foreseeable: car crash on the highway

▼ Cells on wheel operational by several operators around the world for special events coverage & capacity� IRMA (SFR) connected to Caen’s BSC.� Orange coverage / Football WC 1998 for Paris « Stade de France »:

� Specific cells covering Paris Stadium. During games, only small capacity (using joker frequencies). During breaks, some TRX off cells around are turned off, and frequencies are reused for stadium cells.


1.37

1.5 TCH Congestion problem Typical causes (1/2)

� Daily periodic problems

� At peak hour, the cell is not correctly dimensioned. Hardware solution (refer to Annex 1)

� Estimate the offered traffic: – At OMC-R level: Traffic in Erlang/(1- TCH Congestion

rate)

� Use the B-Erlang law to estimate the number of TCHsrequired for a 2% blocking rate, thus the target configuration

� Add TRXs to reach the new target configuration and find ‘joker frequencies’ and / or implement concentric cells.

Annex 1

▼ Warning: “offered traffic” is not the capacity delivered by the system but the traffic asked by the users.


1.38

1.5 TCH Congestion problem Typical causes (2/2)

▼Daily periodic problems

�At peak hour, the cell is not correctly dimensioned. Software solution

� Use specific densification features – Half Rate– Forced Directed Retry – Traffic handover– Fast Traffic handover – Candidate Cell Evaluation (FREEFACTOR /

LOADFACTOR)

▼ Half rate may not only mean “SW” solution. Need of G2 BSC/TC, Evolium TRE or G2 DRFU.




1.6 Deducing the right team for intervention


Coverage problem





Exercises


1.40

1.6 Deducing the right team for intervention Process

Problem characterization

Make assumption causes

Recurrent problem ?

Check the tuning of default radio parameters

Consult the config. db Choose an (other) classical algo

Identify the tunable parameters

Impact estimation

Standard setting ?

No

Yes

Yes

No

No

Yes

Call expert

- Microcell, multiband- Concentric

=N

No

Yes

No

Yes

No

Yes

Parameters modificationDatabase updating

Impact simulation of aparameter modification

No

- Hopping- Marketing

Yes

QOS alarm on the network,on a BSC or some cells

- Indicators (% call drop)- Field measurements/planning- Subscriber complains

QOS team

DHCPEND

Drive test team

DHCPEND

Dimensionning team

OK

Correctionaction

Maintenance team

Planning team

NOK

Cell corrected ?Neighbor cell ?

RFT team - Interferences- Coverage (indoor)- Power budget- Congestion (TCH, SDCCH)- BSS problem

Check ? With QOS ?

Investig problem ?

Planning/BSS causes

Standard parameters ?

On purpose

System problem ? N times Simulation OK ?


1.41

1.6 Deducing the right team for intervention Coverage problem

▼ Coverage problem:

� If the field reality does not match the RNP prediction

�Maintenance team to change physical configuration (tilt, azimuth, antenna height, etc.) and drive test team to check it

� If the field reality matches the RNP prediction

�Deployment team to add sites (tri-sector, micro cellular, indoor cells)


1.42

1.6 Deducing the right team for intervention Others problems

▼ Interference problem:

� Planning team to identify the interference source and correct it (joker frequency, new frequency planning, etc.)

▼ Unbalanced power budget problem:

� Maintenance team to check the impacted BTS (Antennae, TMA, RF cables, LNA, diversity system, etc.)

▼ TCH Congestion problem:

� Traffic team (theoretically always in relation with the marketing team) to manage the need of TRX extension, densification policy, etc.


1.43

Time allowed:

10 minutes

1. Typical radio problems Training exercise

UnbalancedUnbalancedUnbalancedUnbalancedPower BudgetPower BudgetPower BudgetPower Budget

Bad CoverageBad CoverageBad CoverageBad Coverage InterferencesInterferencesInterferencesInterferences TCHTCHTCHTCHCongestionCongestionCongestionCongestion

High rate of UL QUAL HOHigh rate of UL QUAL HOHigh rate of UL QUAL HOHigh rate of UL QUAL HOcausescausescausescauses

Good RXLEV andGood RXLEV andGood RXLEV andGood RXLEV andBad RXQUALBad RXQUALBad RXQUALBad RXQUAL

VSWR aVSWR aVSWR aVSWR alarm larm larm larm (OMC-R)(OMC-R)(OMC-R)(OMC-R)(Voltage Standing Wave(Voltage Standing Wave(Voltage Standing Wave(Voltage Standing WaveRatio)Ratio)Ratio)Ratio)

Bad RXLEV andBad RXLEV andBad RXLEV andBad RXLEV andBad RXQUALBad RXQUALBad RXQUALBad RXQUAL

High Path-loss differenceHigh Path-loss differenceHigh Path-loss differenceHigh Path-loss differencebetween UL and DLbetween UL and DLbetween UL and DLbetween UL and DL

Low incoming HO successLow incoming HO successLow incoming HO successLow incoming HO successraterateraterate

OMC QOS indicators:OMC QOS indicators:OMC QOS indicators:OMC QOS indicators:% TCH ASS failure high% call drop high% QUAL HO% call drop% call failure





1.45

2 ALGORITHMS AND ASSOCIATED PARAMETERS Session presentation

▼Objective: to be able to describe the Power control and Hand-over algorithms and list the associated parameters

▼Program:


2.2 Radio measurements principles

2.3 Averaging windows and book-keeping

2.4 Radio Link Supervision and Power control

2.5 Handover Detection

2.6 Handover Candidate Cell Evaluation

2.7 Handover Management

2.8 Exercise

S1: TYPICAL RADIO PROBLEMS

S2: ALGORITHMS AND ASSOCIATED PARAMETERS



S5: CASE STUDIES






Radio measurements principles

Radio measurements data processing

Radio Link Supervision and Power control

Handover Detection

Handover Candidate Cell Evaluation

Handover Management

Exercise


1.47

JUSTIFICATIONWhen the detected problem does not concern another team (Network planning and

frequency planning, Dimensioning, Radio engineering, Maintenance) or

when the other teams cannot give any solution (too tight frequency planning, no additional TRX available, no financial budget for new sites, etc.)

the Radio Fine Tuning team has to find a compromise between: � High traffic density (Erl/km²/Hz)� High quality of service (Call drop, CSSR, Speech quality, indoor, etc.)

Its role : take charge of radio resources management process▼ This process can be fully described by Power Control and Handover algorithms.

➨ In-depth knowledge of these algorithms is required for tuning

2.1 Theoretical presentation Justification




2.2 Radio measurements principles





Handover Detection


Handover Management

Exercise


1.49

� MS connected (TCH or SDCCH)

� The serving cell gives the MS the list of the neighbor cells to listen to

� Every SACCH, the MS reports to the serving cell: measurement report message

� Received level of 6 best cells (which can change)

� DL level and quality of serving cell

2.2 Radio measurements principlesRadio measurement mechanisms (1/2)

▼ The BTS sends a SYS_INFO_5 message that contains the list of neighbor cells for connected mode. (The SYS_INFO_2 message contains the list of neighbor cells for idle mode).

� Sys info 2bis, 2ter, 5bis and 5ter are also used for multiband networks.� MS reporting depends on EN_INTERBAND_NEIGH and on MULTIBAND_REPORTING parameters.

The MS may report:� 6 strongest cells of any band (MULTIBAND_REPORTING=0), or� 5 strongest cells of the serving band + 1 strongest cell of another band

(MULTIBAND_REPORTING=1), or� 4+2 (MULTIBAND_REPORTING=2), or� 3+3 (MULTIBAND_REPORTING=3).

▼ RXLEV� Range: [-110dBm, -47dBm]� Binary range: [0, 63]; 0=-110dBm, 63=-47dBm� The higher the physical or binary value, the higher the receiving level

▼ RXQUAL� Range: [0.14%, 18.10%]� Binary range: [0, 7]; 0=0.14%, 7=18.10%� The lower the physical or binary value, the lower the bit error rate, the better the quality� 0-2=excellent; 3=good; 4=ok; 5=bad; 6=very bad; 7=not acceptable


1.50

▼ For each MS connected to the BTS (TCH or SDCCH)

� UL received level and quality is measured every SACCH

� The Timing Advance (TA) is computed

� The UL information is gathered into the measurement report

� This is the message result sent by the BTS to the BSC

• The BSC is computing algorithms

• usually using average value (sliding window) of these measurements

Measurements Active channel preprocessing

HO & PC decision

PC execution

Candidate cell evaluation HO execution

BSC

DL measurements UL+DL measurements

measurement report measurement result

2.2 Radio measurements principlesRadio measurement mechanisms (2/2)

▼ The BTS starts sending MEASUREMENT RESULT messages as soon as it receives the RL ESTABLISH INDICATION message from the MS.

▼ The BTS stops sending MEASUREMENT RESULT messages upon receipt of one of the two following messages: � DEACTIVATE SACCH � RF CHANNEL RELEASE

▼ Every SACCH multiframe, the BTS:� receives the MEASUREMENT REPORT message from the MS. For power control and handover algorithms,

this message contains downlink measurements and, in the layer 1 header, the power used by the MS.� does uplink measurements. � reports the uplink and downlink measurements to the BSC in the MEASUREMENT RESULT message.� Input flows

� Uplink radio signal: radio signal received on the Air interface.� BS_TXPWR_CONF: BS transmit power currently used by the BS.� DTX_DL: indicator of downlink DTX use.

� Output flows� Abis MEASUREMENT RESULT message

� Internal flows� Radio measurements:

� Air MEASUREMENT REPORT message (DL) containing DL MS radio measurements.� Uplink radio measurements (quality and level) and a flag indicating whether DTX was used in the

downlink (DTX/DL).� Timing advance: last TA calculated by the BTS.� MS_TXPWR_CONF: last reported value of MS power (reported by the MS).� BS_TXPWR_CONF: value of the BS transmit power currently in use.� BFI_SACCH: bad frame indicator of the SACCH block produced every SACCH multiframe (# 480ms):

� 0 = SACCH frame successfully decoded� 1 = SACCH frame not successfully decoded


1.51

L1 Info

L3 Info

MeasurementReportFrom the MS

2.2 Radio measurements principlesStructure of a measurement result

▼ Basically, the MEASUREMENT RESULT message is composed of:� L1 info: SACCH Layer 1 header containing MS_TXPWR_CONF and TOA.� L3 info: MEASUREMENT REPORT from the MS. This message contains the downlink measurements and

neighbor cell measurements.� Uplink measurements performed by the BTS.� BTS power level used.

▼ SUB frames correspond to the use of DTX� if the mobile is in DTX, the rxlevsub or rxqualsub is used to avoid measuring the TS where there is nothing to

transmit in order not to distort measurements.� else rxlevfull is used that is to say all TSs are measured.

▼ MS TXPOWER CONF: which is the actual power emitted by the MS.

▼ TOA is timing advance.

▼ SACCH BFI: bad frame indicator; 2 values 0 or 1; 0 means that the BTS succeeded in decoding the measurement report.

▼ How the neighbor cells are coded:� BCCH1 index in BA list / BSIC1; BCCH2 index in BA list / BSIC2 � why? because it does not receive LAC/CI (too long ) but BCCH and replies with BCCH/BSIC


1.52

MS BTS BSC MSCTCH ASSIGNMENT PHASE (OC or TC)

< -----------------------------------ASSIGNMENT REQUEST

< --------------------------------------------------------PHYSICAL CONTEXT REQUEST

-------------------------------------------------------- >PHYSICAL CONTEXT CONFIRM

< --------------------------------------------------------CHANNEL ACTIVATION (TCH)

(EMO included)-------------------------------------------------------- >CHANNEL ACTIVATION ACKNOWLEDGE

.

.TCH establishment.

--------TCH---------> .ASSIGNT COMPLETE ----------------------------------- -------------------- >

ASSIGNMENT COMPLETE -------------------------------- --- ><------SACCH-------- ASSIGNMENT COMPLETE

--------SACCH------><------SACCH--------

--------SACCH------><-------SACCH--------

EMO(MAFA freq. List)

.

.

.--------SACCH------>

EMR(MAFA freq. RxLev)

▼Extended Measurement Reporting mechanisms

� Extended Measurement Order includes the MAFA frequencies the MS is asked to measure

� EMO sent once to the MS on SACCH after TCH seizure

� Extended Measurement Results include the average signal level measured on each MAFA frequency over one SACCH mf duration

� EMR received once per call on SACCH

2.2 Radio measurements principlesExtended Measurement Reporting (EMR)

▼ When the BTS receives a CHANNEL ACTIVATION with the Extended Measurement Order (EMO) included, it must send this information on the SACCH to the corresponding mobile only once.

▼ When the BTS has to send this information, it must replace the sending of system information 5, 5bis, 5ter or 6 by this information. At the next SACCH multiframe, the BTS must resume the sending of this system information by the replaced one.

▼ The EMO must be sent after 2 complete sets of SYS_INFO5 and 6, i.e. after the 2nd SYSINFO 6 after the reception of SABM. This guarantees the MS has received a complete set.

▼ Then, the BTS normally receives from the MS an EXTENDED MEASUREMENT RESULT with the level of the frequencies to monitor. The BTS must make the correlation between these levels and the frequencies contained in the latest EMO information, after having decoded them, according to the order of the ARFCN. The ‘EXTENDED_MEASUREMENT_RESULT’ is NOT forwarded to the BSC, instead a ‘MEASUREMENT_RESULT’ with indication ‘no_MS_results’ is sent to the BSC.

▼ In particular, the BTS must identify the level of the BCCH frequency of the serving cell (which must always be part of the frequencies to monitor) and apply it as the RXLEV_DL in the Radio Measurement Statistics. The other frequencies will be considered in the same way as the BCCH frequency of neighbor cells: they will be linked to the neighbor level and C/I statistics.


1.53

Time allowed:

5 minutes

� (BSIC, BCCH index)/(LAC, CI) problem

� As LAC and CI information take up too much space, the MS only reports the decoded BSIC and the BCCH index when it sends measurement on the adjacent cell

� The BSC makes the correspondence between the couple (BSIC, BCCH index) and the real neighbor cell concerned [completely defined by (LAC,CI)]

� WHAT IS THE RISK?

2.2 Radio measurements principlesTraining exercise (1/2)


1.54

2.2 Radio measurements principlesTraining exercise (2/2)

▼ Explain why cell 2 has a very high outgoing HO unsuccessful rate and a high call drop

CI=6169GSM900

CI=1964

GSM900

Cell 3(7, 62)

Cell 2(3,46) Cell 1

(7, 62)

CI=6169GSM900




2.3 Radio measurements data processing





Handover Detection


Handover Management

Exercise


1.56

2.3 Radio measurements data processingFunctional entities

RadioLink Measurements

ActiveChannelPre-processing

Assignment of radio measurements data processing functions in thAssignment of radio measurements data processing functions in thAssignment of radio measurements data processing functions in thAssignment of radio measurements data processing functions in the ALCATEL BSSe ALCATEL BSSe ALCATEL BSSe ALCATEL BSS

BTS BSC

▼ The active channel pre-processing function calculates average values of signal levels, qualities and timing advance provided by the radio link measurements function.

▼ The pre-processing is based on a sliding window averaging technique. The averaging is either weighted or unweighteddepending on the type of the input parameters.


1.57

▼ Active channel pre-processing

� ACTIVATED EACH TIME A MEASUREMENT IS RECEIVED

� AVERAGING VALUES OF SIGNAL LEVELS, QUALITIES, TIMING ADVANCE

� USING “SLIDING WINDOW” TECHNIQUE

� BUILDING A BOOK-KEEPING LIST OF NEIGHBOR CELLS

� The MS is reporting the 6 best cells at one time� They can change from 1 measurement to another� Maximum for 1 call: last 32 best ones (among 64 maximum

declared as neighbor)

2.3 Radio measurements data processingActive channel pre-processing


1.58

▼ Active channel pre-processing – Principles

� HANDLED by the BSC

� ACTIVATED when the BSC receives:

� ESTABLISH INDICATION from the MS on SAPI 0, or� HANDOVER FAILURE from the MS, or� ASSIGNMENT FAILURE from the MS (in case of intracell

handover)

� STOPPED when a HANDOVER COMMAND is emitted in the serving BSC

� AVERAGING VALUES OF SIGNAL LEVELS, QUALITIES, TIMING ADVANCE

� USING “SLIDING WINDOW” TECHNIQUE

� BUILDING A BOOK-KEEPING LIST OF NEIGHBOR CELLS

2.3 Radio measurements data processingActive channel pre-processing - Principles

▼ The pre-processing function is stopped when a HANDOVER COMMAND is emitted by the serving BSC. At this time, the MEASUREMENT RESULT messages are ignored by the pre-processing function and no update of the book-keeping tables or averaging is done anymore.

▼ The pre-processing function is enabled again (in case of failure of an intracell or intercell handover) after reception of either messages listed above, and the old measurements are kept in the book-keeping list and taken into account in the new averaging.

▼ The pre-processing function is completely handled by the BSC. The input parameters of this function are provided by the BTS every SACCH multiframe in the MEASUREMENT RESULT message.

▼ The function calculates average values of levels, qualities and timing advance. The pre-processing method is based on a sliding window averaging technique. The pre-processing is done for every measurement sample, i.e. every SACCH multiframe. The averaging intervals are expressed in terms of SACCH multiframe periods and their range is between 1 and 31.

▼ The averaging process for any variable can start as soon as A_YYYY_XX (YYYY stands for “LEV”, “QUAL”, “PBGT” or “RANGE” and XX for “HO”, “DR”, “PC” or “MCHO”) samples, each with MEAS_VALID bit set to 0 (validity indicator reported by the MS in the MEASUREMENT REPORT message), are actually available except in case of the averagingof the received level from the neighbor cells and the averaging of AV_RXLEV_PBGT_HO, AV_BS_TXPWR_HO andAV_BS_TXPWR_DR.


1.59

▼Avoid reacting too early to some “atypical” measurement(s)

1 2 34 5

67 8

9

average (w indow 8)

average (w indow 4)

level

75,00

80,00

85,00

90,00

95,00

100,00

105,00

sliding window effect

2.3 Radio measurements data processingMeasurement averaging (1/2)

▼ The calculation of levels, qualities and timing advance (i.e. distance information) uses a variety of averaging window sizes as well as specific weighting factors for quality estimates.

▼ One separate window exists for:� power control on the uplink and the downlink (A_LEV_PC , A_QUAL_PC),� emergency handover (A_LEV_HO , A_QUAL_HO , A_RANGE_HO),� fast emergency handover for microcells (A_LEV_MCHO),� better cell handover and better zone handover (A_PBGT_HO) for intra-layer, interlayer and interzone

handovers,� forced directed retry (A_PBGT_DR),� neighbor filtering and ranking for all HOs (A_PBGT_HO),� codec adaptation (A_QUAL_CA_HR_FR , A_QUAL_CA_FR_HR).


1.60

▼ Objective: average measurements to avoid reacting to transient degradation

� Principle: sliding window: level/quality/distance values are averaged for N last samples

N = A_LEV_HO samples for uplink and downlink levelN = A_QUAL_HO samples for uplink and downlink qualityN = A_RANGE_HO samples for distanceN = A_PGBT_HO for level used in power budget equation

� Example (A_LEV_HO=6, A_QUAL_HO=4, A_PBGT_HO=8)

� Experiencessome experiments have shown that the number of HOs is very sensitive to modification of these values

2.3 Radio measurements data processingMeasurement averaging (2/2)

Meas 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

DL Level -90 -92 -93 -98 -100 -99 -98 -90 -80 -75 -72 -71 -110 -70 -69 -68 -78 -88 -95 -98 -100 -110 -110 -110AV-RXLEV -95 -97 -96 -94 -90 -86 -81 -83 -80 -78 -77 -78 -81 -78 -83 -88 -95 -100 -104AV-LEV-PGBT -95 -94 -92 -89 -86 -87 -83 -80 -77 -77 -78 -81 -85 -83 -88 -93 -99DL Qual 2 3 3 4 7 7 7 5 2 1 1 0 6 0 0 0 0 1 2 3 6 7 7 7

AV-RXQUAL 3 4 5 6 7 5 4 2 1 2 2 2 2 0 0 1 2 3 5 6 7

▼ At BSC level, � Input flows

� MEASUREMENT RESULT � Control flows

� active channel pre-processing configuration paramet ers for PC:� A_LEV_PC, W_LEV_PC, A_QUAL_PC and W_QUAL_PC,

� active channel pre-processing configuration paramet ers for HO:� A_LEV_HO, W_LEV_HO, A_PBGT_HO, W_PBGT_HO, A_QUAL_HO, W_QUAL_HO,

A_RANGE_HO, A_LEV_MCHO, W_LEV_MCHO, A_PBGT_DR.� cells list for book-keeping:

� BA_IND_SACCH: indicator of the change of the BA_allocation,� NBR_ADJ: number of declared adjacent cells of the serving cell denoted by n,� for n=1 to NBR_ADJ: BSIC(n) and FREQ(n).

� Output flows� Averaged measurements for power control:

� AV_RXQUAL_UL_PC ; AV_RXLEV_UL_PC: MS power control/threshold comparison,� AV_RXQUAL_DL_PC ; AV_RXLEV_DL_PC: BS power control/threshold comparison.

� Averaged measurements for handover detection:� AV_RXQUAL_UL_HO, AV_RXQUAL_DL_HO, AV_RXLEV_UL_MCHO,� AV_RXLEV_UL_HO, AV_RXLEV_DL_HO, AV_RXLEV_DL_MCHO,� AV_LOAD , averaged traffic load� AV_BS_TXPWR_HO, AV_RANGE_HO,� AV_RXLEV_PBGT_HO, AV_RXLEV_NCELL(n), AV_RXLEV_NCELL_BIS(n).� AV_RXLEV_PBGT_DR,� AV_RXLEV_NCELL_DR(n), n=1..BTSnum.� BFI_SACCH� AV_RXQUAL_xx_CA_HR_FR, AV_RXQUAL_xx_CA_FR_HR

� MS_TXPOWER_CONF / BS_POWER: last power level reported by the MS and transmit power currently used by the BS.


1.61

▼BUILDING A BOOK-KEEPING LIST OF NEIGHBOR CELLS

� The MS reports the measurements of the NO_NCELL_M (≤ 6) best cells every multi-frame

� The adjacent cells reported by the MS can change from one measurement to another

� The book-keeping function keeps a table of the last 32 reported adjacent cells

�Clearing process of non-reported neighbors during 10s (signal level=0)

2.3 Radio measurements data processingneighbor cell measurement book-keeping

▼ An MS is required to measure the BCCH power level of a number of BCCH frequencies. These measurements are used for the power budget computation in the BSC and the candidate cell evaluation in the BSC.

▼ The MS reports to the BTS, in the MEASUREMENT REPORT message, the measurements of the NO_NCELL_M (NO_NCELL_M <= 6) best cells it receives (RXLEV_NCELL, BCCH frequency index and BSIC number) for each multiframe. In case of multiband capability, the mobile reports the best cells of each supported frequency band (if available). This reporting is allowed at BSS level by the flag EN_INTERBAND_NEIGH and it is specified by the parameter MULTIBAND_REPORTING.

▼ The adjacent cells reported by an MS can change over the averaging interval. The book-keeping function keeps a table composed of the last 32 reported adjacent cells, the maximum number of which is NBR_ADJ. The total number of adjacent cells for which measurements reported by the MSs are available within the average interval is BTSnum.

▼ The BSC G1 maintains a table of up to 150 cells, from which up to 64 can be declared as adjacent cells to a given cell.▼ The BSC G2 maintains a list of up to 1000 cells, from which up to 64 can be declared as adjacent cells to a given cell.▼ Because the maximum number of adjacent cells may be greater than 32, the number of adjacent BCCH frequencies is

limited to 32. Moreover, a mechanism for overwriting obsolete entries in the bookkeeping table, when new cells are reported, is provided.

▼ When the variable BTSnum reaches its maximum value of 32 and at least one new cell has to be entered in the list, then the BSC sorts out all cells in the bookkeeping list, which have been reported with signal level = 0 for the last 20 measurements (10 seconds).

▼ This is done by summing the raw measurement values over the last 20 samples. All the corresponding cell entries are cleared from the bookkeeping list, BTSnum is decreased by the number of cleared entries and some of the vacant entries are used to include the new cells.

The end of the comment is on the next page...


1.62

� Measurements averaging

�With ‘averaging window’ excel sheet...

• Compute averaging on quality, distance and level

• Make charts with different sliding averaging windows

2.3 Radio measurements data processingTraining exercise

Time allowed:

10 minutes

▼ Fill up the table with average function. The chart will be automatically processed

▼ The fact that there may not be enough cleared entries to store new measurements is excluded, see justification below:▼ Because the MS must resynchronize at most every 10s with the neighbor cells it monitors, it is useless to keep cells in

the bookkeeping list which have not been reported for more than 10s, it will be impossible to makkes an handover towards these cells.

▼ Therefore, the overwriting mechanism described above will function correctly if there are less than 32 cells reported in every 10s, which makes an average rate of 3 new cells per second.

▼ The potentiality of overflow of the book-keeping list is therefore excluded.▼ The book-keeping is performed according to the BSIC and BCCH frequency couple. This function updates the table

every multiframe except if the measurement report is missing or Measurement Valid Bit is set to not valid. When the level of a cell is not reported, a zero must be entered as measurement value. For each multiframe and for each of the NO_NCELL_M cell measurements it receives, the function has to check the BSIC number and the BCCH frequency index (FREQ(n)).

▼ When the couple (BSIC, BCCH frequency) is not in the reference list (received from the OMC), the corresponding measurements should be discarded.

▼ The BTSnum variable is updated every multiframe except if the measurement report from the MS is missing. It is incremented by the number of new couples (BSIC number, BCCH frequency index) registered as described above.

▼ Remark: Two cells can have the same BSIC number or the same BCCH frequency index. Therefore, the couple of these parameters is needed to define a cell.




2.4 Radio Link Supervision and Power Control





Handover Detection


Handover Management

Exercise


1.64



Assignment of PC functions in the ALCATEL BSS

BTS BSC

RadioLinkSupervision

RadioLinkCommand

PC Threshold Comparison

PC Command

2.4 Radio link supervision and power controlFunctional entities

▼ The two main functions specified in this document and implemented in the ALCATEL BSS are:� Radio link supervision and radio link command:

� These functions handle the detection of the radio link failure so that calls which fail either from loss of radio coverage or unacceptable interference are satisfactorily handled by the network. The radio link supervision is responsible for detection of the loss of the radio link, based on incorrectly received SACCH frames. The radio link command is responsible for commanding to set the power at a maximum level for radio link recovery or to clear the call when the radio link has failed.

� The radio link recovery can be activated or not, depending on a configuration flag (EN_RL_RECOV). The radio link failure procedure is always running and clears the call when the radio link has failed.

� Power control:� This function handles the adaptive control of the RF transmit power from the MS and the BS. The RF

power control aims at minimizing the co-channel interference and also at reducing the DC power consumption of the MS. This function is in charge of detecting a need for a power command and then of applying this power command. Therefore it can be divided into two processes: PC threshold comparison and PC command. MS and BS power control are operating independently, they can be activated or not, depending on configuration flags (EN_MS_PC and EN_BS_PC).

▼ All these functions require directly or indirectly input parameters provided by the function in charge of the radio link measurements.

▼ Most of the input data required by the power control functions are provided by Active channel pre-processing function.

▼ The figure depicts in a general way:� the interconnections between all these functions,� the implementation of these functions in the ALCATEL BSS.


1.65

▼ Principles

� Detection (by BTS) of a radio link failure with an MS

� notification to BSC for radio resource release

� Try to recover an MS when radio becomes poor

� optional mechanism “radio link recovery”

� by requiring BTS and MS to transmit at maximum power

� Equivalent mechanism in MS for Radio Link Failure detection

2.4 Radio link supervision and power controlRadio link supervision

▼ The determination of the radio link failure is based on a counter. According to the GSM Technical Specification 05.08 for the BSS, the criterion for incrementing/decrementing this counter should be based:

� either on the error rate on the uplink SACCH,� or on RXLEV/RXQUAL measurements of the MS.

▼ In the ALCATEL BSS, it is based on the number of SACCH frames which cannot be decoded.▼ It must be stressed that this criterion is related to the first one recommended above but it is not exactly the same. The

ALCATEL criterion is in fact the one recommended by the GSM Technical Specification 05.08 for the MS.


1.66

� For each active radio channel, a counter “S” is

� decremented by 1 each time an SACCH frame cannot be decoded (BFI=1)

� incremented by 2 each time a valid SACCH frame is received

� The value of S gives a measure of the “quality” of uplink radio link

� Initial value of S = BS_RADIO_LINK_TIMEOUT

� if S reaches N_BSTXPWR_M, a radio link recovery is triggered (optional)

� if S reaches 0, a radio link failure is detected

� RADIOLINK_TIMEOUT_BS ≥RADIOLINK_TIMEOUT is important because the mobile must release the radio channel first.

2.4 Radio link supervision and power controlPrinciples of radio uplink supervision

RLTO_BS

18

N_BS_TXPWR_M

13

0

RLTO

(T100)

16

0

SACCH block lost: - 1

SACCH block received: + 2

Radio link Failure

Radio link Recovery

S S

▼ The radio link supervision function is performed in the BTS and it uses three parameters given to the BTS in the TRX configuration data message:

� EN_RL_RECOV: flag enabling/disabling the sending of CONNECTION FAILURE INDICATION by the BTS when the need for radio link recovery is detected,

� N_BSTXPWR_M: threshold for the radio link recovery,� RADIOLINK_TIMEOUT_BS: threshold (number of SACCH messages) for the radio link failure.

▼ In addition, the function handles a counter named S. RADIOLINK_TIMEOUT_BS is the initial and maximum value of S.� For each SACCH not decoded, S is decremented by 1 while for each SACCH decoded, it is incremented by 2.

The incrementation or decrementation is performed if the following condition is met: RADIOLINK_TIMEOUT_BS >= counter S >= 0.

� As soon as the counter S is equal to the threshold N_BSTXPWR_M, the radio link recovery is triggered if EN_RL_RECOV = ENABLE. Therefore, in the case where the shadowing is so strong that all SACCH frames are lost, the radio link recovery will be triggered after (RADIOLINK_TIMEOUT_BS - N_BSTXPWR_M) SACCH periods.

▼ The parameter N_BSTXPWR_M must be set according this simple behavior.▼ If the radio link recovery is not successful, as soon as S reaches 0, the radio link failure procedure is applied.▼ As soon as a radio link failure is detected, the radio link supervision must be started again in the BTS.


1.67

S = f [ BFI(t) ]

0

5

10

15

25

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29SACCH number

Svalue

BFI

S

RADIO_LINK_TIMEOUT_BS

N_BSTXPWR_M

2.4 Radio link supervision and power controlS counter for radio link supervision

▼ Received events� Activate supervision: activation of the radio link supervision from the BTS telecom layer 3,� SACCH, BFI = 1: not decoded SACCH frame,� SACCH, BFI = 0: decoded SACCH frame,

� Note: the BFI flag is internal to the BTS and does not deal with the BFI flag defined by the GSM.� Deactivate supervision: deactivation of the radio link supervision by the BTS telecom layer 3.

▼ Transmitted events� Radio link recovery: indication sent to the radio link command function in order to set the BS and MS powers to

the maximum. � Radio link failure: indication sent to the radio link command function in order to release the call.

▼ These events are sent to the BSC in the CONNECTION FAILURE INDICATION message:� In case of Radio link recovery, the BTS sends only once (to avoid overload of the Abis interface) the

CONNECTION FAILURE INDICATION message to the BSC with cause "set MS/BS-TXPWR-M” (value: '001 1111', reserved for National use). This action (message formatting) is performed by the GSM layer 3.

� In case of Radio link failure, the BTS sends the CONNECTION FAILURE INDICATION message with cause 'Radio link Failure' to the BSC.

▼ Thus, the CONNECTION FAILURE INDICATION message on Abis is not showing any call drop. One should look at the cause of CONFAIL.


1.68

▼The BTS is sending a Connection Failure Indication message

� cause ‘001 1111’ reserved for national usage (ALCATEL: RLR)

�On K1205: “set MS/BS_TXPWR_MAX (Alcatel only)”

▼The BSC is sending BS and MS POWER CONTROL messages

� required for maximum possible values� The MS required level is embedded in the SACCH

header in the downlink

▼Optional mechanism

�EN_RL_RECOV =ENABLE� useless without power control� “master” vs. power control

2.4 Radio link supervision and power controlRadio link recovery

▼ The action consists in increasing the power of the MS and of the BTS to their maximum, in a single step, if the link is failing, i.e. the BTS is not able to decode the SACCH anymore for some period of time.

▼ This functionality is performed upon reception of the CONNECTION FAILURE INDICATION message (cause “set MS/BS-TXPWR-M”) from the BTS. This message can be sent by the BTS only if EN_RL_RECOV = ENABLE. Upon reception of this message, the radio link command function:

1. sends to the BTS a power increase command up to BS_TXPWR_MAX (BS_TXPWR_MAX_INNER if the MS is on the inner zone of a concentric or multiband cell) in the BS POWER CONTROL message.

2. sends to the MS a power increase command up to min(MS_TXPWR_MAX,P) (min (MS_TXPWR_MAX_INNER,P) if the MS is in the inner zone of a concentric or multiband cell) in the message MS POWER CONTROL.

▼ When a radio link recovery occurs, the radio link command function gives an indication to the power control function once the power increase has been commanded.

▼ The maximum power increase of the MS is 2dB per 60 ms. Thus, if MS_TXPWR_MAX=33dBm and MS_TXPWR_MIN=13dBm, the MS coming from MIN to Max will take 600 ms.

Note: the BS Power Control process does not interfere with the recovery procedure since the former comes to a halt when no SACCH multiframe is received. Thus, the BS power control process does not take into account the radio link recovery event.


1.69

▼Radio link failure

� The BTS is sending a Connection Failure Indication message

�Cause ‘radio link failure’

� The BSC is notifying the loss to the MSC

�Usually Clear Request “radio interface failure”

� The BSC is releasing locally the radio resource (TCH or SDCCH)

�Radio frequency Channel Release message sent to BTS

� The call is dropped !

2.4 Radio link supervision and power controlRadio link failure

▼ The task of the radio link command consists in informing the call control function to release the call.▼ Concentric cell or multiband cell▼ The power value BS_TXPWR_MAX_INNER is applied in case of radio link recovery for an MS in the inner zone. The

power value BS_TXPWR_MAX is applied in case of radio link recovery for an MS on an outer zone channel.

▼ Note: the radio link supervision procedure will function also if SACCH frames are not lost continuously, but with a longer reaction time.


1.70

� With the “RLS” excel sheet...

� Taking into account the measurements with BFI and the parameter values (N_BSTXPWR_M and RADIOLINK_TIMEOUT_BS)

� Indicate when � A radio link recovery

is triggered

� A radio link failure is triggered

2.4 Radio link supervision and power controlRadio link supervision: training exercise

Time allowed:

5 minutes


1.71

▼Aims of Power control� Reduce emitted power to the minimum

possible� Minimum power levels:

� GSM: 11dBm, 9dBm, 7dBm and 5dBm

� DCS: 2dBm, 0dBm� Ensuring quality and received level of

peer entity� Adapted in real-time� For Uplink PC: decrease UL interference

and save MS battery

� For Downlink PC, decrease DL interference

BS_TXPWR

RXLEV_DL

downlinkuplink

MS_TXPWR

RXLEV_UL

2.4 Radio link supervision and power controlPower control

GSM-900 DCS-1800

Power level

Output Power(dBm)

Output Power(dBm)

14 15 2

15 13 0

16 11 -

17 9 -

18 7 -

19 5 -

▼ The main objective of the power control, in connection with handover algorithms, is to allow a maximum number of MSsto operate in the network while maintaining a minimum interference level.

▼ The algorithms must ensure that any mobile is connected with the cell in which the output powers from the MS and the BS are as low as possible (to reduce MS power consumption and interference in the network) while keeping a satisfactory link quality.

▼ When on a sufficient duration, the propagation conditions keep worsening, then action must be taken.▼ The first action is to increase the output power levels at the MS or the BS. When the maximum allowed value has been

reached, a handover may become necessary.▼ To reflect this philosophy in macrocells (not in microcellular environment), the algorithm allows for handover on quality

and strength reasons only when the last step of power control has been reached. If propagation conditions worsen rapidly when the MS is at low power, the power control algorithm allows to reach quickly the maximum power.

▼ Nevertheless great care must be taken in choosing the relative values of the thresholds for power control and handover as well as the averaging window sizes (smaller window size and higher threshold for power control than for handover). It must be remembered that, although it is desired that the MS transmits with the lowest possible power, it is more important not to lose a call. Thus early triggering for the power control is possible, by choosing, small values for the averaging window sizes and higher comparison thresholds.


1.72

▼Based on a threshold comparison mechanism

▼Decrease emitted power when received level AND quality measured by peer entity are better than a given value

▼ Increase emitted power when the received level OR quality is lower than a given value

▼Does not decrease power if the resulting level is below the low level threshold

➨ FEATURE REAL FAST PC GIVES REACTIVITY TO THE ALGORITHMS

2.4 Radio link supervision and power controlPower Control principles

▼ The threshold comparison process detects the need to change the MS power level. This detection is done by comparison between the averaged values produced by the active channel pre-processing function and thresholds.


1.73

▼MS Power control (for BS PC, replace MS by BS and UL by DL)

L_RXLEV_UL_P U_RXLEV_UL_P

U_RXQUAL_UL_P

L_RXQUAL_UL_P

POW_RED_STEP_SIZE

-95 -93 -85

1

2

2.4 Radio link supervision and power controlPower Control detection

▼ A need for a PC command is detected when one of the conditions above is true. Then, the information for the execution of the PC command is given to the ‘PC command’ process.

▼ The MS power control function can be disabled with a flag EN_MS_PC. This flag is changeable from the OMC-R.Note: The GSM coding of quality is contra-intuitive, since the value 0 codes for the best quality and 7 for the worst. Thus,

the comparison between two quality values must be understood in the opposite way in terms of quality.Note: POW_RED_STEP_SIZE is used in two ways: for PC_COMMAND (decrease of MS power) and for

PC_THRESHOD_COMPARISON (to avoid ping-pong effect).


1.74

▼Power increase: If� AV_RXQUAL_UL_PC > L_RXQUAL_UL_P + OFFSET_RXQUAL_FH

� AV_RXQUAL_UL_PC ≤ L_RXQUAL_UL_P + OFFSET_RXQUAL_FHand AV_RXLEV_UL_PC < L_RXLEV_UL_P

Then PC_COMMAND(MS, INC, MS_P_INC dB, <min( MS_TXPWR_MAX, P))

▼Power decrease: If� AV_RXQUAL_UL_PC < U_RXQUAL_UL_P

and AV_RXLEV_UL_PC >= L_RXLEV_UL_P + POW_RED_STEP_SIZE

� AV_RXQUAL_UL_PC ≤ L_RXQUAL_UL_P + OFFSET_RXQUAL_FHand AV_RXQUAL_UL_PC ≥ U_RXQUAL_UL_P and AV_RXLEV_UL_PC > U_RXLEV_UL_P

Then PC_COMMAND(MS, RED, MS_P_RED dB, > MS_TXPWR_MIN)

2.4 Radio link supervision and power controlMS PC Threshold comparison

▼ OFFSET_RXQUAL_FH is an internal variable that is equal to 0 in case of Non-Hopping cell and OFFSET_HOPPING_PC in case of BBH or RH.


1.75

▼Power command philosophy:� Target received level TARGET_RXLEV_UL

� middle threshold between U_RXLEV_UL_P and L_RXLEV_UL_P

� Adaptive power step size�According to the average received level� Limited power step size to MAX_POW_INC and

MAX_POW_RED� If only Quality problem: fixed power step size

� POW_INC_STEP_SIZE and POW_RED_STEP_SIZE

� Two weighting factors to modify the algorithm reactivity when level problem� POW_INC_FACTOR for power increase

� POW_RED_FACTOR for power decrease

2.4 Radio link supervision and power controlMS Power Control Command

▼ Whenever any of the threshold conditions occurs, a PC command must be sent to the MS over the air interface.▼ In order to compute the adaptive power step size, the middle threshold between the upper threshold U_RXLEV_UL_P

and the lower threshold L_RXLEV_UL_P is considered.▼ This threshold is regarded as the target received level around which the MS should always stay. The following

algorithm tries to maintain and bring the MS power closer to this target threshold. The size of the power step is limited to MAX_POW_INC for an increase of the MS power and MAX_POW_RED for a decrease of the MS power.

▼ When the received level is between the two thresholds U_RXLEV_UL_P and L_RXLEV_UL_P (i.e. no need to change the level) and a power control on quality cause is triggered, fixed power step sizes are applied: POW_INC_STEP_SIZE for power increase and POW_RED_STEP_SIZE for power decrease.

▼ Two weighting factors POW_INC_FACTOR (for power increase) and POW_RED_FACTOR (for power decrease) allow to modify the reactivity of the algorithm (the more POW_INC_FACTOR is nearby 1, the greater the reactivity of the algorithm is and the larger the power step size is).

▼ The target received level is TARGET_RXLEV_UL for the uplink path.▼ TARGET_RXLEV_UL corresponds to the next higher multiple of 1 dB from (U_RXLEV_UL_P + L_RXLEV_UL_P)/2.


1.76

▼ Example

2.4 Radio link supervision and power controlFast and Normal PC comparison

t/ms0 480 960 1440 1920

0

60

120

180

240

300

360

420

480

t/ms

p/dBm

-110

-100

-90

-80

3. MR

2. MR 4. MR

POW_INC_STEP_SIZE=6dB

Need for PC commanddetected

20dBFast Power Control

Normal Power Control

4 SACCH=1 MeasurementReport (MR)

PC Command


1.77

▼PC_COMMAND (MS, INC, MS_P_INC dB, < power max)� If MS_TXPWR < power maxthen increase MS_TXPWR by min(MS_P_INC, MAX_POW_INC, powermax-

MS_TXPWR)� where MS_P_INC is evaluated by the following algorithm:

if (AV_RXLEV_UL_PC < L_RXLEV_UL_P ) (problem of level)if (AV_RXQUAL_UL_PC ≤ L_RXQUAL_UL_P + OFFSET_RXQUAL_FH)

(sufficient quality)then MS_P_INC = roundup[ POW_INC_FACTOR* (TARGET_RXLEV_UL -

AV_RXLEV_UL_PC)]else MS_P_INC = roundup[ MAX ( POW_INC_FACTOR * (TARGET_RXLEV_UL

- AV_RXLEV_UL_PC ), POW_INC_STEP_SIZE )]else (problem of quality)

MS_P_INC = POW_INC_STEP_SIZE

2.4 Radio link supervision and power controlMS Power Increase Command computation

▼ In the equations: � MS_TXPWR is the last MS_TXPWR_CONF value reported by the BTS.� ‘roundup’ means ‘round to its next higher multiple of 2 dB’.� ‘rounddown’ means ‘round to its next lower multiple of 2 dB’.

▼ The rate of change of MS power is required to be one nominal 2 dB step every 60 msec. Thus a 30 dB step change should be accomplished in 900 msec. The operator should be warned of this as it may impact on the choice of settings for MS_P_CON_ACK and MS_P_CON_INT.

▼ Then the ordered value of the MS transmit power, called MS_TXPWR, is sent to the MS as follows:� The BSC sends the MS POWER CONTROL message to the BTS (i.e. to the TRX handling the relevant

channel) which then forwards the PC command to the MS in the Layer 1 header.� The MS applies the PC command and confirms this action by transmitting the applied power value

(MS_TXPWR_CONF) on the uplink SACCH in the layer 1 header.▼ On SACCH channel, the MS may not send the MEASUREMENT REPORT message (e.g. in case of transmission of

Short Message).� In this case, the BSC receives a MEASUREMENT RESULT message which does not contain the

MEASUREMENT REPORT. The BSC takes into account the MS_TXPWR_CONF variable.


1.78

▼PC_COMMAND (MS, RED, MS_P_RED dB, > power min)� If MS_TXPWR > power min

then decrease MS_TXPWR by min(MS_P_RED, MAX_POW_RED, MS_TXPWR- power min)

� where MS_P_RED is evaluated by the following algorithm:

if (AV_RXLEV_UL_PC > U_RXLEV_UL_P) (good level)if (AV_RXQUAL_UL_PC ≥ U_RXQUAL_UL_P) (sufficient quality)then MS_P_RED = roundup[ MAX(POW_RED_FACTOR*

(AV_RXLEV_UL_PC- TARGET_RXLEV_UL)), 2dB]else MS_P_RED = roundup[ MAX ( POW_RED_FACTOR *

(AV_RXLEV_UL_PC- TARGET_RXLEV_UL), POW_RED_STEP_SIZE )]

else (good quality)MS_P_RED = POW_RED_STEP_SIZE

2.4 Radio link supervision and power controlMS Power Decrease Command computation


1.79

▼OFFSET_RXQUAL_FH

� This variable allows to take into account the frequency hopping in the RxQual evaluation (see Annex 2)

� Defined on a per cell basis

� Algorithm:

If Frequency hopping applied � then OFFSET_RXQUAL_FH = Offset_hopping_PC�Else OFFSET_RXQUAL_FH = 0

2.4 Radio link supervision and power controlFrequency Hopping cases

Annex 2

▼ In order to take into account the frequency hopping in the RXQUAL evaluation, the variable OFFSET_RXQUAL_FH is introduced.

▼ If on the corresponding channel, Frequency hopping is applied then OFFSET_RXQUAL_FH = Offset_Hopping_PC otherwise OFFSET_RXQUAL_FH = 0

▼ Offset_Hopping_PC is a parameter defined on a per cell basis.▼ PC Downlink in Frequency hopping case

� In this case, the BSC inhibits the BS power control on all the channels which use the BCCH carrier. The entity performing the BS power control in the BSC gets all the information concerning a new channel and decides whether to activate the BS power control for this channel. The power control must be inhibited when the frequency used by the new channel is the same as the frequency used for the BCCH in the BTS (cell) in which the channel is activated.

� For any channel which has the BCCH frequency in its hopping sequence (MA), the MS is measuring a very good downlink level each time it hops on the BCCH. To avoid that this results in a too optimistic average, it is possible to require from the MS not to include the BCCH measurement in the averages. This is achieved by setting the PWRC flag to 1 in the SYSTEM INFORMATION type 6 message sent by the BSS on the SACCH.

� If the channel is hopping only on the BCCH frequency (after a transmitter failure), it is considered as a non-hopping channel and it is concerned by the non-frequency hopping case.


1.80

▼Timers

� T_SDCCH_PC allows the inhibition of PC on SDCCH

� When a new power is required, the confirmation is awaited:

�MS_P_CON_ACK�BS_P_CON_ACK

� As soon as the new power is acknowledged, a fixed duration is awaited to trigger a new change of power, if necessary:

�MS_P_CON_INT�BS_P_CON_INT

2.4 Radio link supervision and power controlPower Control timers (1/2)

▼ The timer T_SDCCH_PC allows to inhibit the MS and BS power control on SDCCH.� This timer is changeable at the OMC-R level on a per cell basis. It is triggered upon receipt of the ESTABLISH

INDICATION message after SDCCH activation for immediate assignment procedure. As long as the timer runs, the power control is inhibited on SDCCH.

� If the timer expires, the power control will be enabled again on SDCCH.� If the timer is running at the sending of the RF CHANNEL RELEASE message, the timer is stopped.

▼ T_SDCCH_PC is useful in case of long SDCCH phases.▼ During SDCCH for call establishment, PC disabled should be preferred with a view to secure call setup. Nevertheless,

if SMS usage is very high, SDCCH phases may be long. In this case, to avoid interference, PC will be enabled after T_SDCCH_PC expiry (about 5s).

▼ After any PC command is sent to the MS, some time must be expected before MS_TXPWR_CONF (power confirmation sent by the MS on the uplink SACCH) can reach the desired value. The timer MS_P_CON_ACK is triggered after any power modification command to monitor that the desired transmission power MS_TXPWR is reached.

� If MS_P_CON_ACK elapses before the expected value of MS_TXPWR_CONF is received, the power control decision process is resumed immediately with the last MS_TXPWR_CONF received.

� If the expected value of MS_TXPWR_CONF is received before the timer MS_P_CON_ACK is elapsed, the timer MS_P_CON_ACK is stopped and the timer MS_P_CON_INT is triggered. Then the MS PC threshold comparison process is resumed with MS_TXPWR_CONF for the same MS as soon as MS_P_CON_INT expires.


1.81

▼ IF xx_P_CON_ACK is expiring, it is a system problem:

� Wrong setting of xx_P_CON_ACK (too short)

� No reception of power command by the MS

� a radio link recovery can be activated

� Problem on Abis

� repetition of BS power command

▼The expiry of P_CON_INT is a normal mechanism

2.4 Radio link supervision and power controlPower Control timers (2/2)


1.82

▼ LEVEL and QUALITY USED in EQUATION are average ones with window size A_QUAL_PC and A_LEV_PC

▼ BS POWER CONTROL INHIBITED ON BCCH frequency�BCCH must be emitted at the maximum level

▼MS dynamic constraint�minimum 2dB every 60 ms

▼Emitted power can be changed by radio link supervision algorithm

�Radio link supervision has a greater priority▼Activation of power control can slow down HO decision

� some causes can be triggered only if the MS (BTS) is emitting at the maximum power

2.4 Radio link supervision and power controlExtra information

▼ Interaction with radio link command� The MS power control function is informed of a radio link recovery by the radio link command function. Once the

indication is received, the PC command process is resumed immediately:� timer MS_P_CON_ACK is started (or reset and started if running),� If MS_P_CON_ACK elapses before the expected value of MS_TXPWR_CONF is received, the power

control decision process is resumed immediately with MS_TXPWR_CONF = min(MS_TXPWR_MAX,P).

▼ According to GSM Technical Specification 05.08 section 7.1, the BCCH carrier must be broadcast with a constant power in the cell. In this release of the ALCATEL BSS, this constant value is set to the maximum power allowed in the cell that is defined by the parameter BS_TXPWR_MAX.

� This means that all dedicated channels (TCH, SDCCH) which are on the BCCH frequency must always be transmitted with the maximum power, i.e. the BCCH power must not be changed by the BS power control function.


1.83

2.4 Radio link supervision and power controlPower Control: Training exercise (1/3)

▼Power control UL(Remark: Use the default parameters document)

� What happens if we do not use Frequency Hopping?� Why is it better to have A_LEV_PC=A_LEV_HO/2?� Thresholds:

� Lower QUAL of RX uplink = 3� High QUAL of RX uplink = 2� Lower LEV of RX uplink = -90dBm� Upper LEV of RX uplink = -75dBm� POW_RED_STEP_SIZE= 4� POW_INC_STEP_SIZE= 6

� Put the right threshold in the next slide chart

Time allowed:

25 minutes


1.84

Nb of case --->Nb of case --->Nb of case --->Nb of case ---> 1111 2222 3333 4444 5555 6666

AV_RXQUAL_UL_PCAV_RXQUAL_UL_PCAV_RXQUAL_UL_PCAV_RXQUAL_UL_PC 0 1 2 6 3 4

AV_RXLEV_UL_PCAV_RXLEV_UL_PCAV_RXLEV_UL_PCAV_RXLEV_UL_PC -98 -80 -73 -69 -86 -91

Power controlPower controlPower controlPower control

Delta valueDelta valueDelta valueDelta value

Qual

Lev


▼Power control UL

QUESTIONFor each case• PC triggered?• Step size value?

With POW_INC_FACTOR=0,6And POW_RED_FACTOR=0,6and MAX_POW_INC=MAX_POW_RED=8


1.85


▼Power control DL� Thresholds:

L_RXLEV_DL_P = -85dBm POW_INC_FACTOR = 0.6U_RXLEV_DL_P = -75dBm POW_RED_FACTOR = 0.8L_RXQUAL_DL_P = 2.9 MAX_POW_INC = 16dBU_RXQUAL_DL_P = 1 MAX_POW_RED = 16dBA_QUAL_PC = 4 BS_P_CON_ACK = 3sA_LEV_PC = 4 BS_TXPWR_MIN = -16dB

� Using the Trace Abis Excel file, find each parameter value:POW_INC_STEP_SIZE = ? BS_P_CON_INT = ?POW_RED_STEP_SIZE = ? OFFSET_RXQUAL_FH = 0 or 1 ?

� Which phenomenon can you observe as regards the successive PC commands?




2.5 Handover Detection





Handover Detection


Handover Management

Exercise


1.87

� SEND CONNECTED MS TO ANOTHER CELL

� When needed: “rescue/emergency” handover

� If useful: “better cell” handover

� TOWARD THE “BEST” CELL� From a radio point of view

� power budget

� level

� From a traffic point of view

� less loaded target

� From a dynamic point of view

� MS speed

� “history” of the call

� From an operator point of view

2.5 Handover DetectionHandover main objective

▼ Emergency intercell handovers:� These handovers are triggered when the call conditions deteriorate significantly in order to rescue the call. The

causes are: "too low quality" , "too low level", " too long MS-BS distance", “too short MS-BS distance”, "consecutive bad SACCH frames", "level dropping under high threshold".

▼ Better cell HO:� These handovers are triggered to improve the overall system traffic capacity. This spans: interference

reduction, signaling load reduction, traffic unbalance smoothing. The basic assumption for these handovers is that they should respect the cell planning decided by the operator.

� The causes are: "power budget" , "high level in neighbor lower layer cell for slow mobile", "high level in neighbor cell in the preferred band" and “traffic handover”.


1.88

▼The BSC is analyzing averaged measurement results

� active channel pre-processing (measurements averaging and book-keeping)

▼To detect need/utility to handover

�Handover detection process

▼To choose/rank target cells according to several criteria

�Candidate cell evaluation process

▼To perform the handover

�Handover management process

2.5 Handover DetectionPrinciples


1.89

Assignment of HO functions in the ALCATEL BSSAssignment of HO functions in the ALCATEL BSSAssignment of HO functions in the ALCATEL BSSAssignment of HO functions in the ALCATEL BSS

2.5 Handover DetectionFunctional entities

MSC



BTS BSC

HO DetectionHO CandidateCell Evaluation

HO management

HO protocol

HO Preparation

▼ The HO Preparation function can also be named "handover algorithms" as the algorithms described are the "heart" of this function.

� The ALCATEL handover preparation is derived from the basic algorithm found in Annex A of the GSM Technical Specification 05.08.

� The handover preparation is in charge of detecting a need for handover and proposing a list of target cells. Therefore it can be divided into two processes: handover detection and handover candidate cell evaluation .

▼ The handover detection process analyzes the radio measurements reported by the BTS and triggers the candidate cell evaluation process each time a handover cause (emergency or better cell type) is fulfilled.

▼ The handover candidate cell evaluation works out a list of possible candidate cells for the handover. This list is sorted according to the evaluation of each cell as well as the layer they belong to (in a hierarchical network) and the frequency band they use (in a multiband network).

▼ Once the handover preparation is completed, the handover decision and execution (handover management entity) is performed under the MSC or BSC control. The directed retry preparation is performed by the handover preparation function.

� Once the directed retry preparation is completed, the directed retry is performed either under the BSC control (internal directed retry) or under the MSC control (external directed retry).

▼ An example of implementation of these functions except for directed retry is given in the GSM Technical Specification 05.08.

▼ The handover preparation requires indirectly input parameters provided by the function in charge of the radio link measurements.

▼ Most of the input data required by the handover functions are provided by a function called: Active channel pre-processing.

▼ The figure above depicts in a general way:� the interconnections between these functions,� the implementation of these functions in the ALCATEL BSS.


1.90

▼Based on the contents of the measurement results

▼The BSC is computing the need or utility to trigger a handover

▼HO causes 25, split into 2 main categories:

� emergency handover

� quality, level, distance, etc.

� better cell handover

� power budget, traffic, etc.

▼ Some are specific to hierarchical and concentric architectures

2.5 Handover DetectionHandover causes detection

▼ The process is achieved in the BSC.▼ Each time a set of pre-processed (averaged) measurements is available, this process checks whether a handover is

needed. If the need for a handover is detected, the target cell evaluation process is triggered.

▼ In case of a handover alarm, the handover detection process gives to the cell evaluation process:� the preferred target cell layer: lower, upper or none.� the raw candidate cell list, which can be either all neighbors, or the subset which verify the handover causes

(plus other specific cells in particular cases). With each cell is given one of the handover causes which have been verified.

� The cause of handover.

▼ Four main handover categories are provided, depending on the cause of handover and the context of application. The context of application for a handover is either "intercell" (the handover is performed between two different cells) or "intracell" (the handover is performed in the same cell).

▼ The detection of a need for handover is performed through handover causes which are going to be detailed.▼ The cause of handover is based either on a situation of emergency (this cause is therefore called "emergency cause")

or on the existence of better conditions. In this last case, the name of the cause depends on the context of application: for intercell handovers, it is called "Better cell cause". For intracell handovers, it is called "Better zone cause", as it is applied only in the case of interzone handovers in concentric or multiband cells.


1.91

▼ HO causes for standard networks

� cause 2 : too low quality on the uplink� cause 3 : too low level on the uplink � cause 4 : too low quality on the downlink� cause 5 : too low level on the downlink � cause 6 : Too long distance between the MS and the BTS � cause 10 : too low level on the uplink in the inner zone� cause 11 : too low level on the downlink in the inner zone� cause 15 : high interference on the uplink (intracell HO)� cause 16 : high interference on the downlink (intracell HO)� cause 26 : AMR channel adaptation HO (HR to FR)� cause 24 : general capture Modified in B8� cause 12 : power budget evaluation� cause 23 : traffic Modified in B8� cause 13 : Outer zone level Uplink & Downlink� cause 27 : AMR channel adaptation HO (FR to HR)� cause 20 : forced directed retry� cause 28 : Fast traffic HO� cause 29 : TFO HO

2.5 Handover DetectionHandover causesB8

Better conditions HO

Emergency HO

▼ HO causes for Extended Cells:� Emergency causes

� cause 22: too short MS-BTS distance

▼ HO causes for hierarchical or multiband network:� Emergency causes

� cause 7: consecutive bad SACCH frames received in a microcell� cause 17: too low level on the uplink in a microcell compared to a high threshold� cause 18: too low level on the downlink in a microcell compared to a high threshold

� Better causes� cause 14: high level in neighbor lower layer cell for slow mobile� cause 21: high level in neighbor cell in the preferred band


1.92

▼CAUSE 2: too low quality on the Uplink

AV_RXQUAL_UL_HO > L_RXQUAL_UL_H + OFFSET_RXQUAL_FH

and AV_RXLEV_UL_HO <= RXLEV_UL_IH

and MS_TXPWR = min (P, MS_TXPWR_MAX)

and EN_RXQUAL_UL= ENABLE

� Size of window for averaging quality: A_QUAL_HO

� Size of window for averaging level: A_LEV_HO

2.5 Handover DetectionHandover Cause 2: UL Quality

QUAL

LEV

▼ Quality and Level causes (2, 3, 4, 5, 15, 16)▼ The aim of these causes is to keep the call going when the radio link is degrading otherwise the radio link failure might

be detected and the call released. These causes wait generally for the power control process to increase the BTS and MS power to their maximum values, except for the causes specific to microcellular environment.

▼ Handover on "too low level" is used to avoid situations where the interference level is low, while the attenuation is quite high. These conditions may appear for example in big city streets which enable a line of sight propagation from the BTS antenna. There is in this case a risk of abrupt quality degradation, if the MS moves away from the line of sight street.

▼ In case of simultaneous low-level and low-quality signals, an intercell handover is requested.


1.93

▼CAUSE 3: too low level on the uplink

AV_RXQUAL_UL_HO <= L_RXQUAL_UL_H + OFFSET_RXQUAL_FH

and AV_RXLEV_UL_HO < L_RXLEV_UL_H

and MS_TXPWR = min (P, MS_TXPWR_MAX)

and EN_RXLEV_UL= ENABLE



2.5 Handover DetectionHandover Cause 3: UL Level

QUAL

LEV


1.94

▼CAUSE 4: too low quality on the downlink

AV_RXQUAL_DL_HO > L_RXQUAL_DL_H + OFFSET_RXQUAL_FH

and AV_RXLEV_DL_HO <= RXLEV_DL_IH

and BS_TXPWR = BS_TXPWR_MAX

and EN_RXQUAL_DL= ENABLE



2.5 Handover DetectionHandover Cause 4: DL Quality

QUAL

LEV


1.95

▼ CAUSE 5: too low level on the downlink

AV_RXQUAL_DL_HO <= L_RXQUAL_DL_H + OFFSET_RXQUAL_FH

AV_RXLEV_DL_HO < L_RXLEV_DL_H

BS_TXPWR = BS_TXPWR_MAX

and EN_RXLEV_DL= ENABLE

�Size of window for averaging quality: A_QUAL_HO

�Size of window for averaging level: A_LEV_HO

2.5 Handover DetectionHandover Cause 5: DL Level

QUAL

LEV


1.96

▼CAUSE 6: Too long distance between the MS and the BTS

AV_RANGE_HO > U_TIME_ADVANCE

and EN_DIST_HO= ENABLE

� Size of window for distance averaging: A_RANGE_HO

2.5 Handover DetectionHandover Cause 6: Distance

▼ This cause is used when a dominant cell provides a lot of scattered coverages inside other cells, due to propagation conditions of the operational network. The consequence of these spurious coverages is the probable production of a high level of co-channel interference.

▼ This cause is different from the others as it is more preventive. It does not make use of the propagation conditions of a call. It just does not allow an MS to talk to a BTS if it is too far away.

▼ It may happen for example that some peculiar propagation conditions exist at one point in time that provide exceptional quality and level although the serving BTS is far and another is closer and should be the one the mobile should be connected to if the conditions were normal.

▼ It may then happen that these exceptional conditions suddenly drop and the link is lost, which would not have happened if the mobile had been connected to the closest cell. So for these reasons, this cause does not wait for the power control to react.


1.97

▼Emergency handovers specific to concentric cells� Intracell handovers from inner to outer zone

� cause 10: too low level on the uplink in inner zone

� cause 11: too low level on the downlink in inner zone

▼May be triggered

�From inner zone of a concentric cell�Towards outer zone, same cell

2.5 Handover DetectionHandover algorithms for concentric cells

INNEROUTER


1.98

▼CAUSE 10: too low level on the uplink in the inner zone

AV_RXLEV_UL_HO < RXLEV_UL_ZONE

and MS_TXPWR = min (P, MS_TXPWR_MAX_INNER)

� Averaging window: A_LEV_HO

2.5 Handover DetectionHandover algorithms for concentric cells: cause 10


1.99

▼CAUSE 11: too low level on the downlink in the inner zone

AV_RXLEV_DL_HO < RXLEV_DL_ZONE

and BS_TXPWR = BS_TXPWR_MAX_INNER

� Averaging window: A_LEV_HO

2.5 Handover DetectionHandover algorithms for concentric cells: cause 11


1.100

▼CAUSE 13: too high level on UL and DL in the outer zone� Better condition intracell handover

� If the cell is a multi-band cell, cause 13 is checked only for multi-band MSs

▼May be triggered� From outer zone of a concentric cell

� Towards inner zone, same cell

2.5 Handover DetectionHandover algorithms for concentric cells: cause 13 (1/6)

INNEROUTER


1.101

▼CAUSE 13: too high level on UL and DL in the outer zoneAV_RXLEV_UL_HO > RXLEV_UL_ZONE +

+ ZONE_HO_HYST_UL ++ (MS_TXPWR - MS_TXPWR_MAX_INNER) ++ PING_PONG_MARGIN(0,call_ref)

and AV_RXLEV_DL_HO > RXLEV_DL_ZONE ++ ZONE_HO_HYST_DL ++ (BS_TXPWR - BS_TXPWR_MAX_INNER) ++ PING_PONG_MARGIN(0,call_ref)

and AV_RXLEV_NCELL_BIS(n) <= neighbour_RXLEV(0,n)and EN_CAUSE_13 = ENABLE (B7)and EN_BETTER_ZONE_HO = ENABLE

� Averaging windows: A_LEV_HO and A_PBGT_HO (for n)



1.102

▼ZONE_HO_HYST_UL

� UL static hysteresis for interzone HO from outer to inner� In case of multi-band cell, should take into account the difference of

propagation between GSM and DCS

� Added to cause 10 threshold RXLEV_UL_ZONE

▼ZONE_HO_HYST_DL

� DL static hysteresis for interzone HO from outer to inner� In case of multi-band cell, should take into account the difference of

propagation between GSM and DCS and the difference of BTS transmission power in the two bands

� Added to cause 11 threshold RXLEV_DL_ZONE



1.103

▼PING_PONG_MARGIN(0,call_ref)

� Penalty PING_PONG_HCP put on cause 13 if� The immediately preceding zone in which the call has been is the

inner zone of the serving cell

� And The last handover was not external intracell

� And T_HCP is still running

� PING_PONG_MARGIN(0,call_ref) = 0� If the call was not previously in serving’s inner zone

� Or T_HCP has expired


INNEROUTER


1.104

▼ neighbour_RXLEV(0,n)

� Concentric cells are designed to create an INNER zone� protected from external interferers

� and creating no interferences on other cells

� … to be able to face more aggressive frequency reuse in INNER zone TRXs

� neighbour_RXLEV(0,n) tuning enables to avoid handovers if the MS position will lead to interferences

� the condition is checked towards all neighbor cells belonging to the same layer and band than the serving cell


INNEROUTER

INNER zoneinterferer 1

INNER zoneinterferer 2

?


1.105

▼EN_CAUSE_13

� Load balance between inner and outer zones may be allowed by setting EN_LOAD_BALANCE = ENABLE

� If EN_LOAD_BALANCE = ENABLE� If INNER zone is less loaded than OUTER,

EN_CAUSE_13 = ENABLE

� If INNER zone is more loaded than OUTER,EN_CAUSE_13 = DISABLE

� If EN_LOAD_BALANCE = DISABLE� EN_CAUSE_13 = ENABLE



1.106

▼Outgoing intercell handovers from concentric cells

� As explained here before, the MS located in a concentric cell can make intercell, emergency or better condition HO regardless their current zone�For example, an MS located in the INNER zone of a

concentric cell can make directly a HO cause 12 towards another cell, WITHOUT having to trigger any cause 10 or 11 to the OUTER zone before

2.5 Handover DetectionOutgoing intercell handovers from concentric Cell

▼ The only restrictions are linked to EN_MULTI-BAND_PBGT_HO and EN_BI-BAND_MS parameters.


1.107

▼ Incoming intercell handovers towards a concentric cell

� In case an MS is making an incoming handover towards a concentric cell (due to outer PBGT measurements,etc.), a TCH may be allocated� either in the INNER or in the OUTER zone, as for call setup

� depending on radio conditions

� In case of a multi-band cell, if the MS is not multi-band, it will always be sent to the OUTER zone

2.5 Handover DetectionIncoming intercell handovers towards a Concentric Cell (1/2)

INNEROUTER ??


1.108

▼Use part of Handover cause 13 algorithm on each potential target

▼ IF Cell(n) is external

�The MS is directed to the OUTER zone of (n)

▼ELSE (cell(n) is internal)

� IFAV_RXLEV_NCELL(n) > RXLEV_DL_ZONE + ZONE_HO_HYST_DL +

+ (BS_TXPWR - BS_TXPWR_MAX_INNER)

and EN_BETTER_ZONE_HO = ENABLE

�The MS is directed towards the INNER zone

� ELSE

�The MS is directed towards the OUTER zone

2.5 Handover DetectionIncoming intercell handovers towards a Concentric Cell (2/2)


1.109

▼CAUSE 12: Power budget

� Decision based mainly on comparison of serving and neighbor cells for:

� downlink level of serving and neighbor cells�maximum emitting level of MS

� Aiming at decreasing UL & DL emitted power

� Should be the “normal” handover type

� no matter of emergency

2.5 Handover DetectionHandover Cause 12: Power Budget (1/11)

▼ In this case, there is another cell with a better power budget i.e., the link quality can be improved or maintained with a reduced transmit power of both the MS and the BTS. The radio link is not degraded but there is the opportunity to decrease the overall interference level by changing the serving cell of the given MS.

▼ In conjunction with power control, it presents the advantage to keep the interference as low as possible, since it minimizes the path loss between the BTS and the MS.

▼ This cause is especially designed to cope with the requirement that the mobile should be connected with the cell with which the lowest possible output powers are used. To assess which of the cells is this "best cell", the algorithm performs every measurement reporting period the comparison of the path loss in the current and in the neighbor cell. This is a feature special to GSM which is made possible because the mobile measures the adjacent cell signal levels and reports the six best ones.

▼ This power budget gives the difference in path loss between the current cell and the adjacent cells reported by the mobile.

▼ When PBGT(n) is greater than 0, then the path loss from cell n is less than the path loss from the serving cell and thus the radiated power in the downlink direction, and therefore in the uplink direction as well, will be lower in cell n than in the current cell.

▼ However it would not be advisable to hand over the MS to another cell as soon as PBGT is greater than 0, because the MS would probably oscillate between the two adjacent cells as the propagation conditions vary. An hysteresismechanism is implemented to avoid this undesirable effect.

▼ No PBGT between different layers.▼ Ok between different bands if EN_INTERBAND_PBGT_HO = 1


1.110

▼CAUSE 12: Power budget equation

PBGT(n) = AV_RXLEV_NCELL(n) - AV_RXLEV_PBGT_HO

- (BS_TXPWR_MAX – AV_BS_TXPWR_HO)- (MS_TXPWR_MAX(n) – MS_TXPWR_MAX)- PING_PONG_MARGIN(n, call_ref)


▼ The MS may be handed over from the serving cell indexed 0 to a neighbor cell indexed n only if the power budget exceeds the handover Margin(0,n). The handover Margin(0,n) can be modified according to the traffic situation in the serving cell and the neighbor cell n. In this way, power budget handover can be delayed towards a loaded cell and traffic load handover can be triggered from a loaded cell. Once the MS is handed over, the same algorithm is applied in the new cell, and a new PBGT is computed (which will be close to the opposite value of PBGT in the old cell) and compared to a new HOMargin. (Thus, the global hysteresis (from cell 0 to cell n and back to cell 0) is the sum of the two HOMargins).

▼ However, It is still possible that a ping-pong mechanism is created by different handover causes, for instance a handover may be triggered towards a neighbor cell for bad quality, but in the neighbor cell, a handover back may be triggered for power budget reasons. In order to avoid this, an additional anti-ping-pong mechanism is implemented in the power budget calculation. It enables to penalize for a certain time the cell on which the call has been before.

▼ In case of handover from SDCCH to SDCCH, this cause does not take the traffic situation into account.▼ In multiband cell environment, the mobile can operate in a different band than the frequency band of the BCCHs. This

can lead to circular ping-pong handovers from the inner zone if the new band is DCS 1800 or to the impossibility to trigger PBGT handovers from the inner zone if the preferred band is GSM 900.

▼ To avoid this problem, when the MS is in the inner zone of a multiband cell, it may be handed over from the serving cell indexed 0 to a neighbor multiband cell indexed n only if the power budget exceeds the handover Margin(0,n) plus the offset handover margin which allows to handicap or favor the PBGT (In the inner zone, the cause “power budget” is only checked between multiband cells, in a way to maintain the MS in the preferred band).

▼ The offset handover margin can possibly be used in concentric cells.


1.111


� AV_RXLEV_NCELL� received level of BCCH of neighbor cell

� AV_RXLEV_PBGT_HO� received level of serving cell (BCCH or not)

� AV_RXLEV_NCELL - AV_RXLEV_PBGT_HO� the highest is the best neighbor cell� but serving might not be at the maximum level (with DL

power control)� necessity to have a corrective factor


call_ref)MARGIN(n,PING_PONG_

MAX)MS_TWXPWR_MAX(n)(MS_TXPWR_

)R_HOAV_BS_TXPWMAX(BS_TXPWR_

PBGT(n)

−−−

−−−= BGT_HOAV_RXLEV_PCELL(n)AV_RXLEV_N

▼ ∆∆∆∆ BCCH = AV_RXLEV_NCELL(n) - (AV_RXLEV_PBGT_HO + C)

� with C = BS_TXPWR_MAX - AV_BS_TXPWR_HO.▼ This corresponds to the difference of received BCCH signal levels.

� A correction factor C is taken into account for the serving cell, because the received signal level (i.e. AV_RXLEV_PBGT_HO) may not be measured on BCCH.


1.112


� BS_TXPWR_MAX – AV_BS_TXPWR_HO

� BS_TXPWR_MAX are attenuations, not absolute level� = (“bts_max_power”+BS_TXPWR_MAX) -

(“bts_max_power”+AV_BS_TXPWR_HO)� AV_BS_TXPWR_HO: average of BS_POWER over A_PBGT_HO

measurements� corrective factor used to compensate for the fact that the serving

cell may not emit at the maximum level

� AV_RXLEV_NCELL-[AV_RXLEV_PBGT_HO+(BS_TXPWR_MAX-AV_BS_TXPWR_HO)]� compare received level of neighbor and serving cells as if the

serving one was emitting at the maximum level





PBGT(n)

−−−



1.113


� MS_TXPWR_MAX(n)

� maximum emitting power for the MS in neighbor cell n

� MS_TXPWR_MAX

� maximum emitting power for the MS in the serving cell

� MS_TXPWR_MAX(n) - MS_TXPWR_MAX� Corrective factor to compensate for the difference of maximum power of

each cell� MS_TXPWR_MAX(n) - MS_TXPWR_MAX = bts_max_power(n) -

bts_max_power� which should be the case if delta_path_loss is equilibrated� if not exact, can be corrected with HO_MARGIN(0,n)





PBGT(n)

−−−


▼ Then, another correction factor must be taken into account because the maximum BS powers of the serving and neighbor cells may be different:

∆∆∆∆ TXPWR= MS_TXPWR_MAX(n) - MS_TXPWR_MAX.

▼ As the first step of calculation is based on the downlink parameters, this correction factor should be based on the maximum BS powers used in the serving and neighbor cells.

▼ Two reasons (which are not completely de-correlated) for not using the BS powers can be envisaged:� for a given cell, the GSM does not specify formally the maximum BS power of the neighbor cells. Only

BS_TXPWR_MAX is defined (it is sent on the air interface),� it is not easy for the evaluating BSC to know the maximum BS powers of the neighbor cells.

▼ The use of the maximum MS powers requires that the difference of MS powers is equal to the difference of BS powers. This condition is met in most cases. If it is not the case, the difference can be corrected by the operator with the HO_MARGIN(0,n) parameter (HO hysteresis).

▼ PBGT >0: the neighbor cell is more advantageous as the path loss is lower than in the current cell.▼ PBGT <0: the serving cell is more advantageous than the current cell.


1.114


� Hysteresis to avoid ping-pong HO

� Static hysteresis defined for each couple of cells: HO_MARGIN (0,n)� can also be used to correct delta_path_loss

� “Dynamic” penalty for call coming from cell n: ping_pong_margin(n,call_ref)� penalty applied during a limited duration: T_HCP� not used if call arrived with a forced directed retry� penalty defined on a cell basis




PBGT(n)

−−−



▼ The main drawback of this handover category is the risk of "ping-pong " effect, which is an oscillating back and forth handover between two (or three) cells. As the "better cell" handovers are meant to find the "best cell", the variation of the radio conditions will trigger a big amount of better cell handovers, if the algorithms have a too sensitive reaction. Hence, some mechanisms are forecast, in order to prevent these oscillations from occurring repeatedly at given places.

▼ PING_PONG_MARGIN(n,call_ref) is a penalty put on the cell n if:� it is the immediately precedent cell on which the call has been,� this cell belongs to the same BSC as the serving cell,� the call has not performed a forced directed retry towards the serving cell,� less than T_HCP seconds have elapsed since the last handover.

� In this case PING_PONG_MARGIN(n,call_ref) = PING_PONG_HCP▼ If the call was not precedently on cell n, or if the preceding cell was external, or if the call has just performed a forced

directed retry, or if the timer T_HCP has expired, � then PING_PONG_MARGIN(n,call_ref) = 0


1.115


� ping_pong_margin example


Case 1: OK

Case 2: ping-pong in normal caseOK with ping_pong_margin

Case 3: Not a ping-pong caseOK with ping_pong_margin and T_HCP

▼ This chart shows the efficiency of the anti-ping_pong mechanism.▼ But, never forget that anti-ping-pong mechanism distorts the serving areas of the cells. ▼ This is why interference problems might occur when enabling this mechanism. Tuning PING_PONG_HCP parameter is

thus very important.

▼ Warning: this mechanism is not applied for emergency handovers (new mechanism in B7 exists for capture HO, based on T_INHIBIT_CPT timer).


1.116


If EN_TRAFFIC_HO(0,n)=ENABLE

Then PBGT(n) > HO_MARGIN(0,n) + OFFSET_HO_MARGIN_INNER

+ max(0, DELTA_HO_MARGIN(0,n))(n=1…BTSnum)

Else PBGT(n) > HO _MARGIN(0,n) + OFFSET_HO_MARGIN_INNER

AND AV_RXLEV_PBGT_HO ≤ RXLEV_LIMIT_PBGT_HO

AND EN_PBGT_HO = ENABLE

� Size of window for level averaging: A_PBGT_HO


▼ Cause 12 HO is correlated with HO cause 23. This is why there are two equations according to the activation of HO cause 23 (EN_TRAFFIC_HO).


1.117


� Mechanism to avoid PBGT HO if the level from the serving cell is high enoughRXLEV_LIMIT_PBGT_HO: threshold above which it is not necessary to trigger a

handover on power budget

AV_RXLEV_PBGT_HO: average of the received levels over A_PBGT_HO measurements

� Specific to particular algorithms (not mentioned in this course)OFFSET_HO_MARGIN_INNER: offset which allows to take into account the radio

differences between outer and inner zones (especially incase of multi-band cells)


▼ RXLEV_LIMIT_PBGT_HO: Dense Network Handover Regulator features▼ The feature aims at optimizing the better cell handovers, especially in the microcellular environment.▼ In very dense networks, there is a lot of overlapping between adjacent cells: a better cell handover will be realized very

often. Since B6, the Alcatel BSS tunes the number of handovers performed to the accurate need by taking into account the level received by the serving cell.

▼ Therefore, the best trade-off between quality of speech and intempestive handovers is achieved.▼ Why?

� Especially in microcellular environment (where cell radius is smaller), the better cell HO (based on Power Budget) is likely to be performed at a high rate to the detriment of the quality.

� But it is necessary to maintain the better cell HO.

▼ How?

� With a modification of the power budget triggering cause.

▼ Principles:

� HO cause 12 (Power Budget HO) is modified and takes into account the received downlink level of the serving cell (new criterion): if the received level is high enough, there is no need to perform an HO.

▼ Consequence:

� Less HOs when the number of overlapping cells is high.

W/O B6W/O B6W/O B6W/O B6 WITH B6WITH B6WITH B6WITH B6


1.118


� Specific to traffic considerationsDELTA_HO_MARGIN(0,n) : evaluated according to the traffic situation of the

serving cell and the neighbor cell n (Traffic_load(n)) in the following way:If Traffic_load(0) = high and Traffic_load(n) = low,

DELTA_HO_MARGIN(0,n) = - DELTA_DEC_HO_marginIf Traffic_load(0) = low and Traffic_load(n) = high,

DELTA_HO_MARGIN(0,n) = DELTA_INC_HO_marginElse

DELTA_HO_MARGIN(0,n) = 0

PhilosophyThis mechanism aims at penalizing cause 12 detection when the traffic in the serving cell is low and is high in the cell n.


▼ DELTA_HO_MARGIN(0,n) is evaluated according to the traffic situation of the serving cell and the neighbor cell n (Traffic_load(n)) in the following way:

� If Traffic_load(0)=high and Traffic_load(n)=low� DELTA_HO_MARGIN(0,n)= -DELTA_DEC_HO_margin

� If Traffic_load(0)=low and Traffic_load(n)=high� DELTA_HO_MARGIN(0,n)= DELTA_INC_HO_margin

� else DELTA_HO_MARGIN(0,n)=0where DELTA_DEC_HO_margin allows the cause 23 (traffic handover) detection.

▼ When the traffic in the serving cell is high and is low in the cell n:� DELTA_INC_HO_margin allows to penalize the cause 12 detection when the traffic in the serving cell is low

and is high in the cell n.

Note:In the case of concentric or multiband cells, if the channel is in the inner zone (ZONE_TYPE = INNER),BS_TXPWR_MAX and MS_TXPWR_MAX in equation must be replaced by BS_TXPWR_MAX_INNER and MS_TXPWR_MAX_INNER respectively.If the channel is in the outer zone (ZONE_TYPE = OUTER), the formulation of equation is not changed.

Note: The value of PBGT(n) is calculated every SACCH period for each neighbor cell n whose measures are kept in the book-keeping list.


1.119


� Traffic_load() is a function managed for every cell of a BSC� Traffic_load() can have three values:

� high: cell is loaded� low: cell is unloaded� indefinite: cell is neither loaded nor unloaded

� Traffic_load() value is modified according to the long term traffic evaluation algorithm using the following parameters:�A_TRAFFIC_LOAD, N_TRAFFIC_LOAD,

HIGH_TRAFFIC_LOAD, IND_TRAFFIC_LOAD, LOW_TRAFFIC_LOAD: can be modified per cell

�TCH_INFO_PERIOD: cannot be modified


Annex 3

▼ TCH_INFO_PERIOD = 5s period used by the BSC to count the number of free TCHs.


1.120

2.5 Handover DetectionHandover Cause 23: Traffic (1/2)

▼ CAUSE 23: Traffic Handover

DELTA_HO_MARGIN(0,n) < 0dB

AND PBGT(n) > HO_MARGIN(0,n) + OFFSET_HO_MARGIN_INNER

+ DELTA_HO_MARGIN(0,n) (n=1…BTSnum)

AND EN_TRAFFIC_HO(0,n) = ENABLE

� Size of window for level averaging: A_PBGT_HO

New condition for multiband cells in B8 (see comments)

B8

▼ The principle of this handover is to reduce the size of the serving cell when it is high-loaded relatively to a low-loaded cell.

▼ When the mobile moves away from the BTS, the power budget will increase and a better cell handover will be triggered earlier.

▼ It is recommended to inhibit Traffic handover towards 1-TRX cells. These cells do not have enough resources to receive incoming handovers due to congestion of neighbor cells. Moreover because of the great variation of traffic in the 1-TRX cells, traffic load is never considered as low.

▼ This cause is inhibited for handover from SDCCH to SDCCH.

▼ Cause 23 is checked over all the neighboring cells belonging to the same layer. It means that it is checked between cells whose CELL_LAYER_TYPE is single or upper, between cells whose CELL_LAYER_TYPE is lower, and between cells whose CELL_LAYER_TYPE is indoor.

▼ In addition to the condition on the cell layer type, the cell frequency band condition for checking Cause 23 is as follows whether or not the MS is in the inner zone of a multi-band cell:

� a) The MS is not in the inner zone of a multi-band cell� If the flag EN_MULTI-BAND_PBGT_HO is set to “disabled”, Cause 23 must not be checked between

cells which use different frequency bands (i.e cells having different CELL_BAND_TYPE).� If the flag EN_MULTI-BAND_PBGT_HO is set to “enabled”, Cause 23 will be checked over all the

neighboring cells without any cell frequency band restriction.� b) The MS is in the inner zone of a multi-band cell

� If the flag EN_MULTI-BAND_PBGT_HO is set to “disabled”, Cause 23 is checked over all the neighboring cell multi-band cells (FREQUENCY_RANGE= PGSM-DCS1800 or EGSM-DCS1800) which belong to the same BSC as the serving cell.

� If the flag EN_MULTI-BAND_PBGT_HO is set to “enabled”, Cause 23 will be checked over all the neighboring cells without any cell frequency band restriction.


1.121

2.5 Handover DetectionHandover Cause 23: Traffic (2/2)

▼CAUSE 23: Traffic Handover

� DELTA_HO_MARGIN(0,n) computation is already described in Cause

12 HO

� DELTA_HO_MARGIN(0,n) < 0dB means that

� The serving cell is loaded

� The target cell is unloaded

� PBGT(n) > HO_MARGIN(0,n) + OFFSET_HO_MARGIN_INNER

+ DELTA_HO_MARGIN(0,n) (n=1…BTSnum)

� This constraint is less discriminative than Cause 12

� In specific traffic distribution, this cause is triggered before cause

12

B8


1.122

2.5 Handover DetectionHandover Cause 12 & 23 interworking

▼Cause 12 & 23: A dynamic way to handle traffic load

B8

▼ The figure represents the triggering areas of PBGT and traffic handovers according to the traffic load in the serving cell and in the neighbor cell.


1.123

▼Directed Retry is:

� an SDCCH to TCH intercell handover

� Triggered during call setup procedure

▼ If the serving cell is completely congested, the MS is allocated an SPCCH

▼ If no TCH is available, the MS is queued

� Under certain conditions, the MS obtains TCH in another cell

▼SDCCH-TCH handover on:

� better condition or emergency causes = Directed Retry

� cause 20 = Forced Directed Retry

▼ Internal and External Directed Retries are possible (since B6.2)

2.5 Handover Detection Directed Retry principles


1.124

▼Directed Retry

� Set on a per cell basis with parameter EN_DR

� Same behavior as TCH HO

� Intercell handover causes are checked (i.e. all HO causes except 10, 11 and 13 (concentric cells) and causes 15 and 16 (intracell HO))

� candidate cell evaluation process: same as for TCH HO

2.5 Handover Detection Directed Retry


1.125

▼CAUSE 20: Forced Directed Retry

AV_RXLEV_NCELL_DR(n) > L_RXLEV_NCELL_DR(n)

And EN_FORCED_DR = ENABLE

� EN_FORCED_DR value is only relevant if EN_DR = true

� AV_RXLEV_NCELL_DR(n) is calculated with A_PBGT_DR window

� if less than A_PBGT_DR samples are available, the average value is calculated with the available samples and the averaging window is filled in with -110 dBm

2.5 Handover Detection Forced Directed Retry: cause 20


1.126

▼Pre-ranking

� using PREF_LAYER, PRIORITY(0,n), frequency band

▼Filtering process� AV_RXLEV_NCELL_DR(n) > RXLEVmin(n)

+max(0,MS_TXPWR_MAX(n) - P)

� Number of free TCHs t(n) > FREElevel_DR(n)

▼Remaining cells are sorted according their PBGT_DR(n) (averaging window A_PBGT_DR )� PBGT_DR(n) = AV_RXLEV_NCELL_DR(n) - AV_RXLEV_PBGT_DR

- (BS_TXPWR_MAX - BS_TXPWR) - (MS_TXPWR_MAX(n) - MS_TXPWR_MAX)

2.5 Handover Detection FDR: Candidate cell evaluation


1.127

▼ L_RXLEV_NCELL_DR(n) : level required in the neighbor cell n

�The parameter considered is the one set in the neighborcell

�The default value depends on network architecture�See next slide

▼Freelevel_DR(n) : number of free TCH channels required in the neighbor cell n

�The parameter considered is the one set in the neighborcell

�Default value = 0 to 4 TCHs (linked to the nb of TRXs)

▼A_PBGT_DR : Averaging window

�Default value = 4 SACCHs

2.5 Handover Detection FDR: parameters


1.128

▼CAUSE 24: general capture

� capture handover� Modified in B8:Inhibition of capture handovers for “Single layer

serving cell”

� May be triggered� From all cells

� Towards all cells except serving

� Can be used to capture traffic by any cell, whatever its type, band, etc.

2.5 Handover DetectionCause 24: general captureB8


1.129

▼CAUSE 24: general capture

AV_RXLEV_NCELL(n) > L_RXLEV_CPT_HO(0,n) + max (0, [MS_TXPWR_MAX(n) - P])

and Traffic_load(0) = CAPTURE_TRAFFIC_CONDITIONand Traffic_load(n) ≠≠≠≠ HIGHand EN_GENERAL_CAPTURE_HO = ENABLE

� Size of window for averaging level: A_PBGT_HO� CAPTURE_TRAFFIC_CONDITION can take 3 values:

ANY_LOAD (default), HIGH, NOT_LOW� Anti ping-pong: not checked if T_INHIBIT_CPT is running –

new in B8 for single layer

2.5 Handover DetectionCause 24: general captureB8

▼ Case the serving cell is in the upper or single layer (CELL_LAYER_TYPE(n0) = upper or single):

▼ Condition 1 : The immediately preceding cell n-1 is in the indoor or lower layer, i.e. CELL_LAYER_TYPE(n–1) = lower or indoor, or the frequency band of the immediately preceding cell n-1 is different from the frequency band of the serving cell n0, i.e. CELL_BAND_TYPE(n–1) <> CELL_BAND_TYPE(n0).

▼ Condition 2 : The call has previously performed i) an emergency internal handover on quality (Cause 2, 4, and 7) towards the serving cell or ii) an external handover with the A interface GSM cause “uplink quality or downlink quality” and there is a bi-directional adjacency link between the preceding external cell n-1and the serving cell n0.

� If Conditions 1 and 2 are fulfilled the timer T_INHIBIT_CPT is started


1.130

2.5 Handover DetectionHandover Cause 28: Fast Traffic HO (1/4)

▼CAUSE 28: Fast Traffic HO� Push out of a cell a mobile in dedicated mode to allow a queued

request to be served in the serving cell

� Complement the current traffic HO (Cause 23), for sudden trafficpeaks (no averaging window used)

� More efficient where the overlap of adjacent cells is reduced

HO

New call attempt Most appropriate MS to be pushed out

Congested cell

New call attempt

HOMost appropriated MS

to be pushed out

Upper Layer Cell

� AV_RXLEV_NCELL( n) > L_RXLEV_NCELL_DR( n) + max(0,[MS_TXPWR_MAX( n)-P])� The threshold L_RXLEV_NCELL_DR(n) is the observed level from the neighbor cell n at the border of

the area where fast traffic handovers are enabled. This threshold fixes the size of the overlapping area where fast traffic handovers can be performed. It should be greater than RXLEVmin(n).

� And t(n) > FREElevel_DR(n)� FREElevel_DR(n) is the minimum threshold of free TCHs in the neighbor cell n for forced directed retry

and fast traffic handover.� t(n) is the absolute number of free (dual rate) TCHs in the neighbor cell n.� For external cells, t( n) is fixed to the arbitrary value t(n) = 255. Therefore, setting FREElevel_DR(n) to

255 for an external cell inhibits outgoing external fast traffic handover towards this cell. Setting FREElevel_DR(n) to any other value will allow outgoing external fast traffic handover towards this cell.

� EN_CAUSE_28 = enable� The flag EN_CAUSE_28 is not an OMC flag but a HOP flag.


1.131


▼CAUSE 28: Fast Traffic Handover

� Cause 28 is only checked if the channel of the candidate MS can support the channel rate (HR or FR) required by the queued request:

� HO is triggered when a request is queued at the top of the queue

QUEUED REQUEST CANDIDATE MS

HR HR or

FR on dual rate TRX

FR FR (whatever the TRX type)


1.132


▼CAUSE 28: Fast Traffic Handover equation

AV_RXLEV_NCELL(n) > L_RXLEV_NCELL_DR(n) + max(0,[MS_TXPWR_MAX(n)-P])

AND t(n) > FREElevel_DR(n)AND EN_CAUSE_28 = ENABLEAND EN_FAST_TRAFFIC_HO = ENABLE

� Size of window for averaging level: A_PBGT_DR

�Same thresholds and window as Cause 20 (Forced Directed Retry)

�EN_CAUSE_28 is an internal HOP process variable


1.133


▼CAUSE 28: Fast Traffic Handover process

HandoverHandoverHandoverHandoverManagementManagementManagementManagement

HandoverHandoverHandoverHandoverPreparationPreparationPreparationPreparation

ResourceResourceResourceResourceAllocationAllocationAllocationAllocation

ManagementManagementManagementManagementQueued request referenceChannel rate of queued request

Fast Traffic HO Request

Queued request referenceReference of MS can perform HO

Fast Traffic HO Acknowledge

Cause number (28)

Reference of the call to handover (which corresponds to the first candidate MS received)

Start HO

EN_CAUSE_28=enable

HO alarm

Cause 28?

EN_CAUSE_28=disable

YES

NO

Request still queued?

YESEND NO

Check first 2 conditions of Cause 28

OK

END

NOK

Assignment request queued

T_FILTER is started

▼ HO cause 28 process:� If EN_FAST_TRAFFIC_HO = enable, when an assignment request (or external emergency HO request) is

queued, the RAM process sends to the HOP process a Fast Traffic HO request which contains the queued request reference and its channel rate.

� Then, HO cause 28 becomes checkable (EN_CAUSE_28=enable).� Once an HO alarm for cause 28 is triggered, the flag EN_CAUSE_28 is set to “disable” so as not to perform

more than one handover. In the same time, the HOP process gets back to the RAM process a Fast Traffic HO Acknowledge which contains the queued request reference and the reference of the MS that can perform HO.

� If several answers are sent to the RAM process, only the first one corresponding to the queued request is taken into account.

� The RAM process checks if the request is still queued. If that is so, the RAM process asks the HOP process to start HO for this mobile; otherwise the process is stopped.

� Once the HOP process receives this message, the first two conditions of Cause 28 (good enough level, enough free resources in the target cell) are checked one more time. If the conditions are fulfilled, the HOP process sends an alarm to the HOM entity and the timer T_FILTER is started ; otherwise the process is stopped.

Note : the first two conditions of cause 28 are tested twice in order to be sure that the candidate cells are still valid when the « cause 28 start HO » message is received from the RAM process.


1.134

▼CAUSE 15: High interference on the uplink � Intracell HO

AV_RXQUAL_UL_HO > THR_RXQUAL_CAUSE_15 +OFFSET_RXQUAL_FH

AND AV_RXLEV_UL_HO > RXLEV_UL_IHAND EN_CAUSE_15 = ENABLEAND [ no previous intracell handover for this connection

failed OR EN_INTRACELL_REPEATED = ENABLE ]

� Size of window for averaging quality: A_QUAL_HO� Size of window for averaging level: A_LEV_HO

2.5 Handover DetectionHandover Cause 15: UL Interference

▼ THR_RXQUAL_CAUSE_15 and EN_CAUSE_15 are not parameters but variables defined just after.▼ In B7:

� New causes (26 & 27) introduced due to AMR support� Cause 26 is an emergency condition:� Intracell HO: speech codec from AMR-HR to AMR-FR� Cause 27 is a better condition� Intracell HO: speech codec from AMR-FR to AMR-HR

� Causes 15 & 16 are modified due to AMR support� Specifics enablers and thresholds for AMR calls� AMR emergency HO (cause 26) is triggered if cause 15 or 16 has already been triggered

� Cause 29 is created for intracell handover due to TFO� Codec sharing and optimization for MTM calls


1.135

▼CAUSE 16: High interference on the downlink

� Intracell HOAV_RXQUAL_DL_HO > THR_RXQUAL_CAUSE_16 + OFFSET_RXQUAL_FH

AND AV_RXLEV_DL_HO > RXLEV_DL_IHAND EN_CAUSE_16 = ENABLEAND [ no previous intracell handover for this connection

failed OR EN_INTRACELL_REPEATED = ENABLE ]

� Size of window for averaging quality: A_QUAL_HO� Size of window for averaging level: A_LEV_HO

2.5 Handover DetectionHandover Cause 15: DL Interference

▼ THR_RXQUAL_CAUSE_16 and EN_CAUSE_16 are not parameters but variables defined after.


1.136

2.5 Handover DetectionNew parameters for causes 15 & 16

▼CAUSE 15 and CAUSE 16:

� THR_RXQUAL_CAUSE_15 (or 16) and EN_CAUSE_15 (or 16) are specific to HOP

� THR_RXQUAL_CAUSE_15 (or 16) =

� L_RXQUAL_XX_H for a non AMR call (same threshold as CAUSE 2 or CAUSE 4)

� L_RXQUAL_XX_H_AMR for an AMR call

� EN_ CAUSE _15 (or 16) =

�EN_INTRA_XX for a non AMR call�EN_INTRA_XX_AMR for an AMR call

▼ XX = UL or DL▼ For a non AMR call, the thresholds used are identical to the ones used for CAUSE 2 and CAUSE 4.▼ In this case and if EN_INTRACELL_REPEATED = DISABLE, when aN HO CAUSE 15 (or 16) fails, it can be modified

as UPLINK (or DOWLINK) QUALITY, HO CAUSE 2 (respectively HO CAUSE 4).


1.137

2.5 Handover DetectionAdaptive Multi-rate codec (AMR)

▼Principles:

� Two consecutive encodings: speech coding and channel coding

� With current codecs , the share of each coding is FIXED (not optimized)

11.4 Kbit/s (HR TS)5.6 Kbit/s (HR)

22.8 Kbit/s (FR TS)13 Kbit/s (FR)12.2 Kbit/s (EFR)

Audio Speech coding Channel coding Radio

Speech informationSpeech informationSpeech informationSpeech information“useful part”“useful part”“useful part”“useful part”

Speech protectionSpeech protectionSpeech protectionSpeech protection“against degradation”“against degradation”“against degradation”“against degradation”

▼ Speech coding contains speech information (the “useful” part).▼ Channel coding protects speech information (against radio degradations).

▼ The main speech codec currently used in GSM networks, speech Full Rate, is quite old. It has been specified more than 10 years ago. Around 1992, to increase network capacity, GSM has specified a half rate speech codec. But this codec showed strong limitations in terms of speech quality, especially for mobile to mobile calls (double transcodingdegrades very much the speech quality of the half rate codec) and under poor radio conditions.

▼ Recently, studies on AMR have been launched to provide a solution to:• Increase speech quality in full rate and half rate,• Increase network capacity by offering a good half rate solution,• Use a long-term solution, to avoid adding more and more codecs handled independently from the others.


1.138

2.5 Handover DetectionAMR: codec and channel adaptation

� AMR uses a variable balance between speech coding and channel coding (CODEC Mode Adaptation )

�Choice between FR and HR Codecs: Channel Mode Adaptation

Variable channel Variable channel Variable channel Variable channel coding ratecoding ratecoding ratecoding rate

Variable speech Variable speech Variable speech Variable speech coding ratecoding ratecoding ratecoding rate

11.4 Kbit/s (HR TS)

22.8 Kbit/s (FR TS)

12.212.212.212.210.210.210.210.27.957.957.957.957.47.47.47.46.76.76.76.75.95.95.95.95.155.155.155.154.754.754.754.75

AMR HR 7.95 not supported

Audio Speech coding Channel coding Radio

▼ In order to adapt the intermediate rate, a set of speech codecs has been defined by ETSI to be used by AMR:� When radio conditions are good, increases speech information.� When radio conditions are bad, protects speech information.

▼ Full Rate: Alcatel implementation is fully compliant with GSM recommendations. All these AMR FR codec modes are supported. In particular, the Alcatel BSS has implemented the 7.95, 5.9 and 4.75 codec modes which use polynomials of constraint length 7 to ensure a high protection.

▼ Half Rate: Alcatel implementation supports 5 out of 6 AMR HR codec modes (AMR HR 7.95 is not supported) which are fully compliant with GSM recommendations. In particular, the Alcatel BSS has implemented the 4.75 codec mode which uses polynomials of constraint length 7 to ensure a high protection.

▼ During a call, only a subset out of these 8 codecs is used. The subset can include from 1 to 4 codecs. It is up to the operator to define its own codec subset. In particular, he can define a codec subset limited to the common codec modes supported by all the BSSs of its network (some BSSs may not be able to support all of them due to implementability problems).The codec subset defined by the operator is the same in the uplink and in the downlink.

▼ Codec Mode adaptation:� dynamic change from one codec to another, using the same channel (FR or HR).� metric used: C/I (Carrier over interference ratio).

▼ Channel Mode adaptation:� change from one FR channel to an HR one and vice-versa independently from the codec mode.� metric used: RX_QUAL uplink and downlink.


1.139

� Based on adaptive trade-off between the share of throughput given to speech coding and the one given to channel coding (speech protection)

� Depends on radio conditions estimated in real-time

conditions conditions

Channel coding = speech protection

Speech coding = speech information

Medium radio Bad radio Good radioconditions

2.5 Handover DetectionAMR codec adaptation objective

▼ The AMR principle is to have a set of codecs and, for any radio conditions, to use the one with the best speech quality.� Under good radio conditions, a codec with a high bit rate is used. Speech is encoded with more information so

the quality is better. In the channel coding, only little place is left for redundancy.� Under poor radio conditions, a codec with a low bit rate is chosen. Speech is encoded with less information, but

this information can be well protected due to redundancy in the channel coding.▼ The BSS adapts dynamically the codec in uplink direction and in downlink direction, taking into account the C/I

measured by the BTS (for uplink adaptation) and by the MS (for downlink adaptation).▼ The codec used in the uplink and used in the downlink can be different: the adaptation is independent in each direction.▼ This permits to use an optimal codec for each C/I value of each direction, as indicated in the figure below.

C/I [dB]

SpeechQuality[dBQ]

or[MOS]

High bit rate (for example 12.2 kbit/s: EFR)

Medium bit rate (for example 7.95 kbit/s)

Low bit rate (for example 5.90 kbit/s)

AMR-FR with codec subset (12.2, 7.95, 5.90)


1.140

2.5 Handover DetectionAMR: codec mode adaptation (1/3)

▼Codec mode adaptation� Only a subset out of these codecs can be used� This subset may include from 1 to 4 codecs� The same codec subset is used for both the Uplink and the

Downlink� Uplink codec mode adaptation:

� For each SACCH frame, the BTS compares C/I value to the threshold corresponding to the current codec (belonging to the codec subset defined by the operator)

� Downlink codec mode adaptation:�Same process as uplink adaptation�Nevertheless, the BTS remains the master

� Unrelated processes ⇒ uplink and downlink codecs may be different at a given time


1.141

2.5 Handover DetectionAMR codec mode adaptation (2/3)

▼ The Codec mode can be modified on one frame out of two (CMI / CMC-CMR).

▼ Decision based on thresholds (OMC-R settable), for the uplink and the downlink

AMR_FR_THR_3 + AMR_FR_HYST

High

Low

CODEC_MODE_4 (less robust)

C/Inorm

AMR_FR_THR_3

CODEC_MODE_3

CODEC_MODE_2

CODEC_MODE_1 (most robust)


AMR_FR_THR_2


AMR_FR_THR_1


1.142

2.5 Handover DetectionAMR: codec mode adaptation (3/3)

▼Codec mode adaptation� Uplink adaptation

� Downlink adaptation

Codec Mode Command(new codec mode)

MS BTS TC

C/I evaluation & thresholds comparison

Codec Mode Indication(new codec mode) Codec Mode Indication

(new codec mode)

Codec Mode Request(new codec mode)

MS BTS TC

C/I evaluation & thresholds comparison

Codec Mode Request(new codec mode)

Codec Mode Indication(new codec mode)

Codec Mode Indication(new codec mode)


1.143

2.5 Handover DetectionAMR: codec and channel mode adaptation

▼Codec mode adaptation is dynamically performed through a set of pre-defined “codec modes”:

� In FR mode:

� In HR mode:

▼Choice between HR and FR (Channel mode adaptation) is doneat call setup and during call through HO causes 26 & 27

coding rate

acousticFrom

partsSpeech coding Channel coding To end of chain

Variable speech

11.4 Kbit/s (HR TS)

7.46.75.95.154.75

Fromacoustic

partsSpeech coding Channel coding To end of chain

Variable speechcoding rate

22.8 Kbit/s (FR TS)

12.210.27.957.46.75.95.154.75

▼ Codec mode adaptation:▼ The codec mode adaptation is the dynamic change from one codec to another codec, using the same channel (FR or

HR). This adaptation is performed by the layer 1 of the BTS. It is transparent for the BSC and the layer 3 of the BTS.

▼ The metric used for codec mode adaptation is the evaluation of the ratio: signal over noise.


1.144

2.5 Handover DetectionAMR gain

▼AMR: always gives end user the best satisfaction

� Comparison between different codecs in terms of capacity and quality:

SPEECH QUALITY REQUIREMENT

CAPACITY REQUIREMENT

FR � � EFR ☺ � HR � ☺

AMR-FR ☺ � AMR-HR � ☺

AMR-FR + AMR-HR ☺☺☺☺ ☺☺☺☺

▼ The main speech codec currently used in GSM networks, speech Full Rate, is quite old. It has been specified more than 10 years ago.

▼ Around 1992, to increase network capacity, GSM has specified a half rate speech codec. But this codec showed strong limitations in terms of speech quality, especially for mobile to mobile calls (double transcoding degrades very much the speech quality of the half rate codec) and under poor radio conditions.

▼ A few years later, when GSM started to be introduced in North America, American operators asked for an improved speech codec for full rate channels. Indeed speech quality was a major argument for customers used to have a good speech quality with analog systems. For that issue, EFR was specified for GSM.

▼ Recently, studies on AMR have been launched to provide a solution to:• Increase speech quality in full rate and half rate,• Increase network capacity by offering a good half rate solution,• Use a long-term solution, to avoid adding more and more codecs handled independently from the others,• Take into account Tandem Free Operation (TFO), especially between MSs on half rate on one side and on full

rate on the other side.


1.145

▼FR / HR discrimination� cell load AV_LOAD() computed from

� load samples = NB_BUSY_TS / NB_TS * 100� non sliding window (LOAD_EV_PERIOD) averaging process

2.5 Handover DetectionAMR: TCH allocation

THR_FR_LOAD_U_SV1= 80%

THR_FR_LOAD_U_SV3=60%

THR_FR_LOAD_L_SV1=50%

THR_FR_LOAD_L_SV3=40%

100%

AV_LOAD()

time

▼ Load samples are computed by the BSC every TCH_INFO_PERIOD = 5 seconds.▼ LOAD_EV_PERIOD is the averaging window size for cell load computation. It is equal to 12 but can be changed at the

OMC-R level on a per cell basis.▼ Therefore cell load process has a periodicity of 1mn by default (TCH_INFO_PERIOD*LOAD_EV_PERIOD).▼ The allocation of Half rate resources is decided upon the load evaluation in the serving cell.▼ AMR HR (HR SV3) offers a better speech quality than HR SV1. The Alcatel BSS offers thus the possibility to define a

set of thresholds specific for AMR. If the load increases, AMR HR capable MSs can be the first to be allocated in HR (HR SV3) for load reasons, and if the load still increases, then all the HR capable MSs can be allocated in HR (HR SV1 & HR SV3) for load reasons.

� This is why two variables of load are defined: LOAD_SV3 and LOAD_SV1.▼ Each load variable is calculated through its own threshold set: the thresholds related to the variable LOAD_SV3

(THR_FR_LOAD_U_SV3 and THR_FR_LOAD_L_SV3) are less restrictive than the ones related to the variable LOAD_SV1 (THR_FR_LOAD_U_SV1 and THR_FR_LOAD_L_SV1).

� As a consequence, if the load of the cell increases, then the variable LOAD_SV3 will first equal TRUE, and if the load still increases, the variable LOAD_SV1 will then equal TRUE.

▼ The variable LOAD_SV1 corresponds to a level of load where it is important to put as many MSs on half rate TCH as possible: HR SV3 or HR SV1.

▼ The same computation is done to compute LOAD_SV3 with the thresholds: THR_FR_LOAD_U_SV3 and THR_FR_LOAD_L_SV3 with the following relations:

� THR_FR_LOAD_L_SV3 ≤ THR_FR_LOAD_U_SV3� THR_FR_LOAD_U_SV3 ≤ THR_FR_LOAD_U_SV1� THR_FR_LOAD_L_SV3 ≤ THR_FR_LOAD_L_SV1

Previous stateAV_LOAD

LOAD_SV1 = FALSE LOAD_SV1 = TRUE

AV_LOAD ≤ THR_FR_LOAD_L_SV1 LOAD_SV1 = FALSE LOAD_SV1 = FALSE

THR_FR_LOAD_L_SV1 <AV_LOAD ≤

THR_FR_LOAD_U_SV1

LOAD_SV1 = FALSE LOAD_SV1 = TRUE

THR_FR_LOAD_U_SV1 < AV_LOAD LOAD_SV1 = TRUE LOAD_SV1 = TRUE


1.146

2.5 Handover DetectionCause 26: AMR HR to FR HO (1/4)

▼CAUSE 26: AMR channel adaptation HO (HR to FR)

� Cause 26 is triggered if :

�Current channel rate is HR�Current channel is dual rate and changes are allowed�AMR_FR speech codec is allowed:

�EN_AMR_FR = ENABLE


1.147


▼CAUSE 26: AMR channel adaptation HO (HR to FR) equation

[ a previous intracell HO cause 15 or 16 has been triggered for this call inthe serving cell

OR

EN_INTRA_DL_AMR = DISABLE and EN_INTRA_UL_AMR = DISABLE ]

AND AV_RXQUAL_UL_CA_HR_FR > THR_RXQUAL_CA + OFFSET_CA + OFFSET_RXQUAL_FH and AV_RXLEV_UL_HO > RXLEV_UL_IHORAV_RXQUAL_DL_CA_HR_FR > THR_RXQUAL_CA + OFFSET_CA + OFFSET_RXQUAL_FH and AV_RXLEV_DL_HO > RXLEV_DL_IH

AND EN_AMR_CA = ENABLE

� Size of window for averaging quality: A_QUAL_CA_HR_FR


1.148



� THR_RXQUAL_CA and OFFSET_CA are set as follows :

if LOAD_SV3(0) = false then

THR_RXQUAL_CA = THR_RXQUAL_CA_NORMAL

OFFSET_CA = OFFSET_CA_NORMAL

if LOAD_SV3(0) = true then

THR_RXQUAL_CA = THR_RXQUAL_CA_HIGH

OFFSET_CA = OFFSET_CA_HIGH


1.149



� Calculation of LOAD_SV3(0):If previous value of LOAD_SV3 = false then

if AV_LOAD > THR_FR_LOAD_U_SV3 thenLOAD_SV3 = true

elseLOAD_SV3 = false

Else (if previous value of LOAD_SV3 = true then)if AV_LOAD <= THR_FR_LOAD_L_SV3 then

LOAD_SV3 = falseelse

LOAD_SV3 = trueAnnex 3


1.150

2.5 Handover DetectionCause 27: AMR FR to HR HO (1/2)

▼CAUSE 27: AMR channel adaptation HO (FR to HR)

▼Cause 27 is triggered if :

�Current channel rate is FR�Current channel is dual rate and changes are allowed�AMR_HR speech codec is allowed:

�EN_AMR_HR = ENABLE


1.151

2.5 Handover DetectionCause 27: AMR FR to HR HO (2/2)

▼CAUSE 27: AMR channel adaptation HO (FR to HR) equation

AV_RXQUAL_UL_CA_FR_HR <= THR_RXQUAL_CA + OFFSET_RXQUAL_FH

AND

AV_RXQUAL_DL_CA_FR_HR <= THR_RXQUAL_CA + OFFSET_RXQUAL_FH

AND EN_AMR_CA = ENABLE

� Size of window for averaging quality: A_QUAL_CA_FR_HR


1.152

2.5 Handover DetectionCause 26 & 27 interworking

▼Cause 26 & 27 interaction

THR_RXQUAL_CA_NORMAL

THR_RXQUAL_CA_HIGH

good quality: 0

bad quality: 7

THR_RXQUAL_CA_NORMAL +OFFSET_CA_NORMAL

THR_RXQUAL_CA_HIGH +OFFSET_CA_HIGH

Load = FALSE Load = TRUE

HO cause 26

HO cause 27

Full Rate

Half Rate

Quality

HO cause 26

HO cause 27

Full Rate

Half Rate


1.153

2.5 Handover DetectionIntroduction to TFO (1/2)

▼Tandem Free Operation (TFO) solution

Double transcoding without TFO

No transcoding with TFO

µ

TC TC

Codec GSM (a)(8 or 16 Kbit/s)

Codec GSM (b)(8 or 16 Kbit/s)

A/ law(64 Kbit/s)

TC TC

Codec GSM(8 or 16 Kbit/s)

▼ The Tandem Free Operation (TFO) feature is a way to avoid double transcoding in mobile to mobile speech calls.▼ Indeed without TFO, one GSM codec type is used between the first mobile and the first transcoder, then the speech is

transcoded into A/µ law between transcoders and finally this speech is transcoded again into a second GSM codec type (which may be the same as the first one) between the second transcoder and the second mobile.

▼ With TFO, after call establishment, both BSSs at each side are able to negotiate a common GSM codec type which is then used from one mobile to the other mobile. This negotiation is performed through in-band signaling between transcoders.


1.154

2.5 Handover DetectionIntroduction to TFO (2/2)

▼ Applicability: Only MS to MS speech calls

▼ TFO is based on information exchanged between transcoders

MS TRAU TRAU MS

BTS BTSBSCBSC MSC MSC

64 kbit/s Speech Samples carrying- TFO Frames on the LSB containing

- Compressed speech samples- Control bits- TFO Messages

- Original PCM speech samples on the MSB

IPE IPE


1.155

2.5 Handover DetectionTFO principles

▼ In the case of first allocation (normal assignment at call setup, inter-BSS handover, intra-BSS handover where no TFO was previously on-going):

Codec match

TFO mode ON

Intracell HO

Codec mismatch

Exchange of codeccapabilities

Common codecfound

New callsetup



Normal operation

No common codecfound


1.156

2.5 Handover DetectionCause 29: TFO HO

▼CAUSE 29: TFO HO

� Intracell HO used in case of codec mismatch between two MSs calling, in order to match their speech codec

� No radio measurements needed � No priority and may be triggered at any time

� Conditions:

HO_INTRACELL_ALLOWED = ENABLEAND

EN_TFO_MATCH = ENABLE


1.157

2.5 Handover DetectionCause 29: TFO parameters (1/5)

▼EN_TFO� enables/disables the feature, per cell

▼EN_TFO_MATCH� enables/disables resolution of codec mismatch, per cell

▼EN_TFO_OPT� enables/disables codec optimization, per cell

▼FORCE_TFO_VS_AMR� enables/disables the basic functions of TFO for GSM EFR, FR and

HR codec types when the current codec is AMR FR or AMR HR

▼FORCE_TFO_HR_WHEN_LOADED� controls the establishment of TFO in HR when the cell is loaded

▼KEEP_CODEC_HO� indicates if the BSC tries to keep the same codec in case of internal

intercell HO

▼ Codec mismatch:

� At call setup for a mobile to mobile speech call, when both BSSs do not use the same codec type, a codec mismatch occurs. If a common codec type can be found, either one or possibly both BSSs perform an intracellhandover to use the common codec type found. Afterwards TFO can be started using this common codec type. Codec mismatch resolution is authorized in the BSC using an O&M flag: EN_TFO_MATCH. This flag is forwarded to the TC, via the BTS.

▼ Codec optimization:

� At call setup for a mobile to mobile speech call, it can occur that a first common codec type can be found but a better speech quality would be provided with another common codec type. Once both BSSs operate in Tandem Free, they exchange their complete codec capabilities, to try to find a better codec type than the current one. Codec optimization is authorized in the BSC using an O&M flag : EN_TFO_OPT. This flag is forwarded to the TC, via the BTS.

▼ Classification of codec types :

� In all cases, TFO is considered better as any tandeming configuration. In TFO, EFR is considered as better than FR, considered as better than HR.

▼ Force TFO vs. AMR :� TFO + AMR is not supported in this implementation of TFO. In the normal operation, a call established with

AMR will not initiate a TFO negotiation. The goal of the function Force TFO vs. AMR is to allow a call, established with AMR to initiate a TFO negotiation and, if possible, to change of codec type to FR, HR or EFR to establish TFO.

▼ In-Path Equipments (IPEs):

� TFO can only be activated if TFO frames (at 8 or 16 Kbit/s) can be sent transparently through the public switching network. In-path equipments are equipments such as echo cancelers or A/µ law converters that modify the 64 Kbit/s speech signal. Such equipments need to be deactivated for TFO calls.


1.158


▼EN_TFO_OPT: enables/disables codec optimization, per cell

� Allows new TFO negotiation on an on-going MTM call to find a better common codec

� For example, HR is used at both sides, but FR is possible too

�HO cause 29 will be triggered on both sides towards best codec


1.159


▼FORCE_TFO_VS_AMR:

� TFO AMR not specified

�Call setup in AMR is not followed by TFO negotiation� FORCE_TFO_VS_AMR enables HO cause 29 after

AMR call establishment towards best TFO codec

MS / cell cap: AMR/HR/EFR/FR MS / cell cap: HR/EFR/FR

DISABLED

(ETSI implementation)

ENABLED

(Alcatel patent)

The MS using AMR, could use HR/EFR/FR

The MS can use HR/EFR/FRTFO not possible

The MS can only use HR/EFR/FR

The MS can use HR/EFR/FR EFR + TFO

FORCE_TFO_VS_AMR


1.160


▼FORCE_TFO_HR_WHEN_LOADED:

� Gives control on load regulation precedence vs. TFO

� 3 values: TFO_HR_NOT_FORCED, TFO_HR_ONLY, TFO_HR_PREFERRED enable different behaviours in case of loaded cell

Loaded cellMS / cell cap: HR/EFR/FR

Unloaded cell MS / cell cap: HR/EFR/FR

DISABLED

(ETSI implementation)

ENABLED

(Alcatel patent)

The MS can use HR/EFR/FR

The MS can use HR/EFR/FR EFR + TFO

The MS can only use HR

The MS can use HR/EFR/FR HR + TFO

FORCE_TFO_HR_WHEN_LOADED


1.161


▼ KEEP_CODEC_HO� keeps the same codec type in the new cell in case of internal intercell

HO in order to avoid resolving a new mismatch codec situation� Avoids double speech quality transition:

TFO --> non-TFO --> TFO� 3 possible behaviors:

� TFO_CALLS_ONLY: codec is preferably kept in case of internal intercell HO for TFO calls only

� ALL_CALLS: codec is preferably kept in case of internal intercellHO for all calls (whatever the TFO state)

� FREE: the choice of the codec type is free and depends on the situation in the target cell


1.162

XXXXXXXXXXXXXXXXXXXXXXX

DONWLINK INTERFERENCECAUSE 16 Intracell

UPLINK INTERFERENCECAUSE 15 Intracell

HR to FR CHANNEL ADAPTATIONCAUSE 26 Intracell

TOO LOW LEVEL DL INNERCAUSE 11

TOO LOW LEVEL UL INNERCAUSE 10

DISTANCECAUSE 6

DOWNLINK LEVELCAUSE 5

UPLINK LEVELCAUSE 3

DONWLINK QUALITYCAUSE 4

UPLINK QUALITYCAUSE 2

EMERGENCYHANDOVER

FAST TRAFFIC HOCAUSE 28

FORCED DIRECTED RETRYCAUSE 20

FR to HR CHANNEL ADAPTATIONCAUSE 27 Intracell

OUTER UL/DL LEVELCAUSE 13

TRAFFICCAUSE 23

POWER BUDGETCAUSE 12

CAPTURE HANDOVERCAUSE 24

BETTER CONDITIONHANDOVER

HANDOVERPRIORITIES

2.5 Handover DetectionHandover causes priorities

▼ The causes 24, 12 and 23 have the same priority. Nevertheless, if a cell is a candidate for both causes, triggered in the same time, it is kept only for cause 12.

▼ Dealing with all available causes, we get the following list:� Emergency: 7 > 17 > 18 > 2 > 4 > 3 > 5 > 6 > 22 > 10 > 11 > 26 > 15 > 16� Better conditions: 21=14=24=12=23 > 13 > 27 > 20 > 28� 29 has no priority


1.163

▼Emergency causes1- What is the HO cause 2?

2- Which is the flag to activate the HO cause 2?

2.5 Handover DetectionTraining exercises (1/16)

Time allowed:

45 minutes


1.164

Nb of case Nb of case Nb of case Nb of case ------------>>>> 1111 2222 3333 4444 5555 6666

AV_RXQUAL_UL_HOAV_RXQUAL_UL_HOAV_RXQUAL_UL_HOAV_RXQUAL_UL_HO 4 1 3 4 4 4

AV_RXLEV_UL_HOAV_RXLEV_UL_HOAV_RXLEV_UL_HOAV_RXLEV_UL_HO - 81 - 79 - 75 - 70 - 69 - 72

Current MS PowerCurrent MS PowerCurrent MS PowerCurrent MS Power 33 33 33 33 33 29 (0,8 w)

HO cause 2HO cause 2HO cause 2HO cause 2 : YES/NO: YES/NO: YES/NO: YES/NO ????

▼Emergency causesComplete the diagram below and fill in the chart with:

L_RXQUAL_UL_H = 3

RXLEV_UL_IH = -70 dBm

QUAL

LEV



1.165

▼Better condition causes (simple case)� There are only 2W cells and 2W MS

�EN_TRAFFIC_HO(0,n)=Disable� No Ping-Pong margin� HO_MARGIN(0,n)=5 dB

Fill up the chart:

Nb of case --->Nb of case --->Nb of case --->Nb of case ---> 1111 2222 3333 4444 5555 6666

AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n) - 70 - 70 - 80 - 70 - 70 - 75

AV_RXLEV_PBGT_HOAV_RXLEV_PBGT_HOAV_RXLEV_PBGT_HOAV_RXLEV_PBGT_HO - 80 - 70 - 75 - 75 - 79 - 96

PBGT(n)PBGT(n)PBGT(n)PBGT(n)

HO cause 12: YES/NOHO cause 12: YES/NOHO cause 12: YES/NOHO cause 12: YES/NO ? ? ? ?

Serving cell

Ncell

MS



1.166

Serving cell

▼Better condition causes (ping-pong case)� EN_TRAFFIC_HO(0,n)=Disable

� Ping-Pong margin PING_PONG_HCP=15db

T_HCP =15s� HO_MARGIN(0,n)=5 dB A_PBGT_HO = 8 SACCH

A n to 0 HO has just been triggered, what happens after 4s?

Ncell

MS


AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n) - 70 - 70 - 80 - 70 - 70 - 75

AV_RXLEV_PBGT_HOAV_RXLEV_PBGT_HOAV_RXLEV_PBGT_HOAV_RXLEV_PBGT_HO - 80 - 70 - 75 - 75 - 79 - 96

PBGT(n) PBGT(n) PBGT(n) PBGT(n) « a » only 10 0 - 5 5 9 21

HO cause 12: YES/NOHO cause 12: YES/NOHO cause 12: YES/NOHO cause 12: YES/NO ???? PBGT > HO margin

YES NO NO NO YES YES

PING_PONG_HCP=15 PING_PONG_HCP=15 PING_PONG_HCP=15 PING_PONG_HCP=15 ----> PBGT(n)> PBGT(n)> PBGT(n)> PBGT(n)

HO cause 12: YES/NOHO cause 12: YES/NOHO cause 12: YES/NOHO cause 12: YES/NO ????



1.167

▼Training exercise: Handover Detection

� Better condition causes (traffic case)

� EN_TRAFFIC_HO(0,n)=Enable

� No Ping-Pong margin

� HO_MARGIN(0,n)=5 dB

� DELTA_DEC_HO_margin=5dB

� DELTA_INC_HO_margin=5dB

Serving cell

Ncell

MS

2.5 Handover DetectionTraining exercise (5/16)


1.168

Serving cell

▼Better condition causes (traffic case)Fill up the chart:

DELTA_HO_MARGIN(0,n)

0: tr lowN: tr low

0: tr highN: tr low

0: tr highN: tr low

0: tr lowN: tr high

Traffic distribution

Cause 23 HO: YES/NO?

Cause 12 HO: YES/NO?

PBGT(n)

-80 dBm-80 dBm-80 dBm-80 dBmAV_RXLEV_PBGT_HO

-71 dBm-76 dBm-71 dBm-71 dBmAV_RXLEV_NCELL(n)

4321Number of case

Ncell

MS



1.169


▼ Channel adaptation (cause 26 and cause 27)

1- Why is it recommended to have A_QUAL_CA_FR_HR ≥ A_QUAL_CA_HR_FR ?

2- An operator may be willing to:

- Under normal load, use only HR calls for quality 0

- Under high load, use HR calls for qualities 0 to 3, with an hysteresis of 1

Find the thresholds and offsets for normal and high load:THR_RXQUAL_CA_NORMAL = ? OFFSET_CA_NORMAL = ?

THR_RXQUAL_CA_HIGH = ? OFFSET_CA_HIGH = ?


1.170


▼Channel adaptation (cause 26 and cause 27)� EN_INTRA_XX_AMR = Disable� RXLEV_XX_IH = -110 dBm� OFFSET_RXQUAL_FH = 0� A_QUAL_CA_FR_HR =4 and A_QUAL_CA_HR_FR = 2

Use the previous thresholds and fill up the chart:

UL_QUAL 0 1 2 3 3 1 1 0 0 1 DL_QUAL 0 0 1 1 1 0 0 2 4 3 LOAD_SV3 false false false false true true true true true true AV_RXQUAL_UL_CA_HR_FR AV_RXQUAL_DL_CA_HR_FR AV_RXQUAL_UL_CA_FR_HR AV_RXQUAL_DL_CA_FR_HR CHANNEL TYPE FR FR FR


1.171

▼capture HO (Cause 24 )� There are only 2W cells and 2W MS

� L_RXLEV_CPT_HO(0,n) = -85dBm�EN_GENERAL_CAPTURE_HO = ENABLE

Fill up the chart:


AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n)AV_RXLEV_NCELL(n) - 70 - 70 - 80 - 70 - 70 -85

CAPTURE_TRAFFIC_CONDITION NOT_LOW HIGH ANY_LOAD HIGH HIGH HIGH

TRAFFIC_LOAD(0) HIGH LOW INDEFINITE HIGH LOW HIGH

TRAFFIC_LOAD(n) HIGH LOW INDEFINITE LOW LOW LOW

HO cause 24HO cause 24HO cause 24HO cause 24: Y: Y: Y: YES/NOES/NOES/NOES/NO ????

Serving cell

Ncell

MS



1.172


▼Fast Traffic HO (cause 28)Find the appropriate candidate MS for this queued request:� Channel rate required: HR� L_RXLEV_NCELL_DR(n) = -85 dBm (whatever n)� FREElevel_DR(n) = 1 (whatever n)� Channel rate: MS1�FR on Full rate TRX, MS2�HR, MS3�FR on Dual rate TRX� t(n) for neighbor cells: t(1)=1, t(2)=2, t(3)=2� AV_RXLEV_NCELL(n) in dBm:

MS Neighbors 1 2 3

1 -82 -85 -78

2 -79 -86 -92

3 -90 -82 -89


1.173


▼TFO HO (cause 29): after call setupFind the 2 speech version types of the following MS to MS call � EN_TFO = enable, EN_TFO_MATCH = enable

� FORCE_TFO_HR_WHEN_LOADED = TFO_HR_NOT_FORCED


Unloaded cell MS / cell cap: EFR/FR

After call setup TCH = ? TCH = ?

After TFO negotiation TCH = ? TCH = ?


1.174



� FORCE_TFO_HR_WHEN_LOADED = TFO_HR_ONLY






1.175



� FORCE_TFO_HR_WHEN_LOADED = TFO_HR_PREFERRED






1.176









1.177


▼TFO HO (cause 29): after handoverFind the speech version types of the following MS to MS call � EN_TFO = enable, EN_TFO_MATCH = enable


1. KEEP_CODEC_HO = TFO_CALLS_ONLY

2. KEEP_CODEC_HO = FREE


MS1


MS2

HO


MS2

call setup+ TFO negotiation

?

?TFO?

?

TFO?


1.178

?


▼TFO HO (cause 29): after handoverFind the speech version types of the following MS to MS call � EN_TFO = enable, EN_TFO_MATCH = enable


� KEEP_CODEC_HO = TFO_CALLS_ONLY

1. EN_TFO_OPT = disable

2. EN_TFO_OPT = enable

Unloaded cellMS / cell cap: HR/EFR/FR

MS1

Loaded cell MS / cell cap: HR/EFR/FR

MS2

HO


MS2

call setup+ TFO negotiation

?TFO?

?

TFO?




2.6 Handover Candidate Cell Evaluation





Handover Detection


Handover Management

Exercise


1.180

▼Used to rank potential target cells:

� Ranking based on radio characteristics

� Ranking based on operator preferences

� Ranking based on traffic intensity

2.6 Handover Candidate Cell EvaluationPrinciples



BTS BSC


HO management

MSC

HO protocol

HO Preparation

▼ Handover candidate cell evaluation� The process is performed in the BSC.� Once a need for handover is detected, this process looks for possible target cells (except if it is an intracell

handover or an interzone handover) and provides the BSC entity in charge of the HO decision and execution entity with a list of candidate cells and their respective HO cause.


1.181

MEASUREMENT PREPROCESSING

according

• A_LEV_HO• A_QUAL_HO• A_PBGT_HO• A_RANGE_HO

Performed every SACCH

Measurem

entresult

HO_DETECTION

cause 2: uplink qualitycause 3: uplink levelcause 4: downlink qualitycause 5: downlink levelcause 6: distancecause 12: power budget

Performed every SACCH

Preprocess

measurem

ent

Raw cell list

cell 1: cause C2cell 2: cause C2cell 3: cause C2cell 4: cause C2cell 5: cause C2cell 6: cause C2cell 7: cause C2cell 8: cause C2…

Max 32 cells

max E

veryS

AC

CH

HO CANDIDATE CELLS EVALUATION

Priority (0,n) = 0cell 2: cause C2cell 3: cause C2cell 4: cause C2Priority (0,n) = 1cell 1: cause C2Priority (0,n) = 2Priority (0,n) = 3cell 5: cause C2cell 6: cause C2cell 7: cause C2cell 8: cause C2Priority (0,n) = 4Priority (0,n) = 5

PRE-RANKING

PBGT_FILTERING

HO_MARGIN_XX(0,n)

Priority (0,n) = 0cell 2: cause C2cell 3: cause C2cell 4: cause C2

Priority (0,n) = 1Priority (0,n) = 2

Priority (0,n) = 3cell 6: cause C2cell 8: cause C2Priority (0,n) = 4Priority (0,n) = 5

CELLS EVALUATION PROCESSOrder or Grade

Grade

Priority (0,n) = 0cell 4 cause C2cell 2 cause C2Cell 3 cause C2

Priority (0,n) = 1

Priority (0,n) = 2

Priority (0,n) = 3cell 6: cause C2cell 8 cause C2Priority (0,n) = 4Priority (0,n) = 5

Order

Priority (0,n) = 0cell 4 cause C2cell 3 cause C2cell 2 cause C2

Priority (0,n) = 1

Priority (0,n) = 2

Priority (0,n) = 3cell 6: cause C2cell 8 cause C2Priority (0,n) = 4Priority (0,n) = 5

2.6 Handover Candidate Cell EvaluationEvaluation process

▼ The HO candidate evaluation process is run after all intercell handover alarms.▼ In case of intracell handover alarm (HO causes 10, 11, 13, 15, 16), the candidate cell evaluation process is skipped:

the target cell is the serving cell.▼ The handover detection gives as indication the raw cell list (built from book-keeping list) and the preferred layer for the

handover. In case of emergency handover alarms or cause 20 alarm, the cell evaluation will order the cells given in the raw list, putting in the first position the cells belonging to the preferred layer, having the highest priority (if EN_PRIORITY_ORDERING=ENABLE) and/or having the same frequency band type as the serving cell. In case of an intercell handover alarm, if the serving cell belongs to the raw cell list (emergency handover from the DCS 1800 inner zone of a multiband cell), this cell is put at the end of the candidate cell list with the MS zone indication OUTER.

▼ In case of better condition handover alarms (except cause 20), the cell evaluation will order the cells given in the raw list, putting in the first position the cells belonging to the preferred layer and having the highest priority (if EN_PRIORITY_ORDERING=ENABLE).


1.182

▼Pre-ranking in hierarchical or multi-band networks:

2.6 Handover Candidate Cell EvaluationPre-ranking

Cell_layer_type = Pref_layer

Cell_layer_type ≠ Pref_layer

List of candidate cells n

Cell_band_type = serving cell

Cell_band_type ≠ serving cell

Priority(0,n) = 0

Priority(0,n) = 1

Priority(0,n) = 5

Priority(0,n) = 0

Priority(0,n) = 1

Priority(0,n) = 5


1.183

▼ with priority(0,n) settingswith priority(0,n) settingswith priority(0,n) settingswith priority(0,n) settings, the operator can, for each couple of cells:�tag the target cell with a defined priority (from 0 = max to 5 = min)�this definition has an higher priority than usual order/grade ranking

▼ especially useful for multi band/hierarchical architecturesespecially useful for multi band/hierarchical architecturesespecially useful for multi band/hierarchical architecturesespecially useful for multi band/hierarchical architectures: �a simple way to force a target cell whatever its RxLev level and PBGT�nevertheless can be skipped over by filtering processes�low interest for standard networks

Serving cell

Candidate cell 1: RxLev: - 70 dBm, pbgt: + 10 dB

Candidate cell 2: Rxlev: - 90 dBm, PBGT: + 5dB

P0

P1

2.6 Handover Candidate Cell EvaluationPre-ranking

▼ Cell ordering according to target layer and target band▼ In hierarchical or multiband environment, cells are characterized by the layer they belong to or/and the frequency band

they use. The candidate cell evaluation process takes into account these characteristics in the candidate cell ordering.▼ In hierarchical environment, the HO detection process can indicate a preferred layer where the handover must be

directed to. If this indication is used, the candidate cell evaluation puts in the first places of the list, the candidate cellsbelonging to the preferred layer. They are followed by the cells of the other layer, providing they are also correct candidates.

▼ After this possible distinction, in each part of the list, the candidate cell evaluation sorts the candidate cells according to the parameter PRIORITY(0,n) (parameter on line changeable from the OMC-R).

▼ The cells having the highest priority are put in the first place of the list. They are followed by the cells having the lowest priorities. The PRIORITY(0,n) is only used when the flag EN_PRIORTY_ORDERING is set to “enable”.

▼ In case of emergency handover, for each category (preferred layer and other layer) and between cells having the same priority, the candidate cell evaluation sorts the candidate cells according to the frequency band they use: the cells which use the same frequency band as the serving cell are put first and they are followed by the cells which use the other frequency band.

▼ The cell evaluation function is then applied to the different candidate cell lists defined from the preferred layer indication, the PRIORITY(0,n) parameter and the frequency band of the serving cell (only in case of emergency handover).


1.184

▼ PBGT filtering:�optional, flag EN_PBGT_FILTERING

�filter out cells from the target list� inhibited for better cell handovers�based on power budget�per couple of cells

�was needed for multiband architecture

PBGT(n) >PBGT(n) >PBGT(n) >PBGT(n) > HO_MARGIN_XX (0,n)HO_MARGIN_XX (0,n)HO_MARGIN_XX (0,n)HO_MARGIN_XX (0,n) ++++ OFFSET_HO_MARGIN_INNEROFFSET_HO_MARGIN_INNEROFFSET_HO_MARGIN_INNEROFFSET_HO_MARGIN_INNER

�HO_MARGIN_XX (0,n) = HO_MARGIN_QUAL (0,n) for cause 2,4�HO_MARGIN_XX (0,n) = HO_MARGIN_LEV (0,n) for cause 3,5�HO_MARGIN_XX (0,n) = HO_MARGIN_DIST (0,n) for cause 6

�OFFSET_HO_MARGIN_INNER is only applied when the MS is in the inner zone of a concentric or multi band cell

�The averaging window is A_PBGT_HO

2.6 Handover Candidate Cell EvaluationPBGT Filtering

▼ The filtering process allows to filter out cells from the target list before sending them to the ORDER or GRADE evaluation process.

▼ It can be enabled/disabled on-line on a per cell basis from the OMC-R with the flag EN_PBGT_FILTERING.▼ The candidate cells are filtered on their power budget in relation to a handover margin threshold based on the

handover cause.

Note : the averaging window used for this process is A_PBGT_HO (even for emergency handovers, where a handover alarm could have been raised through A_LEV_HO or A_QUAL_HO samples)


1.185

▼ ORDER cell evaluation process

Cell "n" is ranked among other accordingly:

If EN_LOAD_ORDER = ENABLE and cell n is internal to the BSC

ORDER (n) = PBGT(n) + LINK_FACTOR(0,n) + FREEfactor(n) - FREEfactor(0)-HO_MARGIN_XX(0,n)

� Link_factor (0,n) is an operator parameter to give a bonus/penalty to a cellex: avoid external HO, decrease incoming flow of HO to a cell from another

� FREEfactor is TCH traffic based bonus/penalty to rank cells

If EN_LOAD_ORDER = DISABLE or cell n is external to the BSC

ORDER (n) = PBGT(n) + LINK_FACTOR(0,n) - HO_MARGIN_XX(0,n)

Cell "n" is kept if:

� AV_RXLEV_NCELL (n) > RXLEVmin (n) + max [0;( MS_TXPWR_MAX(n)-P)][dBm]

2.6 Handover Candidate Cell EvaluationORDER evaluation

▼ Two types of cell evaluation algorithms can be used: ORDER and GRADE.▼ ORDER and GRADE are two different methods of cell ranking. They both consist in giving a mark or ’figure of merit’ to

each candidate cell.▼ The basic differences between ORDER and GRADE are that:

� with ORDER� The candidate cell evaluation process interacts with the handover detection by use of cause-dependent

handover margins.� The candidate cell evaluation process takes into account the number of free TCHs in the candidate

cells.� with GRADE

� The candidate cell evaluation process does not interact with the handover detection.� The candidate cell evaluation process takes into account the relative load of traffic channels in the

candidate cells.▼ The type of cell evaluation is chosen by the operator on a (serving) cell basis and is provided to the BSC with the

parameter CELL_EV.

▼ For any handover cause, the first cell in the list is taken as a target cell, i.e. the cell with the highest value of ORDER(n). The cells do not need to fulfil any other condition.

▼ If no cell fulfils the condition and the serving cell does not belong to the target cell list, the target cell list is empty andno further action is carried out.

Note : the A_PBGT_HO averaging window is used for this process.


1.186

▼ GRADE cell evaluation process

Cell "n" is ranked among other accordingly:

If EN_LOAD_ORDER = ENABLE and cell n is internal to the BSC

GRADE (n) = PBGT(n) + LINK_FACTOR(0,n) + LOADfactor(n)

� Link_factor (0,n) is an operator parameter to give a bonus/penalty to a cell

� LOADfactor(n) is a weighting factor that takes into account the relative load of traffic channels in a cell

If EN_LOAD_ORDER = DISABLE or cell n is external to the BSC

GRADE (n) = PBGT(n) + LINK_FACTOR(0,n)

� Cell "n" is kept if:

� AV_RXLEV_NCELL (n) > RXLEVmin(n) + max [0;( MS_TXPWR_MAX(n)-P)]

2.6 Handover Candidate Cell EvaluationGRADE Evaluation

Annex 4

▼ LINKfactor(0,n) is a parameter set by OMC command for each cell(n).▼ LINKfactor(n1,n2) allows the operator to handicap or to favor the cell n1 with respect to its neighbor cell n2. In

particular, it can be used to disadvantage an external cell when an internal cell is also a possible candidate.

▼ For any handover cause, the first cell in the list is taken as a target cell, i.e. the cell with the highest value of GRADE(n). If no cell fulfils the condition and the serving cell does not belong to the target cell list, the target cell list isempty and no further action is carried out.

Note : the A_PBGT_HO averaging window is used for this process


1.187

2.6 Handover Candidate Cell EvaluationTraining exercise (1/2)

▼Emergency HO detected�With the “Candidate

evaluation.xls” excel sheet...� Filtering simulation for a

list of candidate cells� Ranking simulation for a

list of candidate cells

Book-keeping list1-

Time allowed:

15 minutes


1.188

2.6 Handover Candidate Cell EvaluationTraining exercise (2/2)

▼Emergency HO detected

Averaging measurement

PBGT Filtering

GRADE evaluation process

Target Cell

2-

3-

4-

5-?

?

?

?




2.7 Exercise





Handover Detection


Handover Management

Exercise


1.190

2.8 Exercise

▼List all the parameters involved in the detection of cause 23

▼List all the causes impacted by the parameter DELTA_INC_HO_MARGIN

▼List all the causes impacted by the parameter L_RXQUAL_UL_H

▼List all the causes impacted by the parameter BS_TXPWR_MAX

▼List all the causes impacted by the parameter BS_P_CON_ACK

Time allowed:

10 minutes



3 OTHER ALGORITHMS

B8


1.192

3 OTHER ALGORITHMSSession presentation

▼Objective: to be able to describe LCS, SDCCH Dynamic allocation, TCH resource allocation, MS reselection algorithms and list the associated parameters

▼Program:

3.1 LCS

3.2 Dynamic SDCCH allocation

3.3 TCH resource allocation algorithm

3.4 MS Reselection algorithms

B8



S3: OTHER ALGORITHMS


S5: CASE STUDIES



3 OTHER ALGORITHMS

3.1 LCS

B8

LCS

Dynamic SDDCH allocation

MS Reselection algorithms


1.194

3.1 LCS Definitions

▼ New end-user services which provide the geographical location of an MS:

� On MS request to know its own location � On network request (especially during Emergency calls)� On external request (LCS Client)

▼ Several positioning methods:

� Cell-ID or Cell-ID + TA (Timing Advance)� Conventional (standalone) GPS� Assisted GPS (with A-GPS server help to compute location)

� MS-based (MB): the MS is able to perform a pre-computation

� MS-assisted (MA): the MS sends info, Network computes

B8

▼ Assisted GPS Method:

� Mobile-based: The MS performs OTD signal measurements and computes its own location estimate. In this case, the network provides the MS with the additional information such as BTS coordinates and the RTD values. These assistance data can be either broadcast on the CBCH (using SMSCB function) or provided by the BSS in a point-to-point connection (either spontaneously or on request from the MS).

� Mobile-assisted: The MS performs and reports OTD signal measurements to the network and the network computes the MS’s location estimate.

� With� OTD: Observed Time Difference: the time interval that is observed by an MS between the receptions of

signals (bursts) from two different BTSs.� RTD: Real Time Difference: This means the relative synchronization difference in the network between

two BTSs.▼ Finally, 4 methods are possible for positioning:

� Cell ID+ TA,� This is the simplest method for determining the location of a mobile. It relies on the hypothesis that the

geographical coverage of a cell corresponds to that predicted by radio coverage studies. When an active mobile is connected to a base station, the mobile is assumed to be located geographically within the area predicted to be best served by this base station

� Conventional (MS equipped with GPS System),� MS-based Assisted GPS,� MS-Assisted GPS.


1.195

3.1 LCS LCS architecture

LCS function: Architecture

B8

SMLCBTS

BTS

MS

BSC

MSC

HLR

GMLC

OSP

Lg

Lh

ExternalLCS client

LeA

Abis

Abis

Lb

SMLC function integrated in MFS:

- receives the loc. Request from the GMLC through the MSC/BSC

- Schedules all the necessary actions to get MS location

- Computes MS location

- Provides the result back to the GMLC

MFS

A-GPS server

SAGI

GPS reference network

LCS: Location ServicesSMLC: Serving Mobile Location Center GMLC: Gateway Mobile Location CenterA-GPS: Assisted GPS

Where is my son?

Where am I?

Where is the accident?Emergency call

MS Request

Network Request

External Request

2

1

3

3

2

1


1.196

3.1 LCS LCS Positionning procedureB8

GMLC

LCS Client

1. Location request

2. Routing information

8. Locationresponse

4. P

agin

g, a

uth

e nti

c ati

on

,ci

ph

e rin

g, n

oti

fica

tio

n

3. P

rovi

de s

ubsc

riber

loca

tion

6. In

divi

dual

posi

tioni

ng

5. P

rovi

de

sub

scri

ber

loca

tio

n7. Location report

BSC MSC GMLC

HLR

SMLC

MFS

OSP

▼ If the MS is in idle mode, the MSC first performs a CS paging, authentication and ciphering in order to establish an SDCCH with the MS. The MS subscriber is not aware of it, i.e. no ringing tone, except towards GPRS MS in Packet Transfer Mode which may suspend its GPRS traffic in order to answer to the CS Paging (i.e. not fully transparent for the subscriber).

▼

▼ When the MS is in dedicated mode (after a specific SDCCH establishment for location, or during an on-going call), the MSC sends the location request to BSC in the existing SCCP connection for the current call, which forwards it to the SMLC


1.197

3.1 LCS LCS protocol (1/2)B8

Signaling Protocols between the SMLC and the MS (CS domain)


1.198

3.1 LCS LCS protocol (2/2)

Example: Mobile terminated location request success (External request)

B8

SMLCMS BSCBTS LCS ClientMSC

BSSAP-LE Perform_Location_Request

.

GMLC

BSSMAP Perform_Location_Request

BSSAP-LE Perform_Location_Response

BSSMAP Perform_Location_Response

BSSMAP Clear Command and Release

Adequate positioning method chosen by SMLCwith optional additional scenario

HLR

Paging

Authentication + Ciphering

LCS Service Response

LCS Service Request

Send_Routing_Info rqst

Send_Routing_Info resp

Provide_Subscriber_Location

Provide_Subscriber_Location Result

Starts T_location

Stop T_location

▼ T_location_Longer used in case of optional additional scenario (see graph):Upon receipt of the MS POSITION COMMAND message from the SMLC (optional additional scenario), the BSC stops

the T_Location timer, and starts instead the T_Location_Longer timer. This timer is stopped only at the end of the location procedure in the BSC, i.e. when an 08.08 PERFORM LOCATION RESPONSE message is sent back to the MSC.

▼ Aborts:� Abort by MSC

Depending on the location procedure and its current state of execution, upon PERFORM LOCATION ABORT message receipt, the BSC sends immediately to the MSC a PERFORM LOCATION RESPONSE message (when no exchange on the Lb interface is on-going), or to the SMLC either a PERFORM LOCATION ABORT or an ABORT message. The BSC starts the timer T_Loc_abort to supervise the SMLC response.

� Abort by BSSIf an ongoing location request is interrupted at the BSC level for the following reasons:

� by an inter-BSC handover, or

� if the main signaling link to the target MS is lost or released, or� the SCCP connection on the A interface is released, or� if the timer T_Location expires,

the BSC must send either a PERFORM LOCATION ABORT message or a ABORT message to the SMLC and starts the timer T_Loc_abort

▼ The useful B8 content of the received PERFORM LOCATION REQUEST message is:� Location type,� Classmark information 3, � Requested QoS: provides service requirement concerning geographic positioning and response time

� accuracy, the response time category (Low Delay or Delay Tolerant),� Current Cell Id + TA information are always provided to the SMLC.

▼ The time of transfer of the assitance data on the SDCCH is estimated about 14s for a 1000 octets information,


1.199

3.1 LCS Positioning methods : CI+TA positioning

▼ Principles of CI + TA Positioning Method

B8

TA

553m

TA553m

Site Geographical

Information

(Lat/Long)

Main Beam

Directio

ngiven by

the azimuth

3dB points given

by the azimuth

and the HPBW

MS EstimatedLocation

Annex 5

▼ With the TA positioning method, no signalling exchange is required between the SMLC and the MS (i.e. RRLP protocol is not required). The TA positioning method is applicable to all the MSs (supporting LCS or not).

▼ Based on:� Cell Identity (CI) of the serving cell and � Timing Advance (TA) value reported by MS➨ intersection point of a line from the BTS antenna in their main direction with a circle which radius is

corresponding with the propagation delay (timing advance) is the MS estimated position➨ Omni-directional cells: MS position = site position

▼ Parameters:▼ EN_LCS – flag to enable/disable the Location Services per BSS

� 0 = Enabled; 1= Disabled; Default = 0➨➨➨➨ IF EN_LCS=1, CI+TA method is enabled in all the BSS cells� LCS_LATITUDE

� Latitude of the BTS supporting the cell� LCS_LONGITUDE

� Longitude of the BTS supporting the cell� LCS_AZIMUTH

� Antenna direction orientation for the sector supporting the cell� HALFPWR_BEAM_WIDTH

� Antenna half power beamwidth for the sector supporting the cell

▼ Optimization parameters:� ARC_SIZE_FACTOR

� Factor used in the computation of the width in degree of the ellipsoid arc returned by the MFS when computing location estimate based on TA positioning method.

� MIN_RADIUS_FACTOR� Factor used in the computation of the minimum radius of the ellipsoid arc returned by the MFS when

computing location estimate based on TA positioning method

� MAX_RADIUS_FACTOR

� Factor used in the computation of the maximum radius of the ellipsoid arc returned by the MFS when computing location estimate based on TA positioning method


1.200

3.1 LCS Positioning methods : Conventional GPS

▼Conventional GPS location procedure

� This optional location procedure is chosen by the SMLC (if the MS support it) upon reception of a Perform Location Request message from the BSC

MS BTS BSC SMLCPerform

Location Request

Measurement Position Request

Measurement Position Response (X,Y)

(X,Y)

PerformLocation Response (X,Y)(X,Y) : Computed position

(X,Y)

Location Request

Location Response

B8

� The MS continiously computes its position� Terminal searches for satellites, acquires all the GPS data, computes its own position and finally provides the

location estimation to the SMLC

Location Request

Location Response


1.201

3.1 LCS Positioning method : Assisted GPS Positioning 1/3

▼ Assisted GPS Positioning Method (A-GPS)

� Assistance GPS Positioning Method is split into:� MS Based A-GPS method� MS Assisted A-GPS method

B8

A-GPS Server

GPS referencenetwork

Assistance data on requestAssistance data on request

GPS acquisition assistanceNavigation model (almanac, ephemeris)Ionospheric modelTime integrity

GPS acquisition assistanceNavigation model (almanac, ephemeris)Ionospheric modelTime integrity

GPS MS

▼ Assistance data gathered from a GPS reference network receiver is broadcasted to the GPS MS

▼ Flags/Parameters� EN_LCS = 1� EN_MS_BASED_AGPS – enables/disables the positioning method MS Based A-GPS per CELL

� 0 = disabled; 1 = enabled; default = 0� EN_MS_ASSISTED_AGPS – enables/disables the positioning method MS Assisted A-GPS per CELL

� 0 = disabled; 1 = enabled; default = 0


1.202


▼A-GPS location procedure / MS Based A-GPS

BTS BSC SMLCPerform

Location Request

Assistance Data

Assistance Data Ack

GPS info RequestGPS info Response


Measurement Position Response (X,Y)

A-GPSServer

Perform(X,Y) Location Response

MS

(X,Y) : Computed position

(X,Y)

POSITIONING CALCULATIONlatitude, longitude (and

altitude)

POSITIONING CALCULATIONlatitude, longitude (and

altitude)

Location Request

Assistance Data

Position Request

Position Response

Location Response

B8

▼ Using assistance data, MS computes by itself the position and sends it back to the SMLC


1.203


▼A-GPS location procedure / MS Assisted A-GPS

BTS BSC SMLCPerform

Location Request

Assistance Data

Assistance Data Ack

GPS info RequestGPS info Response

Location Request

Assistance Data

Position Request

Position Response

Location Response


Measurement Position Response (M) GPS LocationRequest (M)

GPS LocationResponse (X,Y)

A-GPSServer

Perform(X,Y) Location Response

(X,Y) : Computed position

MS

(X,Y)

Pseudo-range MEASUREMENTS

Pseudo-range MEASUREMENTS

B8

� Using a reduced set of assistance data, the MS makes pseudo–range measurements and sends the result to the A-GPS server, which fixes the position in the end


1.204

3.1 LCS LCS impact on HO 1/3B8

▼HO preparation

� Inhibition of “better cell handovers”�Other HO


BSSAP-LE Perform_Location_Request

.

GMLC


HLR

Paging

Authentication + Ciphering

LCS Service Request

Send_Routing_Info rqst

Send_Routing_Info resp

Provide_Subscriber_Location

Starts T_location

EMERGENCY HO

DETECTION

BSSLAP - Reset

� HO needed during LCS procedure


1.205


▼HO management

� Internal HO


BSSLAP-Reset

.

GMLC


HLR

HO COMPLETE BSSMAP PERFORM LOCATION RESPONSE (Cause = “Intra-BSC Handover Complete”)


INTRA BSC HO

On going

� Mobile in communication


1.206


▼HO management

� external HO

SMLCMS SERVING BSCBTS LCS ClientMSC GMLC HLR


external BSC HO BSSMAP HO REQUIRED

BSSAP-LE Perform_Location_Abortt



1.207

3.1 LCS BSS ParametersB8

▼ FLAGS ▼ Timers

�EN_LCS

�EN_SAGI

�T_Location

�T_Location_longer

�T_Loc_Abort

�T_LCS_delay_tolerant

�T_LCS_LowDelay

�T_RRLP_low_delay

�T_RRLP_delay_tolerant

�ARC_SIZE_FACTOR

�MIN_RADIUS_FACTOR

�MAX_RADIUS_FACTOR

▼ OPTIMIZATION DATA

▼ BSS PARAMETERS� EN_LCS (BSC)

� Flag which enables or disables the LCS feature in the BSS.� EN_SAGI

� Flag indicating whether SAGI is configured or not for this BSS� T_Location:

� BSC timer on a per call basis to guard the response from the SMLC in case of Location Request, when no RRLP exchange is triggered with the MS.

� T_Location_longer:� BSC timer on a per call basis to guard the response from the SMLC in case of Location Request, when an

RRLP exchange is triggered with the MS. Replace T_Location timer in case of Conventional GPS, MS-Assisted A-GPS, MS-Based A-GPS.

� T_Loc_Abort� BSC timer to guard the response from the SMLC in case of Location Abort.

� T_LCS_LowDelay� SMLC timer to guard the calculation of the MS position (including the RRLP message exchange

with the target MS) in case of a Low Delay Location Request.� T_LCS_DelayTolerant

� SMLC timer to guard the calculation of the MS position (including the RRLP message exchange with the target MS) in case of a Delay Tolerant Location Request.

� T_LCS_LowDelay� SMLC timer to guard the calculation of the MS position (including the RRLP message exchange

with the target MS) in case of a Low Delay Location Request. � T_RRLP_Low_delay

� Timer to guard the RRLP exchange between the SMLC and the MS . � T_RRLP_delay_tolerant

� Timer to guard the RRLP exchange between the SMLC and the MS.

� Optimization data:� ARC_SIZE_FACTOR



computing location estimate based on TA positioning method � MAX_RADIUS_FACTOR



1.208

3.1 LCS Cell ParametersB8

▼ FLAGS ▼ SITE DATA

�EN_CONV_GPS

�EN_MS_ASSISTED_AGPS

�EN_MS_BASED_AGPS

�LCS_LATITUDE

�LCS_LONGITUDE

�LCS_SIGNIFICANT_GC

�LCS_AZIMUTH

�HALF_POWER_BANDWIDTH

▼ CELL PARAMETERS� EN_CONV_GPS

� Flag to enable/disable the Conventional GPS positioning method.� EN_MS_ASSISTED_AGPS

� Flag to enable/disable the MS Assisted A-GPS positioning method.� EN_MS_BASED_AGPS

� Flag to enable/disable the MS Based A-GPS positioning method. � LCS_LATITUDE

� Latitude of the BTS supporting the cell (used by the MFS to compute location estimate based on TA positioning method).

� LCS_LONGITUDE � Longitude of the BTS supporting the cell (used by the MFS to compute location estimate based on TA

positioning method). � LCS_SIGNIFICANT_GC

� Indicates whether latitude and longitude are significant or not � LCS_AZIMUTH

� Antenna direction orientation for the sector supporting the cell (used by the MFS to compute location estimate based on TA positioning method).

� HALF_POWER_BANDWIDTH � Half power beam width of the antenna for the sector supporting the cell (used by the MFS to compute

location estimate based on TA positioning method).

� Remark: To have LCS supported for a cell, the operator must activate LCS on the BSS handling this cell but he must also activate GPRS for this cell (i.e. setting of MAX_PDCH to a value > 0, the cell being kept locked for GPRS if the operator does not want to have GPRS running on this cell) and configure all the required transmission resources (Ater and Gb resources) on the GPU(s) connected to this BSC


1.209

3.1 LCS Exercise

▼Where is implemented the SMLC function?

▼What are the LCS impacts on cell dimensioning?Time allowed:

10 minutes

B8



3 OTHER ALGORITHMS

3.2 Dynamic SDCCH allocation

B8

LCS




1.211

3.2 Dynamic SDCCH allocationPurpose

▼SDCCH/8 time slots can be dynamically allocated on demand on a cell-by-cell basis.

� “Dynamic SDCCH/8 time slots”. � “Static SDCCH time slots”

StaticSDCCH

time slotsTCH

Capacity

Allocated DynamicSDCCH/8time slots

Max

Min

B8

▼ Definitions

A Static SDCCH timeslot is a physical timeslot fixed allocated on the air interface. It contains 3, 4, 7 or 8 SDCCH sub-channels depending on whether the timeslot is an SDCCH/3, SDCCH/4, SDCCH/7, or SDCCH/8 timeslot.


1.212

3.2 Dynamic SDCCH allocationPrinciple (1/2)

▼Principles

� Too few SDCCH time slots could result in high blocking rateon SDCCH (Configuration 1)

� Too many SDCCH time slots could lead to a lack of TCH resources (Configuration 2)

SDCCHtime slots

TCHCapacity

SDCCHtime slots

TCHCapacity

Configuration 1 Configuration 2

B8

▼ Definition

An SDCCH is a logical SDCCH sub-channel mapped on a Static SDCCH timeslot or a Dynamic SDCCH/8 timeslot.

▼ Signaling load cases

Timeslot split between signaling and traffic channels depends on the network signaling load. The main cases are:

- Normal signaling load cells:Rural area cells in center of Location Areas(e.g. 1 SDCCH timeslot for a 3-TRX cell)

- High signaling load cells:Urban or suburban area cells in the center of a Location AreaRural area cells at the border of Location Areas(e.g. 2 SDCCH time slots for a 3-TRX cell)

- Very high signaling load cells:Urban or suburban area cells at the border of a Location AreaCells with high SMS load (more than one SMS per call)(e.g. 3 SDCCH time slots for a 3-TRX cell)


1.213

3.2 Dynamic SDCCH allocationPrinciple (2/2)

▼Allocation and de-allocation of Dynamic SDCCH/8 time slots

• An additional dynamic SDCCH/8 timeslot is allocated by the BSC if there is no SDCCH sub-channel free in the cell.

• A dynamic SDCCH/8 timeslot is de-allocated by the BSC after T_DYN_SDCCH_HOLD (10s) delay if all of its SDCCH sub-channels become free

+

-

BCC SDC

TCH TCH

TCH TCH

BCC: BCCHSDC: Static SDCCHSDD: Dynamic SDCCH

TCHTCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

TCH

SDC

SDD

SDC

SDD

SDD

B8

▼ The location of the Dynamic SDCCH/8 time slots are fixed by O&M configuration.

▼


1.214

3.2 Dynamic SDCCH allocationTIMESLOT types

▼NEW TIMESLOT TYPES :

� SDCCH Pure SDCCH or “ static SDCCH “

� TCH Pure TCH

� TCH/SDCCH “ dynamic SDCCH”

� TCH/SPDCH

� MPDCH

B8

▼The OMC-R provides the BSC with the following O&M type of radio timeslots:

� Main BCCH timeslot (BCC): It is a timeslot carrying FCCH + SCH + BCCH + CCCH.� Main combined BCCH timeslot (CBC): It is a timeslot carrying FCCH + SCH + BCCH + CCCH +

SDCCH/4 + SACCH/4.� Static SDCCH timeslot (SDC): It is a timeslot carrying SDCCH/8 + SACCH/8.� Dynamic SDCCH/8 timeslot (SDD): It is a timeslot carrying TCH + SACCH or SDCCH/8 + SACCH/8� TCH timeslot (TCH): It is a timeslot carrying TCH + SACCH or PDCH

▼In RAM point of view, a radio timeslot can be defined as:� Pure BCCH timeslot: The BCCH timeslot is the radio timeslot configured as BCC by O&M. Such a

timeslot only carries common CS signalling.� Pure SDCCH timeslot: A pure SDCCH timeslot is a timeslot configured as a CBC or SDC by O&M.

Such a timeslot can carry SDCCH traffic.� Pure TCH timeslot: A pure TCH timeslot is a timeslot configured as TCH by O&M. Such a timeslot only

carries TCH traffic.� TCH/SDCCH timeslot: A TCH/SDCCH timeslot is a timeslot configured as SDD by O&M. Such a

timeslot is dynamically allocated as TCH or as SDCCH depending on the usage of the timeslot. It can carry TCH traffic or SDCCH traffic.

� TCH/SPDCH timeslot: A TCH/SPDCH timeslot is a timeslot configured as TCH by O&M. Such a timeslot is dynamically allocated as TCH or as SPDCH depending on the usage of the timeslot. It can carry TCH traffic or PS traffic.

� MPDCH timeslot: A MPDCH timeslot is a timeslot configured as TCH by O&M. Such a timeslot can only carry common PS signalling.

▼A pure SDCCH timeslot can carry x SDCCH sub-channels where x equal to:− 4 in case of combined CCCH and when CBCH is not configured on the timeslot,− 7 in case of non-combined CCCH and when CBCH is configured on the timeslot,− 3 in case of combined CCCH and when CBCH is configured on the timeslot,− 8 for a normal SDCCH timeslot.▼When allocated as SDCCH, a TCH/SDCCH timeslot can carry up to 8 SDCCH sub-channels.


1.215

3.2 Dynamic SDCCH allocationAllocation algorithmB8

?

Are they any free SDCCH

sub-channel among Static

SDCCH timeslots ?

Selection of one

SDCCH sub-channel

Yes

No

?

Are they any free SDCCH

sub-channel among Dynamic

SDCCH/8 already allocated ?

Select one SDCCH

sub-channel

No

?

Are they any Dynamic

SDCCH/8 timeslots available

and free in the cell ?

SDCCH request

rejected !!!

NoYes

Yes

Allocate

one dynamic

SDCCH/8

timeslot

SDCCH mapped on” TCU

very high load state” removalSDCCH REQUEST

Principle 1 : Preference is given to pure SDCCH timeslotsPrinciple 2 : Balance TCU processor load between different TCUs

in fact before entering in this algorithm ( see slide) the first step is :Removal of all the SDCCH subchannels mapped on TCU in « Very High Overload » statePrinciple 3 : FR TRX preference


1.216

3.2 Dynamic SDCCH allocationSDCCH sub-channel selectionB8

TS with LOWEST TCU

LOAD

TS with lowest index on TRX with lowest

TRX_ID

TS with MAXIMUM FREE

SDCCH Sub channels

TS on FR TRX

Pure SDCCH

Timeslots

TCH/SDCCH TS allocated as

SDCCH

TCH/SDCCH

allocated as TCH

Note that a SDCCH request can not access the timeslots reserved by NUM_TCH_EGNCY_HO. If all remaining TCH/SDCCH timeslots are reserved by NUM_TCH_EGNCY_HO, then the SDCCH request shall be rejected.


1.217

3.2 Dynamic SDCCH allocationDe allocation algorithmB8

Yes

No

No

Yes

TCH/SDCCH

deallocatedT_DYN_SDCCH_HOLD

stopped

?

T_DYN_SDCCH_HOLD

running ?

No

T_DYN_SDCCH_HOLD

started

all SDCCH

sub-channels

of a TCH/SDCCH

timeslot

in the cell are free ?

Yes

T_DYN_SDCCH_HOLD

Expires ?

??No

Yes

At least one TCH/SDCCH allocated

The de-allocation algorithm ensures that :· TCH/SDCCH timeslots are not allocated too fast to TCH after de-allocating them

TCH/SDCCH timeslots are not re-allocated too frequently to SDCCH

Note : · while T_DYN_SDCCH_HOLD is running: ¨ the dynamic SDCCH/8 timeslot marked as “HOLD” is still considered as allocated to SDCCH (and can not be

allocated to TCH);¨ If a subsequent dynamic SDCCH/8 timeslot (used as SDCCH and in the same cell) becomes free: a) If this just freed dynamic SDCCH/8 timeslot has a higher priority, T_DYN_SDCCH_HOLD is re-started and

precedent dynamic SDCCH/8 timeslot in “HOLD” state is de-allocated immediately;b) If this just freed dynamic SDCCH/8 timeslot has lower priority, and T_DYN_SDCCH_HOLD is re-started and

the just freed dynamic SDCCH/8 timeslot is de-allocated immediately.


1.218

3.2 Dynamic SDCCH allocationO&M configuration 1/2

▼Massive modification by script

� 10 templates �Template

customization �Template

launched through PRC

B8

12

3 10

14 7 8

692

5

11

▼ selection of static or dynamic SDCCH

�Timeslot configuration menu

▼Dynamic sdcch rules▼The CBCH must be configured on a static SDCCH/8 or SDCCH/4 timeslot.▼Combined SDCCHs (SDCCH/4 + BCCH) are always static.▼To avoid incoherent allocation strategy between SDCCH and PDCH, a dynamic SDCCH/8 timeslot cannot have the characteristic of being a PDCH (it cannot carry GPRS traffic).▼The operator must configure at least one static SDCCH/8 or SDCCH/4 timeslot on BCCH TRX in a cell.▼In cells with E-GSM, only the TRX, which do not belong to the G1 band, can support dynamic and staticSDCCHs.▼In multiband and concentric cells, only the TRX, which belongs to the outer zone, can support dynamic and staticSDCCHs. ▼Up to 24 static/dynamic SDCCH sub-channels can be configured per TRX.


1.219

3.2 Dynamic SDCCH allocationO&M configuration 2/2

▼ Default configuration for a cell which has only Full rate TRX

B8

Note1: For one TRX, dynamic SDCCHs are over-dimensioned because of the granularity of 8. According to the Alcatel traffic model, all dynamic SDCCHs will not be used.

Note2: An additional dynamic SDCCH/8 must be provided for each DR TRX (these are expected mainly on small cells).

▼ rules:� At least one static SDCCH/4 or SDCCH/8 on BCCH TRX

� Up to 24 static/dynamic SDCCH sub-channels per TRX� Up to 32 static/dynamic SDCCH sub-channels per TCU� Up to 88 static/dynamic SDCCH sub-channels per CELL



3 OTHER ALGORITHMS

3.3 TCH resource allocation algorithm

LCS




1.221

3.3 TCH resource allocation algorithmRadio Allocation and Management

▼ Radio resource allocation and management (RAM) aims at:

� Managing pools of TCH radio resources by:

� defining TCH radio timeslots as a function of the cell radio configuration from the operator

� sorting these TCH TS according to their radio capabilities (FR or DR, frequency band (G1 or GSM/DCS))

� Allocating dedicated TCH radio resources by:

� selecting the TCH pool in which the TCH should be chosen according to:

� the requested channel rate (FR or HR)

� the radio capability of the mobile

� the TRE DR capability and the TRE band

� selecting the best TCH resource among the available TCH channels of this pool according to several criteria

B8


1.222

3.3 TCH resource allocation algorithmRadio Timeslot of a cell : Operator view

▼On the OMC-R the operator can configure the following Radio TS per cell:

� Main BCCH timeslot (BCC): TS carrying FCCH + SCH + BCCH + CCCH

� Main combined BCCH timeslot (CBC): TS carrying FCCH + SCH + BCCH + CCCH + SDCCH/4 + SACCH/4

� Static SDCCH timeslot (SDC): TS carrying SDCCH/8 + SACCH/8

� Dynamic SDCCH/8 timeslot (SDD): TS carrying TCH + SACCH or SDCCH/8 + SACCH/8

� TCH timeslot (TCH): TS carrying TCH + SACCH or used as a PS timeslot (PDCH)

B8

▼ The operator has to choose between a Combined BCCH (CBC TS) or a Non-combined BCCH configuration (BCC TS).

▼ A PDCH is a radio timeslot used for PS traffic or signalling.▼ It can carry either PS traffic or PS signalling but not both.

� If it carries traffic it is called a Slave PDCH (SPDCH) TS and it carries the logical channels PDTCH+PACCH+PTTCH.

� If it carries signalling it is called a Master PDCH (MPDCH) TS and it carries:� either the logical channels PBCCH+PPCH+PAGCH+PRACH: it is then called a Primary MPDCH� or only PPCH+PAGCH+PRACH: it is then called a Secondary MPDCH

▼ SDD TS can carry either TCH or SDCCH channels but not both at the same time.▼ TCH TS can carry either CS traffic channel TCH or PS logical channels but not both at the same time.


1.223

3.3 TCH resource allocation algorithmRadio Timeslot of a cell : RAM view

▼ In the BSS the RAM software module maps the OMC-R cell radio configuration to its own types of TS :

� Pure BCCH timeslot: BCC TS carrying only common CS signalling (BCCH+CCCH)

� Pure SDCCH timeslot: CBC or SDC TS carrying only dedicated CS signalling (SDCCH)

� Pure TCH timeslot: TCH TS carrying only TCH traffic� TCH/SDCCH timeslot: SDD TS carrying either CS traffic

(TCH) or dedicated CS signalling (SDCCH) � TCH/SPDCH timeslot: TCH TS carrying either CS traffic

(TCH) or PS traffic (SPDCH channels) � MPDCH timeslot: TCH TS carrying common PS signalling

(PBCCH+PCCCH or PCCCH only)

B8

▼ TCH/SDCCH timeslots are allocated as TCH or SDCCH according to an SDCCH dynamic allocation algorithm presented in the “Introduction to Radio Fine Tuning B8” training course.

▼ TCH/SPDCH timeslots are allocated as TCH or SPDCH according to a SPDCH dynamic allocation algorithm presented in the “Introduction to GPRS & E-GPRS Quality of Service Monitoring B8” training course.


1.224

3.3 TCH resource allocation algorithmRadio Timeslot : OMC-R / RAM mapping

▼NB_TS_MPCH MPDCH TS are defined on the BCCH TRX :

� on the timeslots configured as TCH TS on the OMC-R

� having the lowest timeslot index

▼TCH/SPDCH TS are defined as being part of an SPDCH group

▼Pure TCH timeslots are OMC-R TCH TS neither defined as MPDCH TS nor in an SPDCH group

OMC-R radio TS RAM radio TS

BCC

CBC

SDC

SDD

TCH

Pure BCCH

Pure SDCCH

TCH/SDCCH

TCH/SPDCH

MPDCH

Pure TCH

B8

▼ MPDCH TS are defined on the BCCH TRX even if the corresponding TRX_PREF_MARK is different than 0.


1.225

3.3 TCH resource allocation algorithmDefinition of a TCH/SPDCH TS

▼For PS traffic resource allocation, an SPDCH group is defined ona per TRX basis and is made of consecutive timeslots:

� mapped on OMC-R TCH TS

� located on a PS capable TRX (TRX_PREF_MARK = 0)

� not defined as MPDCH TS

� having the same radio configuration (MA, MAIO)

▼ If several SPDCH groups can be defined on a given TRX, the BSS chooses the SPDCH group of timeslots having the highest number of consecutive timeslots.

▼A radio timeslot belonging to one of the different SPDCH groups of the cell is identified in RAM as a TCH/SPDCH timeslot.

B8

▼ The timeslots shall be consecutive on a given TRX means that there shall be no hole in the SPDCH group.▼ If several SPDCH groups can be defined on the same TRX and having the same number of consecutive timeslots then

the group that is located on the left side of the TRX (i.e. the timeslots having the lowest index) shall be chosen.


1.226

TRX1

TRX2 SDC

TRX3

TRX4

3.3 TCH resource allocation algorithmExercise 1

▼A non hopping cell is configured on the OMC-R

TRX30 TCH TCH TCH TCH TCH TCHTCHTCH

TRX_PREF_MARK 0 1 2 3 4 5 6 7TRX1 BCC SDC0 TCH TCH TCH TCH TCH TCH

TRX2 SDC0 SDCSDD TCH TCH TCH TCH TCH TCH

TRX41 TCH TCH TCH TCH TCH TCHTCH TCH

▼ Find the radio TS configuration in RAM if NB_TS_MPDCH= 2

PBC: Pure BCCH

PSD: Pure SDCCH

TSD: TCH/SDCCH

TSP: TCH/SPDCH

MPD: MPDCH

PTC: Pure TCH

B8

▼ The timeslots shall be consecutive on a given TRX means that there shall be no hole in the SPDCH group.▼ If several SPDCH groups can be defined on the same TRX and having the same number of consecutive timeslots then

the group that is located on the left side of the TRX (i.e. the timeslots having the lowest index) shall be chosen.


1.227

3.3 TCH resource allocation algorithmTCH pools

▼ 3 pools of TCH resources are managed per cell:

� G1 pure TCH pool: contains all the free TCH sub-channels (FR or HR) free on the pure TCH TS of the G1 TRXs

� GSM/DCS pure TCH - TCH/SPDCH pool: contains all the free TCH sub-channels (FR or HR) free on the pure TCH TS and on the TCH/SPDCH TS of the GSM/DCS TRXs

� GSM/DCS TCH/SDCCH pool: contains all the free TCH sub-channels (FR or HR) free on the TCH/SDCCH TS of the GSM/DCS TRXs

▼ any pure TCH, TCH/SPDCH, TCH/SDCCH TS can be:

� Busy: if it is not free to serve a FR TCH request

� Free: if it is free to serve a FR TCH request

B8

▼ A DR TS (timeslot on a DR TRX) is free if no FR TCH or HR TCH is allocated for a call on this timeslot.

▼ A DR TS is busy if at least one TCH is allocated for a call on this timeslot:

� 1 FR TCH

� or 1 HR TCH (HR 0 TCH or HR 1 TCH)

� or 2 HR TCHs (HR 0 TCH and HR 1 TCH)


1.228

3.3 TCH resource allocation algorithmTCH sub-pools

▼FR TCH channels can be allocated on both FR and DR TRXswhereas HR TCH channels can only be allocated on DR TRXs

▼Each of the three TCH pools is divided in three sub-pools:

� FR sub-pool: contains all the free FR TCH sub-channels available on the FR TRX

� DR: sub-pool: contains all the free FR TCH sub-channels available on the DR TRX

� HR sub-pool: contains all the free HR TCH sub-channels whose mate HR TCH sub-channel is busy(always located on the DR TRX)

B8

▼ Inputs for TCH allocation function:


1.229

3.3 TCH resource allocation algorithmTCH allocation process 1/2

TCH request

TCH rejected TCH queued

TCH allocation

Select a TCH sub-pool

radio capability of the mobilechannel type (FR,HR,DR)speech version (FR, HR, EFR, AMR FR, AMR HR)request type (NA or HO)

cell channel type capabilitycell codec type capability

cell load

TCH free?

Queuing?

no yes

no yes

Select a TCH in this sub-pool

TCH selected

B8

Inputs for TCH allocation function:

▼ radio capability of the MS:

� the BSS knows the radio capability of the mobile from the MS CLASSMARK after the Radio Link Establishment procedure

▼ requirements from the MSC:

channel type (mandatory) is one of the following:

list of preferred speech version (optional):

� GSM full rate speech version 1 = FR

� GSM full rate speech version 2 = EFR

� GSM full rate speech version 3 = AMR FR

� GSM half rate speech version 1 = HR

� GSM half rate speech version 3 = AMR HR

▼ capabilities of the cell:

� FR TCHs only if only FR TRXs / FR+HR TCHs if some DR TRXs

� codec supported among: FR, EFR, AMR FR, HR, AMR HR

FR Full Rate only

HR Half Rate only

DR FR P NCA Dual Rate Full Rate Preferred No Changes Allowed after first channel allocationas a result of the request

DR FR P CA Dual Rate Full Rate Preferred Changes Allowed after first channel allocation as aresult of the request

DR HR P NCA Dual Rate Half Rate Preferred No Changes Allowed after first channel allocationas a result of the request

DR HR P CA Dual Rate Half Rate Preferred Changes Allowed after first channel allocation as aresult of the request

DR SV P NCA Dual Rate No Changes of channel rate Allowed after first channel allocation as aresult of the request

DR SV P CA Dual Rate Changes of channel rate Allowed after first channel allocation as aresult of the request


1.230

3.3 TCH resource allocation algorithmTCH allocation process 2/2

TCH rejected TCH queued

TCH allocation

Select a TCH sub-pool

TCH free?

Queuing?

no yes

no yes

Select a TCH in this sub-pool

TCH selected

B8

NUM_TCH_EGNCY_HO

T_QHO

ALLOC_ANYWAY

T11

T11_ FORCED

- The timer T11 corresponds to normal assignment with queuing authorised.- The timer T11_FORCED corresponds to normal assignment

i) when the queuing is not authorised by the MSC but forced by the BSC (QUEUE_ANYWAY = TRUE),

or ii) when the queuing is not authorised but the request has its pre-emption indicator set and has already forced the release of a lower priority pre-emptable on-going call.

The QUEUE_ANYWAY flag is checked by the Normal Assignment (NASS) entity.- The timer T_qho corresponds to an external channel change with queuing authorised or to an

external channel change when the queuing is not authorised but the request has its pre-emption indicator set and has already forced the release of a lower priority pre-emptable on-going call.


1.231

3.3 TCH resource allocation algorithmTCH sub-pool selection

▼The BSS selects the TCH sub-pools in which a TCH channel can be allocated according to:

� the requested channel rate and the cell load situation

� favour HR if cell is loaded

� a priority given to generic resources

1. G1 pool (E-GSM mobile only)2. GSM/DCS pure TCH - TCH/SPDCH pool3. GSM/DCS TCH/SDCCH pool

� an optimisation of FR/HR resources

� favour FR pool over DR pool for a FR TCH request� favour HR pool over DR pool for an HR TCH request

� the availability of a TCH channel in the sub-pool

B8

▼ TCH allocation without list of preferred speech versions� FR request: FR pool � DR pool� HR request: HR pool � DR pool� DR FR Preferred request:

� cell load=False: FR pool � DR pool � HR pool � cell load=True: HR pool � DR pool � FR pool

� DR HR Pref. request: HR pool � DR pool � FR pool

▼ TCH allocation with a list of preferred speech versions� FR SV then HR SV: FR pool � DR pool � HR� HR SV then FR SV: HR pool � DR pool � FR� FR SV only: FR pool � DR pool� HR SV only: HR pool � DR pool

▼ favour G1 pool for an E-GSM mobile (penetration of E-GSM mobile is low and GSM/DCS traffic is high)▼ disfavour TCH allocation on TCH/SDCCH TS (favour signalling over traffic)

▼ Example : E-GSM mobile / DR FR P NCA / no sub-pool is empty� BSS selects a TCH in the G1 pure TCH pool / FR sub-pool


1.232

3.3 TCH resource allocation algorithmTCH selection

▼ sub-pool of the GSM/DCS pure TCH - TCH/SPDCH pool� optimise CS/PS traffic resources

1. Favour TCH allocation on pure TCH TS2. Optimise PS traffic on TCH/SPDCH TS

� TCH allocated on TRX of highest TRX rank– and on TS of highest TS index

� SPDCH allocated on TRX of lowest TRX rank– and on TS of lowest TS index

▼ 2 modes of TCH selection� on pure TCH or TCH/SDCCH timeslots� on TCH/SPDCH timeslots

B8

▼ TCH selection on pure TCH or TCH/SDCCH timeslots if:� there is at least one candidate TCH free on pure TCH TSOR� there is no candidate TCH free on TCH/SPDCH TS

� only the candidate TCH sub-channels available on pure TCH TS and on TCH/SDCCH TS are kept as candidate

▼ TCH selection on TCH/SPDCH timeslots if:� there is at least one candidate TCH free on a TCH/SPDCH TSAND� there is no candidate TCH free on pure TCH TS

� only the candidate TCH sub-channels available on TCH/SPDCH TS are kept as candidate


1.233

3.3 TCH resource allocation algorithmTCH selection on pure TCH or TCH/SDCCH TS

▼The TCH is chosen from the selected sub-pool according to the following criteria:

TCH candidatesof the selected TCH sub-pool

Highest TRX_PREF_MARK

BiggestMobile Allocation

Best Interference Band

EN_MA_SELECTION= true

non hopping cell

Highest TRX identity

Highest TS index

HR 0 TCH sub-channel

Selected TCH

FR allocationor HR allocation on busy TS

B8

▼ The BSS attempts to offer the best quality of service for TCH calls in accordance with the privileged order between the groups of TRXs (if any) defined by the operator. Among a group of TRXs the BSS attempts to allocate traffic channels that have the best quality characteristics (channels using frequency with low reuse factor, large hopping frequency sets, low measured interference).

▼ The benefits from this type of allocation are that the operator has the possibility to define groups of TRXs and to favour (or to disadvantage) them on the other if he wants to do so. Among a group of pure TCH or TCH/SDCCH timeslots, the overall interference is kept as low as possible, thus the user will perceive a better quality of service.

▼ The BSS chooses the best TCH among the sub-channels of the selected TCH sub-pool applying criteria below in the specified order of priority:

1. TCH on TS with the highest TRX Preference Mark

� According to the frequency plan, the coverage and interference probability of a cell (or according to measurements), the operator may know which TRX should be a priori favored for TCH selection. For that purpose, it is possible for operators to give a preference mark to each TRX of a cell. This mark is given through the parameters TRX_PREF_MARK (TPM) changeable at OMC-R side per TRX. The range of TRX_PREF_MARK will be from 0 (lowest priority) to 7 (highest priority). The TCH selection function favours the channels with the highest TPM.

� Note that a few Pure TCH TS should be available in a cell on a TRX of TRX_PREF_MARK value of 0 since TCH/SPDCH TS may also be defined on this TRX according to PS radio resource configuration.

2. TCH on TS with the biggest Mobile Allocation (for hopping cell only)

� Considering that the number of frequencies is a key factor for the average quality of channels, the TCH selection function favors the TS with the biggest MA (i.e. with the most frequencies in their frequency hopping sequence). This selection criterion is enabled/disabled via the flag EN_MA_SELECTION changeable at the OMC-R side on a per cell basis.

3. TCH on TS from the best Interference Band

� Considering that the uplink received level measured by the BTS on an idle channel is a means to assess the quality when in connected mode, the TCH selection function favours the TS belonging to the best Interference Band (IB). Five IBs are defined through 5 parameters INTFBD1 to INTFBD5 where INTFBD(i)< INTFBD(i+1) and INTFBD5 = -47 all changeable at the OMC-R side on a per BTS basis.

4. TCH on TRX with the highest TRX identity

5. TCH on TS with the highest TS index

6. HR 0 TCH if the two sub-channels remaining candidates are the 2 HR TCH of the same free TS


1.234

3.3 TCH resource allocation algorithmTCH selection on TCH/SPDCH TS

▼The TCH is chosen from the selected sub-pool according to the following criteria:

� TRX rank is determined by the TRX Ranking algorithm described in the “GPRS & EGPRS Radio Algorithms Description” training course

HighestTRX rank

Highest TS index

HR 0 TCH sub-channel

TCH candidatesof the selected TCH sub-pool

Selected TCH

FR allocationor HR allocation on busy TS

B8

▼ The BSS tends to allocate to the MFS the TCH/SPDCH timeslots so as to avoid conflicts between CS and PS allocations on PS capable TRX.

▼ In order to be able to allocate as much slave PDCHs as possible to a given TBF, it is important to avoid any mix ofallocation between TCHs and SPDCHs (e.g. avoid on a TRX a configuration such as TCH – TCH – SPDCH – SPDCH –TCH – SPDCH – SPDCH – SPDCH). For that purpose, a TRX rank is assigned to each PS capable TRX. The TRX having the highest TRX rank is preferentially selected for TCH allocations, whereas TRX having the lowest TRX rank is preferentially selected for SPDCH allocations

▼ This rule only applies on PS capable TRX. On a given PS capable TRX, TCH are preferentially allocated on the right side of the TRX (highest TS index), whereas SPDCH are preferentially allocated on the left side (lowest TS index).


1.235

3.3 TCH resource allocation algorithmExercise 2 - 1/3

▼A cell is configured on the OMC-R and TRE are mapped by BSS

0 1 2 3 4 5 6 7TRX1 BCC TCHSDC TCH TCH TCH TCH TCH

TRX2 TCHSDCSDD TCH TCH TCH TCH TCH TCH

TRX3 TCH TCH TCH TCH TCH TCHTCHSDC

TRX4 TCH TCH TCH TCH TCH TCHTCH TCH

TRX5

TRX_PREF_MARK

0

0

1

0

1 TCH TCH TCH TCH TCH TCHTCH TCH

TRE

G4 MP FR

G4 MP DR

G3 DR

G4 MP FR

G3 DR

B8

Time allowed:

10 minutes


1.236


▼ Find the radio TS configuration in RAM if NB_TS_MPDCH= 2

TRX_PREF_MARK

0

0

1

0

1

0 1 2 3 4 5 6 7TRX1

TRX2 SDC

TRX3

TRX4

TRX5

TRE

G4 MP FR

G4 MP DR

G3 DR

G4 MP FR

G3 DR

PBCPure BCCH TS

PSDPure SDCCH TS

TSDTCH/SDDCH TS

PTC

TSP

MPD

Pure TCH TS

TCH/SPDCH TS

MPDCH

B8


1.237


▼ Find which TCH sub-channel is allocated:

1. For MS1: E-GSM, DR

2. For MS2: GSM/DCS, DR

3. For MS3: GSM, FR

4. For MS4, MS5, …., MSn: E-GSM, DR n = ?

H

0 1 2 3 4 5 6 7TRX1 P P P

TRX2 P P PP FSDCTSD

F F F F F FTRX3

P P P P P P PTRX4

F FTRX5

TRX Rank

2

3

-

1

-

TRE

GSM/FR

GSM/DR

GSM/DR

GSM/FR

G1/DRH HHH HH H

P: SPDCH TS

F: FR TCH callH: HR TCH call

Cell load = true

Pure TCH TS

TCH/SPDCH TS

TCH/SDDCH TS as TCH TS

B8



3 OTHER ALGORITHMS

3.4 MS Reselection algorithms

LCS




1.239

▼At startup (IMSI Attach), the MS is selecting a cell with

� best C1

� once “camped on” one cell (in idle mode)…

▼…the MS can decide to reselect on another one if:

� C1 criteria is too low

� the MS cannot decode downlink messages

� the current cell is becoming forbidden (e.g. barred)

� the MS cannot access the cell

� there is a better cell, regarding C2 criteria

3.4 MS Reselection algorithmsSelection and reselection principles

▼ Idle mode� Status null:

the mobile station (MS) is off� Status search BCCH:

the MS searches a broadcast channel with the best signal level (cell selection and reselection)� BCCH list: up to 36 BCCH frequencies plus BSIC can be saved on SIM per visited network.� Look if frequencies of the BCCH list can be used.� No entries in the BCCH list, or the location is completely different: scan frequency band.

� Status BCCH:the MS is synchronized on a BCCH. The MS camps on a cell.

� The BTS sends the neighbor cells list (BCCH allocation BA) on BCCH in System Information (SI) 2, 2bis and 2ter if BSS parameter EN_INTERBAND_NEIGH in dual band networks:

� GSM900 serving cell- GSM900 neighbor cells put into SI 2- GSM1800 neighbor cells put into SI 2ter/2bis

� GSM1800 serving cell- GSM900 neighbor cells put into SI 2ter- GSM1800 neighbor cells put into SI 2/2bis

� The MS measures RXLEV from BCCH of the serving and neighbor cells.� Camping on a cell is performed using C1 criteria only (the chosen cell is the one with the best C1)

� The MS needs to have access to the network.� The MS needs to be accessible by the network.

� Reselection is done using the mechanisms referenced above� ‘handover algorithms’ in idle mode


1.240

▼C1

� ensure that, if a call was attempted, it would be done with a sufficient downlink and uplink received level

� based on 2 parameters, broadcast on BCCH

�RXLEV_ACCESS_MIN [dBm]�minimum level to access the cell

�MS_TXPWR_MAX_CCH [dBm]�maximum level for MS emitting

3.4 MS Reselection algorithmsC1 criteria (1/2)


1.241

▼C1

� evaluated every 5 sec (minimum)

� C1 = A - MAX(0,B) > 0

� A = RxLev - RXLEV_ACCESS_MIN

� assess that the MS received level is sufficient

� B = MS_TXPWR_MAX_CCH - P

�P maximum power of MS� assess that the BTS received level will be sufficient� if MS_TXPWR_MAX_CCH < P

� If A > 0 & B < 0 OK, if B > 0, it can be compensated by A

�A >> 0 means that the MS is closer to the BTS

3.4 MS Reselection algorithmsC1 criteria (2/2)


1.242

▼C2� CELL_RESELECT_PARAM_IND= not present THEN C2=C1 else

� C2 = C1 + CELL_RESELECT_OFFSET - TEMPORARY_OFFSET (T) (if PENALTY_TIME ≠ 31) � if T > PENALTY_TIME, TEMPORARY_OFFSET(T) = 0� used to avoid locating on “transient cell”� CELL_RESELECT_OFFSET used to favor cell among other (e.g.

micro-cell vs. umbrella, once T > PENALTY_TIME)

� Or C2 = C1 - CELL_RESELECT_OFFSET (if PENALTY_TIME = 31)

� CELL_RESELECT_OFFSET used to handicap some cells among others

� One reselection criterion is compared to C2s� C2neighbor > C2current if cells belong to same LA� C2neighbor > C2current+Cell_Reselect_Hysteresis if cells from a

different LA

3.4 MS Reselection algorithmsC2 criteria

▼ Note:� CRO: from 0 to 126 dB, step 2dB� PENALTY_TIME: from 1=20s to 30=600s, step: 20s; 31=infinite� TEMPORARY_OFFSET: from 1=10dB to 6=60dB; 7 = infinite

▼ The use of a second formula (Penalty_time = 31) is restricted to very special cases, as we do not like to penalize a cell. If a cell is parametered with PT=31, it will be penalized compared to ALL its neighbors. To penalize a cell compared to one neighbor, one should better boost the neighbor cell (using the first formula).

▼ The first formula is very useful for favoring indoor cell or microcell.

▼ Cell selection and cell reselection considering CELL_BAR_QUALIFY� in case of phase 2 MS and CELL_RESELECT_PARAM_IND=1, it is possible to set priorities to cells� CELL_BAR_QUALIFY

� Two values:� 0 = normal priority (default value)� 1 = lower priority

� Interacts with CELL_BAR_ACCESS (barring cell) � A phase 2 MS selects the suitable cell with the highest C2 (C1>0) belonging to the list of normal priority.� If no cell with normal priority is available then the MS would select the lower priority cell with the highest C2

(C1>0).


1.243

3.4 MS Reselection algorithmsTraining Exercise (1/2)

▼On this network example

� List the parameters involved in the selection / reselection process

CI=6169GSM900

CI=1823

GSM900

CI=1964

GSM900

CI=6270

GSM900

CI=6271GSM900 Cell (8557,1823)

Cell (8564,6169)Cell (8564, 1964)

Time allowed:

5 minutes


1.244

3.4 MS Reselection algorithmsTraining Exercise (2/2)

� Find the selected cell by the MS

5

4

3

2

1

Measurements

-78-85-89

-82-87-88

-87-90-88

-100-90-84

-104-96-80

RxLev (3)RxLev (2)RxLev (1)

CI=6169GSM900

CI=1823

GSM900

CI=1964

GSM900

CI=6270

GSM900

CI=6271GSM900

Cell 3(8557,1823)

Cell 2(8564,6169) Cell 1

(8564, 1964)





1.246

4 ALGORITHMS DYNAMIC BEHAVIOR Session presentation

▼Objective: to be able to Estimate qualitatively the impact of parameters change

▼Program:


4.2 Examples and exercises





S5: CASE STUDIES






Examples and Exercises


1.248

▼ SESSION OBJECTIVES

� Be able to estimate qualitatively the impact of a parameter change

▼ JUSTIFICATION

� Tuning is not an exact science

� The optimizer has to control every parameter change and predict qualitatively what the consequences will be

▲ Note: Each change of parameter and its justification have to beregistered in a database for operation convenience

▼ DETAILED PROGRAM

� Three Example/Exercises

4.1 Theoretical presentationSession objectives




4.1 Examples and Exercises


Examples and Exercises


1.250

4.2 Examples and ExercisesOverview

▼Example 1: Optimization of handover algorithms

� Sliding averaging window

▼Example 2: Optimization of power control algorithms

� Sliding averaging window

▼Example 3: Traffic load sharing

� Parameters qualitative influence


1.251

▼Search for best tuning of HO parameters to decrease call drop

HO/Call

Call drop

4.2 Examples and ExercisesExample 1: Optimization of Handover Algorithms (1/4)


1.252

▼Main Objective: make the HO algorithm as efficient as possible

� Minimize call drop rate

� trigger HO soon enough� toward the “best” neighbor

� while keeping a good speech quality

� avoid HO due to quality: “too late”� avoid having HO/call rate too high



1.253

▼ Method

� Collect Abis trace chart

� Search for HO level to avoid quality lower than 4 (or even 3)

� sufficient number of “bad quality” samples

� low standard deviation� problem when HO already

activated

< R x Q u a l _ D L > = f ( A V _ R x L e v _ D L )

0

1

2

3

4

5

6

7

N b _ s a m p l e s

0

2 0 0

4 0 0

6 0 0

S t a n d a r d D e v i a t i o n

0

0 .5

1

1 .5

2

< R x Q u a l _ U L > = f ( A V _ R x L e v _ U L )

0

1

2

3

4

5

6

7

N b _ s a m p l e s

02 0 04 0 06 0 08 0 0

1 0 0 0

S t a n d a r d D e v i a t i o n

0

1

2

3

▼ Then tune according to QoS indicators (OMC-R) by repetitive process

� A_PBGT_HO/A_LEV_HO/A_QUAL_HO

� L_RXLEV_UL_H, L_RXLEV_DL_H, L_RXLEV_UL_P, L_RXLEV_DL_P

� OK as soon as HO success rate stabilized


▼ Never forget that Abis information takes into account the traffic distribution in the cell. Any parameter tuning done after an Abis study has to be checked periodically as the distribution in the cell can change from one week to another.

▼ Use the pivot table function (Excel) to build this graph.

RxQUAL

0

1

2

3

4

5

6

7

-110

-108

-106

-104

-102

-100 -98 -96 -94 -92 -90 -88 -86 -84 -82 -80 -78 -76 -74 -72 -70 -68 -66 -64 -62 -60 -58 -56 -54 -52 -50 -48

RxQUAL


1.254

▼ neighboring relationship cleanup

� Remove useless relationships (A interface statistics, PM Type 180)

� Remove the common BCCH/BSIC couple

� Add new relationships when a new site is created

▼ Finally, check the main QoS indicators

� Call drop rate

� HO failure rate

� HO/call rate



1.255

▼According to the Abis results and some parameters already set,

▼ tune qualitatively the sliding averaging windows:

� A_QUAL_HO� A_LEV_HO

4.2 Examples and ExercisesExample 1: training exercise

4??A_LEV_HO

?66A_QUAL_HO

-90dBm-90dBm-85dBmL_RXLEV_DL_H

-90dBm-96dBm-80dBmLevel at RxQual=3

Time allowed:

5 minutes


1.256

▼Optimization of Downlink Power Control

� Decrease of downlink interference� Risks of delay of HO (without fast power control)

▼Optimization of Uplink Power Control

� Decrease of Uplink interference� MS battery saving� Risks of delay of HO (without fast power control)

4.2 Examples and ExercisesExample 2: Optimization of Power Control Algorithms (1/2)


1.257

▼The main tuning problem is the interaction with handover, which can slow down HO decision, and debase call drop rate

� Power control threshold must be within HO ones

� Dynamic step size must be activated if possible

4.2 Examples and ExercisesExample 2: Optimization of Power Control Algorithms (2/2)

▼ In the example below, a dynamic MS PC is activated. The MS power changes are really reactive and control the UL level between -80 and -90dBm. In this example, the HO threshold is -98 dBm.

RxLev_UL

-100

-95

-90

-85

-80

-75

-701 39 77 115 153 191 229 267 305 343 381 419 457 495 533 571 609 647 685 723 761 799 837 875 913 951 989 1027

RxLev_UL

13

15

17

19

21

23

25

27

29

31

33

1 40 79 118 157 196 235 274 313 352 391 430 469 508 547 586 625 664 703 742 781 820 859 898 937 976 1015

MS_PwrLevel


1.258

▼Explain qualitatively the impacts of some parameter changes

�What happens if:

�we increase POW_INC_FACTOR?�we increase MAX_POW_INC?�We increase A_LEV_PC?

4.2 Examples and ExercisesExample 2: Training Exercise

Time allowed:

5 minutes


1.259

▼ Used to unload cell with too high traffic, without HW extension

▼ Trade-off between traffic sharing/radio quality

▼ Different algorithm

� Fast Traffic Handover: Cause 28

� Traffic Handover: Cause 23 and 12 with DELTA_HO_MARGIN(0,n)

� Static (couple of cells): HO_MARGIN, LINK_FACTOR

� On a local traffic basis:

� Load_Factor/Free_Factor

� Forced Directed Retry

4.2 Examples and ExercisesExample 3: Traffic Load Sharing (1/12)


1.260


▼Fast Traffic HO

� Useful in case of sudden traffic peaks as the process response is instantaneous (no averaging window)

� The principle is to force handover towards neighbor cells which have lower traffic when a request is queued in the serving cell.

� Interaction with Forced DR due to the use of same thresholds

� Optimization method (repetitive process)

� Tunes L_RXLEV_NCELL_DR(n), FREElevel_DR(n)� Applies new values, checks traffic peaks, QoS

indicators


1.261


▼The Pros and cons of Fast Traffic HO� Efficiency depends on

� traffic location in the loaded cell� capacity of neighbor cells

➘ increase of the number of HO/call➘ increase of incoming HOs fail rate (risk of ping-pong

effect)� in case of internal HO: use PING_PONG_HCP with T_HCP

or/and enable HO CAUSE 23

➘ heavy to tune (has to be done for each couple of cells)➚ Adapted to instantaneous traffic modification➚ can be used to send traffic towards a cell external to

the serving BSC➚ adapted to hierarchical network, but also to standard

ones


1.262

▼DELTA_HO_MARGIN (0,n)

▼CHANGE DYNAMICALLY TRAFFIC DISTRIBUTION WITH HO:� Traffic HO Cause 23

� Ease outgoing better condition HO on a traffic point of view

� Slow down outgoing better cell HO (to be tuned for a given couple of cells)� When the better cell in radio condition is the worst cell

in traffic terms� Optimization method (repetitive process)

� Tune DELTA_DEC_HO_MARGIN and DELTA_INC_HO_MARGIN

� Apply new values, check traffic, QoS indicators and possibly speech quality



1.263

▼The Pros and cons of DELTA_HO_MARGIN (0,n) method� Efficiency depends on

� traffic location in the loaded cell� cells overlap� capacity of neighbor cells

➘ increase the number of HO/call➘ cannot be used to send traffic toward a cell external to

the serving BSC➘ the call has to be first established on a loaded cell,

before being “exported”� it can be rejected

➚ Easy to tune (dynamic process)➚ Adaptability to instantaneous and long term traffic

modifications



1.264

▼HO_MARGIN / LINK_FACTOR▼CHANGE STATICALLY TRAFFIC DISTRIBUTION WITH HO:

� Ease outgoing better cell HO (to be tuned for a given couple of cells)� Decrease HO_MARGIN (can make a cell “candidate”)� Increase LINK_FACTOR (used to rank candidate cells)

� Optimization method (repetitive process)� Look for neighbor cells able to carry extra traffic� Use Abis trace to check if these cells are candidate

� if yes, use LINK_FACTOR to favor them� if not, use HO_MARGIN and LINK_FACTOR

� Apply new values, check traffic, QoS indicators and possibly speech quality



1.265

▼The Pros and cons of LINK_FACTOR/HO_MARGIN� can be efficient (up to 20% increase of capacity) in

some cases� cell overlap� capacity of neighbor cells

➘ increase the number of HO/call➘ the call has to be first established on a loaded cell,

before being “exported”➘ it can be rejected

➘ heavy to tune (has to be done for each couple of cells)➘ no adaptability to instantaneous and long term traffic

modifications➚ can be used to send traffic toward a cell external to the

serving BSC



1.266

▼FREE_FACTOR/LOAD_FACTOR▼ Taking into account the current load of cells, send the MS toward the less

loaded cell with HO

� Ease outgoing better cell HO, according to

� Load_Factor (% of TCH occupancy) of serving and “target” cells� Free_Factor (number of free TCHs) of serving and target cells

(order only)� cannot make a “candidate” cell, only change ranking

� Tuning method (repetitive)

� to be activated locally for each cell with default parameter setting� look for QoS indicators (esp. traffic intensity and blocking rate)� tune tables accordingly



1.267

� The Pros and cons of load/free factors method

➘ Lower efficiency compared to LINK_FACTOR/HO_MARGIN

➚ No increase of HO/call rate➘ Calls have to be established on a loaded cell before

being “exported”➘ Tuning is performed on a cell-per-cell basis➘ Cannot be used to send traffic toward an external cell➚ Adapted to dynamic change of traffic and capacity (for

Load_Factor)



1.268

▼Forced directed retry method

� Mechanisms

� the MS is connected on an SDCCH of cell1

� it must switch on TCH

� no TCH is free on cell1

� there is at least 1 neighbor cell which has

�sufficient DL level seen by the MS�enough free TCHs

� the MS is handed over to TCH towards this cell

� if there are several cells, the one with the best PBGT is selected



1.269

▼method: trade-off between traffic and radio quality

� mainly L_RXLEV_NCELL_DR(n) parameter to tune� the lower, the better the traffic sharing� the lower, the higher the interference risks

� QoS indicators and field tests (speech quality) are necessary for tuning

cell 2 : 45 cell 3 : 23

cell 1 : 24


▼ Forced directed retry� The following condition is checked every measurement reporting period and if at least one input pre-processed

parameter AV_RXLEV_NCELL_DR(n) is available.� CAUSE = 20 (high level in neighbor cell for forced directed retry)

� AV_RXLEV_NCELL_DR(n) > L_RXLEV_NCELL_DR(n) (n = 1 ... BTSnum)� and EN_FORCED_DR = ENABLE

� The threshold L_RXLEV_NCELL_DR(n) is the observed level from the neighbor cell n at the border of the area where forced directed retry is enabled. This threshold fixes the size of the overlapping area where forced directed retry can be performed. It should be greater than RXLEVmin(n).


1.270

▼The Pros and cons of Forced directed retry

➚ highest efficiency (up to 30%)

➚ no increase of HO/call rate

➚ can be used to send traffic toward an external cell

➚ adapted to dynamic change of traffic

➚ adapted to hierarchical networks, but also to standard

ones

➘ tuning is performed on a cell-per-cell basis



1.271

▼Draw qualitatively the new serving areas on the pseudo map when enabling traffic HO with:

� DELTA_DEC_HO_MARGIN=6dB� DELTA_INC_HO_MARGIN=4dB

4.2 Examples and ExercisesExample 3: training exercise (1/3)

Loaded cell 0 Unloaded cell n

PBGT(n) = 5PBGT(0) = 5

EN_TRAFFIC_HO = 0

Cause 12Cause 12

PB

GT

(0)=

PB

GT

(n)=

0SolutionTime allowed:

5 minutes


1.272


▼What happens when EN_FAST_TRAFFIC_HO = ENABLE and EN_TRAFFIC_HO(0,n) = DISABLE


PBGT(n) = 5PBGT(0) = 5P

BG

T(0

)=P

BG

T(n

)=0

Queued Ass Req Av_Rxlev_Ncell(n) = -82dBm Av_Rxlev_Ncell(0) = -74dBmAv_Rxlev_PBGT_HO = -82dBm

L_RLEV_NCELL_DR(n) = -85dBm

Time allowed:

5 minutes


1.273


▼What happens when EN_FAST_TRAFFIC_HO = ENABLE and EN_TRAFFIC_HO(0,n) = ENABLE


PBGT(0) = 5

PB

GT

(n)

= -

1

PB

GT

(0)

= 9

Queued Ass Req

Av_Rxlev_Ncell(n) = -82dBm

Av_Rxlev_Ncell(0) = -74dBmAv_Rxlev_PBGT_HO = -82dBm

Time allowed:

5 minutes



5 CASE STUDIES


1.275

5 CASE STUDIES Session presentation

▼Objective: to be able to propose a set of parameters to solve typical radio problems

▼Program:5.1 Theoretical presentation5.2 TUNNEL Case5.3 RADAR Case5.4 TOWER Case5.5 RESURGENCE Case5.6 FOREST Case5.7 HIGHWAY Case5.8 TCH/SDCCH CONGESTION Case5.9 INDOOR CELL CONGESTION Case





S5: CASE STUDIES



5 CASE STUDIES



TUNNEL CaseRADAR CaseTOWER Case

RESURGENCE CaseFOREST Case

HIGHWAY CaseTCH/SDCCH CONGESTION Case

INDOOR CELL CONGESTION Case


1.277

▼ SESSION OBJECTIVES

� Be able to propose an appropriate set of parameters to solve

typical field problems

▼ JUSTIFICATION

� Some typical problems due to particular field configuration always

occur in a GSM network

▼ DETAILED PROGRAM

� Eight typical case studies

5.1 Theoretical presentationSession objectives



5 CASE STUDIES

5.2 Tunnel Case







1.279

5.2 Tunnel Case

▼Radiating cable in a tunnel� Question:

� Risks of such a configuration� tune the right parameters for the tunnel cell

�catch quickly ‘car traffic’ �avoid the pedestrian traffic

Repeater

Radiatingcable

Tunnel

OutdoorBTS Pedestrian

mobile

‘IndoorBTS’



5 CASE STUDIES

5.3 Radar Case







1.281

5.3 Radar Case

▼Radar situation� A “radar” cell situated on top of a hill provides a wide coverage area.

� An industrial zone in the valley is covered by small cells but also by the “radar” cell. The serving areas in the IZ are not clearly defined.

▼Objective� Give a parameter set to prevent the radar cell from catching any traffic

in the industrial zone by HO assignment



5 CASE STUDIES

5.4 Tower Case







1.283

5.4 Tower Case

▼Tower situation� The indoor mobile selects in idle mode the outdoor cell (same LA)

▼Objective� Define a set of parameters to avoid that effect

IndoorAntenna

BS1



5 CASE STUDIES

5.5 Resurgence Case







1.285

5.5 Resurgence Case

▼Resurgence situation� In rural network, especially in hilly landscape, many resurgences occur

from very far cells.

▼Objective� Define a set of parameters to avoid radio link establishment to those

cells and TCH traffic on those cells

Resurgence from cell A

25 km

Cell A

Cell B



5 CASE STUDIES

5.6 Forest Case







1.287

5.6 Forest Case

▼Forest situation: A highway crosses a forest� High call drop rate (radio cause) on the cell and drive tests: strong level

attenuation at the entrance of the forest

▼Objective� Define a set of parameters to avoid radio link failure

-75dBm -90dBm

Highway

ForestATT=10dBevery 100m

BTS



5 CASE STUDIES

5.7 Highway Case







1.289

5.7 Highway Case

▼ Highway situation:� A highway is slightly covered (best coverage on 200m) by an

‘orthogonal’ cell (cell C on the map)

▼ Objective� Define a set of parameters to avoid traffic in the ‘orthogonal cell’

Cell A

Cell B

Cell C



5 CASE STUDIES

5.8 TCH/SDCCH congestion case







1.291

5.8 TCH/SDCCH Congestion Case

▼SDCCH congestion situation� A railway station is located at the frontier of two LAs. Every train

stopping in this station comes from LA 1 and then return to LA 1after the stop.

▼Objective� Define a set of parameters to avoid SDCCH congestion on cell B

(LA 2)

LA2 LA1

Cell B

LA frontier

Cell A



5 CASE STUDIES

5.9 Indoor cell congestion case







1.293

▼An indoor microcell has been introduced within a multi-layer network (macro + micro)

▼When the indoor microcell is congested, FDR may not be working as some the MSs can be covered only by this cell� Define parameter settings to find a good solution in case of indoor cell

congestion

5.9 Indoor cell congestion


1.294

END SESSION



ANNEXES

TYPICAL MODULE STRUCTURE objective(s)

theoretical presentationtraining exercises and/or cases study + feedback



ANNEXES

Annex.1 Erlang B law

Erlang B law

Frequency hopping influence on PCHO process

Load & Traffic evaluation

Training exercises solutions


1.297

▼ERLANG: unit used to quantify traffic

� T = (resource usage duration)/(total observation duration) [ERLANG]

� Example:

� 1 TCH is observed during 1 hour� one can observe 1 call of 80 sec and 1 call of 100 sec� the observed traffic is T = (80+100)/3600 = 0.05 ERLANG

Annex.1 Erlang B law Erlang definition


1.298

▼ERLANG <-> CALL MIX

� CALL MIX EXAMPLE

� 350 call/hour� 3 LU/call� TCH duration: 85 sec�SDCCH duration: 4.5 sec

� ERLANG COMPUTATION

� TCH = (350 * 85)/3600 = 8.26 ERLANG�SDCCH = [ (350 + 350*3) * 4.5 ] / 3600 = 1.75

ERLANG

Annex.1 Erlang B law Call mix definition

▼ 350 calls * 85 sec / 1 hour(3600 sec):� TCH = (350 * 85)/3600 = 8.26 ERLANGS

▼ 350 calls means 350 SDCCH phases.▼ 3 LU/call means 3 * 350 LUs so 1050 SDCCH phases more.▼ 1 SDCCH phase is 4.5 sec:

� SDCCH = [ (350 + 350*3) * 4.5 ] / 3600 = 1.75 ERLANG


1.299

Annex.1 Erlang B law Erlang B (1/5)

▼ERLANG B LAW

� relationship between

� offered traffic� number of

resources� blocking rate

▼ In a telecom system, call arrival frequency is ruled by the POISSON LAW

call/second

0

1

2

3

4

5

6

7

8

9

10

1 5 9

13

17

21

25

29

33

37

41

45

49

53

57

61

65

69

73

77

81

85

89

93

97

call/second

0

1

2

3

4

5

6

7

8

9

10

1 5 9

13

17

21

25

29

33

37

41

45

49

53

57

61

65

69

73

77

81

85

89

93

97

▼ The offered traffic is the traffic asked by the customers.

▼ The graph gives the number of connection requests per second during 35 seconds.

▼ 83/30s => 83 * 2 * 60 = about 10 000 / hour

▼ Real example in Paris on 1 BSC (LA FOURCHE).


1.300

▼ call request arrival rate (and leaving) is not stable

� number of resources = average number of requests * mean duration

� is sometime not sufficient => probability of blocking

▼ => Erlang B law

� Pblock: blocking probability

� N: number of resources

� E: offered traffic [Erlang]

� good approximation when the blocking rate is low (< 5%)

P b lo ck N

k

N

k

k

N

E

E=

=∑

!

!0



1.301

▼There is two different ways to use this law

� Using Abacus

� Using SW (here Excel)

�Pblock = f (T, Nc)�Offered = f (Nc, Pblock)�Channels = f (T, Pblock)



1.302

▼Example:

We have a BTS of 8 TRXs (about 60 channels (Nc))

We do not want more than 2% of blocking (Pblock)

=> The traffic is not to be greater than 50 Erlangs (T)

� 83% of resources used to reach 2% of blocking



1.303

▼But be careful, the law is not linear:

▼ In B4, we use for example a combined BCCH with a micro BTS.

4 SDCCHs, Pblock = 2% => T = 1.1 E

25% of resources used to reach 2% of blocking

▼ In B5, if we decide to provide SMSCB (Cell Broadcast information)

1 subchannel SDCCH is therefore used.

3 SDCCHs, Pblock = 2% => T = 0.6 E

25% of resources less => 50% of Traffic less !!



1.304

▼CELL DIMENSIONING

� Given an Offered traffic, compute the number of TRXs (and SDCCHs) needed to carry it

� default blocking rate

�RTCH: 2%�SDCCH: 0.5%� (TTCH: 0.1%)

Annex.1 Erlang B law Cell dimensioning (1/5)


1.305

▼CELL DIMENSIONING

� to handle an offered traffic of 12 Erlangs (TCH), compute the number of channels, then the number of TRXs

� Channels (12;2%) = 19

� example: 3 TRXs , 21 TCHs, 1 BCCH, 2 SDCCH8



1.306

▼CELL DIMENSIONING, based on field measurement

� one is measuring a traffic of 15 Erlangs, with a blocking rate of 10%

� how to dimension the cell?

� Offered traffic = 15 / (1-10%) = 16.7 Erlangs !!!!

� Channels (16.7;2%) -> 25 TCHs -> 4 TRXs needed



1.307

▼FORECASTING TRAFFIC/CRITICAL TRAFFIC

� traffic forecasting must be calculated according to offered traffic

✭ not directly on measured traffic

� In order to plan necessary actions soon enough, one must calculate regularly the date when the traffic of a cell will become critical

� critical traffic: when offered traffic will induce 2% of blocking



1.308

▼WARNING: IN CASE OF TOO HIGH BLOCKING RATE

� FIRST CHECK THAT THERE IS NO OUTAGE ON THE BTS

� BEFORE STARTING A DIMENSIONING/TUNING ACTION



1.309

▼TRAINING EXERCISE Complete this form in order to get less than 2% of blocking in all cases.

cellcellcellcell call mix infocall mix infocall mix infocall mix info Erlang TCHErlang TCHErlang TCHErlang TCHOffered trafficOffered trafficOffered trafficOffered traffic

traffic forecasttraffic forecasttraffic forecasttraffic forecast proposed configproposed configproposed configproposed config

12, 743 450 call/hourmean TCH call duration : 80secblocking rate TCH : 0.8%

10.08 Erlang TCH 30 % offered trafficincrease

13.1 Erlang TCH - > 20 TCH3 TRX

12,675 330 call/hourmean TCH call duration 129secblocking rate 4%

30 % offered trafficincrease

12,865 600 call/hourmean TCH call duration 96secblocking rate 8 %

30 % offered trafficincrease

Annex.1 Erlang B law Training exercise

Back

cellcellcellcell call call call call mix infomix infomix infomix info Erlang TCHErlang TCHErlang TCHErlang TCH traffic traffic traffic traffic forecastforecastforecastforecast proposed proposed proposed proposed configconfigconfigconfig

12, 743 450 call/hourmean TCH call duration : 80secblocking rate TCH : 0.8%

10 Erlang TCH

(450*80)/3600=1010/.992=10.081

30 % TCH increase

10,081*1.3=13.1

13,1 Erlang TCH - > 20TCH

3 TRX

12,675 330 call/hourmean TCH call duration 129secblocking rate 4%

(330*129)/3600=11.825/0.96=12.3177

30 % TCH increase

12.3177*1.3 =16

16 Erlang TCH -> 24 TCH

4 TRX

12,865 600 call/hourmean TCH call duration 96secblocking rate 8 %

(600*96)/3600=16/.92 = 17.4

30 % TCH increase

17.4*1.3 = 22.6

22.6 Erlang TCH -> 31 TCH

5 TRX



ANNEXES

Annex.2 Frequency Hopping influence on PCHO process

Erlang B law





1.311

Annex.2 Frequency Hopping influence on PCHO process(1/4)

▼Signal decoding process

� In a GSM system, the number of frames that are not erased are sent as an input to the voice decoder

DEMOD DECODER

ENCODER

FrameErasureDecision

VoiceDecoder

-

RXQUAL Frame Erasure Rate

DeinterleaveError correct.

Inside the mobile stationAir


1.312

▼Quality impact of frequency hopping on the reception chain

� In non-hopping networks, the RXQUAL and voice quality are

correlated

� In hopping networks, the voice quality is sooner correlated

to the FER. This is due to interferer averaging and due to

the non-linear mapping of BER to RXQUAL values.



1.313

▼Quality impact of frequency hopping on the reception chain

� FER is improved when frequency hopping is activated (cyclic or random)

� RxQual is not impacted whereas the speech quality is better


Average (RxQ & FER)

0

0,2

0,4

0,6

0,8

1

1,2

1,4R

ef

Cyc

li

Ran

dom

0,00%

0,50%

1,00%

1,50%

2,00%

2,50%

RxQ Average

FER Average


1.314

▼Conclusion

� When frequency hopping is activated

� We can accept in Power Control and Handover processes, a threshold increase:

� OFFSET_HOPPING_PC and� OFFSET_HOPPING_HO

Annex.2 Frequency Hopping influence on PCHO processConclusion (4/4)

Back



ANNEXES

Annex.3 Load & Traffic evaluation

Erlang B law





1.316

Annex.3 Load & Traffic evaluationCell TCH radio resource evaluation usage

BackCause 12

Load evaluation Period Usage

short term TCH_INFO_PERIOD FREEfactors

LOADfactors

medium term LOAD_EV_PERIOD x TCH_INFO_PERIOD Speed discrimination for hierarchical network

Full Rate / Half Rate channel allocation

long term N_TRAFFIC_LOAD x A_TRAFFIC_LOAD x TCH_INFO_PERIOD Power budget Handover

Traffic Handover

Multiband capture Handover

General capture Handover

BackCause 26


1.317

Annex.3 Load & Traffic evaluationLoad evaluation (1/5)

▼medium term measurement of the load of a cell

� corresponds to function AV_LOAD(cell)

� a new sample of the “Nb free TCH” in the cell is available every TCH_INFO_PERIOD seconds

� AV_LOAD() is a non-

� sliding window load average from Nb free TCH samples updated every LOAD_EV_PERIOD x TCH_INFO_PERIOD sec


1.318


▼AV_LOAD(cell n) calculated from N Nb free TCH samples available during LOAD_EV_PERIOD x TCH_INFO_PERIOD sec

� LOADfactors and FREEfactors also determined from Nbfree TCH samples every TCH_INFO_PERIOD seconds (short term evaluation)

� LOADlevels are boundaries of load intervals associating a LOADfactor (db) to a Nb of free TCH samples

� FREElevels are boundaries of Nb of free TCH intervals associating a FREEfactor (db) to a Nb of free TCH samples

100*)(n) TCHTot Nb

(n) TCH free Nb1(

Nsamples

1 = AV_LOAD(n)

Nsamples

1 = i∑ −


1.319


▼ LOADfactor determination:

� LOADlevel in %

� LOADfactor in dB

t = (1 - Nb free TCH/Total Nb TCH) x 100 LOADfactor t <= LOADlevel_1 LOADfactor_1LOADLevel_1< t <= LOADlevel_2 LOADfactor_2LOADLevel_2< t <= LOADlevel_3 LOADfactor_3LOADLevel_3< t <= LOADlevel_4 LOADfactor_4LOADLevel_4< t LOADfactor_5


1.320


▼FREEfactor determination:

� FREElevel in absolute number of TCH

� FREEfactor in dB

Nb free TCH FREEfactor t <= FREElevel_1 FREEfactor_1FREELevel_1< t <= FREElevel_2 FREEfactor_2FREELevel_2< t <= FREElevel_3 FREEfactor_3FREELevel_3< t <= FREElevel_4 FREEfactor_4FREELevel_4< t FREEfactor_5


1.321

▼ example: cells with 4 TRXs (28 TCHs)

Nb free TCH FREEfactor t <= 3 - 16 dB3 < t <= 8 - 8 dB8 < t <= 15 0 dB15 < t <= 21 +7 dB21 < t +10 dB


Load = (1-Nb free TCH/Total TCH)x 100 LOADfactor t <= 10% +10 dB10% < t <= 25% +5 dB25% < t <= 50% 0 dB50% < t <= 80% -10 dB80% < t -15 dB

Nb free TCHs = 4Load = 85.7%

Cell 0

FREEfactor(0) = -8 dBmLOADfactor(0) = -15 dBm

Nb free TCHs = 20Load = 28.6%

Cell n

FREEfactor(n) = +7 dBmLOADfactor(n) = 0 dBm

• in ORDER(n): + FREEfactor(n) - FREEfactor(0) = +7 -(-8) = +15 dB

• in GRADE(n): + LOADfactor(n) = +0 = 0 dB

in cell evaluation of cell n for outgoing HO from cell 0:

HO?


1.322

Annex.3 Load & Traffic evaluationTraffic evaluation (1/4)

▼ long term measurement of the load of a cell

� corresponds to function Traffic_load(cell)

� Traffic_load() value is determined from a number N_TRAFFIC_LOAD of consecutive non-sliding window load averages AV_TRAFFIC_LOAD calculated from Nb of free TCH samples updated every A_TRAFFIC_LOAD x TCH_INFO_PERIOD sec


1.323


� 3 possible values for Traffic_load(): high, low, indefinite� initialization: Traffic_load() = indefinite� Traffic_load() becomes:

� high if the last N_TRAFFIC_LOAD consecutiveAV_TRAFFIC_LOAD load averages are all greater than HIGH_TRAFFIC_LOAD threshold

� low if the last N_TRAFFIC_LOAD consecutiveAV_TRAFFIC_LOAD load averages are all lower than LOW_TRAFFIC_LOAD threshold


1.324

� Traffic_load() becomes indefinite if:

� Traffic_load() was high and the last AV_TRAFFIC_LOAD load average is lower than LOW_TRAFFIC_LOAD (or IND_TRAFFIC_LOAD if not 0%)

� Traffic_load() was low and the last AV_TRAFFIC_LOAD load average is greater than HIGH_TRAFFIC_LOAD (or IND_TRAFFIC_LOAD if not 0%)

� Traffic_load(n) is always equal to indefinite if cell n is external to BSC

� HIGH_TRAFFIC_LOAD ≥ IND_TRAFFIC_LOAD ≥LOW_TRAFFIC_LOAD



1.325

▼Example with N_TRAFFIC_LOAD = 3


BackCause 12

BackCause 26


1.326

Annex.3 Load & Traffic evaluationCell TCH radio resource evaluation usage



ANNEXES

Annex.4 Handover Management

Erlang B law





1.328

▼Handover Management made up of:

� Cell Filtering Process (according to call history)

� Handover Decision (according to the best cell in the list)

▼Handover Management followed by:

� Handover Protocol

Annex.4 Handover Management Principles



BTS BSC


HO management

MSC

HO protocol

HO Preparation


1.329

Annex.4 Handover Management Global Handover Process

RawRawRawRaw cellcellcellcell listlistlistlist

HandoverHandoverHandoverHandover

DetectionDetectionDetectionDetection

HandoverHandoverHandoverHandover ManagementManagementManagementManagement HandoverHandoverHandoverHandover protocolprotocolprotocolprotocolHandoverHandoverHandoverHandover PreparationPreparationPreparationPreparation

CandidateCandidateCandidateCandidate

CellCellCellCell

EvaluationEvaluationEvaluationEvaluation

Filtered target Filtered target Filtered target Filtered target Cell listCell listCell listCell list

CellCellCellCell

FilteringFilteringFilteringFiltering

ProcessProcessProcessProcess

HandoverHandoverHandoverHandover

DecisionDecisionDecisionDecision

Ordered target Ordered target Ordered target Ordered target cell listcell listcell listcell list

ExternalExternalExternalExternal ororororInternalInternalInternalInternalChannelChannelChannelChannelchangechangechangechange

Execution target Execution target Execution target Execution target Cell listCell listCell listCell list


1.330

▼Three cell lists:

� Ordered target Cell list

� target cells provided by Candidate Cell Evaluation

� REJ_CELL_LIST

� cells internally rejected by the MSC or BSC

� MS_CELL_REJ_LIST

� cells to which the MS failed to hand over

Annex.4 Handover Management Cell Lists usage

▼ Since B6 release, some changes have been provided to the HO management process which is in charge of the HO execution triggering, when the need of handover is detected by the HO preparation process.

▼ These changes are :

� use of the T_FILTER parameter in a different way than for B5,

� the parameter NBR_HO_ATTEMPTS which was used for internal HO in B5 is removed,

� use of the T7 parameter and of the REJ_CELL_LIST list also for internal HO in B7,

� same behavior in case of internal and external HO in B7,

� immediate attempt after rejection or failure without waiting for a new alarm in case of internal and external HO in B7,

� implicit rejection of cells in B7 with the help of the target cell identity in the HO command received from the MSC.


1.331

� T_FILTER: controls the global handover procedure� started: when a cell list is to be sent by Candidate Cell Evaluation� expiry ⇒ empty target cell list sent to the Handover Management

� T7: controls the clean-up of REJ_CELL_LIST� started: when a target cell list is to be sent to Handover Protocol

� expiry ⇒ empty REJ_CELL_LIST� T_MS_CELL_REJ: clean-up of MS_CELL_REJ_LIST

� started: when an MS reports a failure to seize the target channel

� expiry ⇒ empty MS_CELL_REJ_LIST� T_HO_REQ_LOST: to supervise answer of MSC (no HANDOVER

REQUIRED REJECT message sent)

�Started: HO REQUIRED sent�Stopped: HO COMMAND received �Expiry ⇒ external channel change procedure is

terminated.

Annex.4 Handover Management Timers usage

▼ If the candidate cell list provided by the candidate cell evaluation process is different from the previous one (the number of cells is different or same number of cells but new cells in the list), an alarm is sent to the HOM process. In B7, if T_FILTER expires, it means that the HO is no more necessary.

▼ For both internal and external HOs in case of HO failure from the MS, the cell is filtered until the expiry of the T_MS_CELL_REJ timer. When the T_MS_CELL_REJ timer expires, the rejected cell may be a candidate.

▼ In B7 release, T7 timer is used to manage the REJ_CELL_LIST list and a subsequent HO REQUIRED can be sent to the MSC before T7 expiry if the target cell list has changed (new cell or removed cell).

▼ The REJ_CELL_LIST list is used for both internal and external Hos.

▼ T_HO_REQD_LOST Expiry

� This timer is used to supervise response from the MSC. It is started when sending the first HANDOVER REQUIRED to the MSC and it is stopped in the following cases:

• when HANDOVER COMMAND is received from the MSC or

▼ when HANDOVER REQUIRED REJECT is received from the MSC only if the same number of HANDOVER REQUIRED REJECT messages have been received from the MSC than the number of HANDOVER REQUIRED messages sent to the MSC for this channel change procedure) (i.e. no message crossing over A interface).

� In case where more HANDOVER REQUIRED messages have been sent to the MSC, the timer T_HO_REQD_LOST is not stopped upon HANDOVER REQUIRED REJECT receipt, as there is no way for the BSC to know if the received HANDOVER REQUIRED REJECT is a response to the last HANDOVER REQUIRED message or a response to a previous one (message crossing over A interface).

� On expiry, an O&M error report is raised only when no message has been received from the MSC since the last HANDOVER REQUIRED message, and the external channel change procedure is terminated.


1.332

Annex.4 Handover Management Handover Execution Process

Ordered target cell Ordered target cell Ordered target cell Ordered target cell listlistlistlist

cell 4cell 2cell 8

List of cells previously List of cells previously List of cells previously List of cells previously rejected from MSC or rejected from MSC or rejected from MSC or rejected from MSC or BSCBSCBSCBSC

cell 4

REJ_CELL_LIST list cleared at T7 expiry

List of cells previously List of cells previously List of cells previously List of cells previously rejected for MS failurerejected for MS failurerejected for MS failurerejected for MS failure

cell 8

MS_CELL_REJ_LIST list cleared at T_MS_CELL_REJ expiry

Cell FilteringCell FilteringCell FilteringCell FilteringProcessProcessProcessProcess

remove cells previously rejected from MSC or BSC

remove cells previously rejected for MS failure reason

remove cells not suitable due to O&M reason

Handover Handover Handover Handover decisiondecisiondecisiondecision

relevant handover protocol is chosen according to the type of GSM procedure ongoing and the first target cell of the list

T7 is started

Filtered Target cell Filtered Target cell Filtered Target cell Filtered Target cell listlistlistlist

cell 2

Handover ManagementHandover ManagementHandover ManagementHandover Management

Internal Internal Internal Internal Handover Handover Handover Handover

External External External External Handover Handover Handover Handover

HandoverHandoverHandoverHandover protocolprotocolprotocolprotocol


1.333

Annex.4 Handover Management HO execution example

ordered1. Cell12. Cell23. Cell3


filtered1. Cell12. Cell23. Cell3

HO mgt

-

MSempty

BSC/MSCempty

Rej. lists

HO prot

HO failson cell1

ROC

update

Cell1 ->MS

rej list


filteredCell1

1. Cell22. Cell3

HO mgt

-

MScell1

BSC/MSCempty

Rej. lists

HO prot

HO failson cell2

ROC

updateCell2 ->

MSrej list

Cell1 ->BSC

rej list

update

T_MS_CELL_REJ

Expires

MS rej listempty


filteredCell1Cell2

1. Cell3

HO mgt

-

MScell2

BSC/MSCcell1

Rej. lists

HO prot

HO tocell3


1.334

▼End of Handover procedure = T_FILTER timer expiry

� T_FILTER restarted each time a target cell list is to be sent

by Candidate Cell Evaluation to the Handover Management

(same list than the one previously sent or not)

� The target cell list is sent to the Handover Management if

different from the last target cell list previously sent

� T_FILTER expiry means no handover is needed anymore

Annex.4 Handover Management T_FILTER controls HO procedure (1/2)


1.335

Annex.4 Handover Management T_FILTER controls HO procedure (2/2)

New candidate cell list from thecandidate cell evaluation function

IsT_FILTERrunning ?

Start T_FILTERAn HO alarm containing the candidatecell

list issent to the HO management entity

NoYes

Is the candidate celllist different fromthe previous one ?

Restart T_FILTERRestart T_FILTER

An HO alarm containing the candidatecelllist issent to the HO management entity

Yes No

No handover is on-going

A handover is on-going

A handover is NOW on-going

T_FILTER is restarted each time the alarm is still on

Back



ANNEXES

Annex.5 ellipsoid arc determination

Erlang B law





1.337

Annex.5 ellipsoid arc determinationPositioning methods : CI+TA positioning 2/2B8

θ

r1

Point (o)

r2

North

β

▼ ellipsoid arc definition:

� Point (O)= serving BTS site coordinate

� θ= serving cell antenna azimuth -β /2

� β =A*width of serving cell sector in [°], calculated from bisector angles of co-sited antenna azimuths

� r1= inner radius of TA ring-(B-0.5)*554 in [m]

� R2=(B+C)*554 in [m]

– A: ARC_SIZE_FACTOR– B: MIN_RADIUS_FACTOR– C: MAX_RADIUS_FACTOR Back

▼ An ellipsoid arc is a shape characterised by the co-ordinates of an ellipsoid point o (the origin), inner radius r1, uncertainty radius r2, both radii being geodesic distances over the surface of the ellipsoid, the offset angle (θ) between the first defining radius of the ellipsoid arc and North, and the included angle (β) being the angle between the first and second defining radii. The offset angle is within the range of 0° to 359,999…° while the included angle is within the range from 0,000…1° to 360°. This is to be able to describe a full circle, 0° to 360°

▼ For CI+TA method which is default one , the answer is given by description of "ellipsoid arc".

▼ Optimization parameters:� ARC_SIZE_FACTOR



computing location estimate based on TA positioning method

� MAX_RADIUS_FACTOR


Alcatel Optimization

Documents

double transcoding

ellipsoid

radio link

daily periodic

radio link

sudden traffic

radio link

optional additional