62012957 Lapd Overload

50
BSS (BSC) Traffic Handling Capacity, Network Planning and Overload Protection DN9812388 Issue 9-0 en 29/02/2008 # Nokia Siemens Networks 1 (50) BSC3153 Nokia GSM/EDGE BSS, Rel. BSS13, BSC and TCSM, Rel. S13, Product Documentation, v.1

Transcript of 62012957 Lapd Overload

Page 1: 62012957 Lapd Overload

BSS (BSC) Traffic HandlingCapacity, Network Planning andOverload Protection

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 1 (50)

BSC3153Nokia GSM/EDGE BSS, Rel. BSS13, BSC andTCSM, Rel. S13, Product Documentation, v.1

Page 2: 62012957 Lapd Overload

The information in this document is subject to change without notice and describes only theproduct defined in the introduction of this documentation. This documentation is intended for theuse of Nokia Siemens Networks customers only for the purposes of the agreement under whichthe document is submitted, and no part of it may be used, reproduced, modified or transmitted inany form or means without the prior written permission of Nokia Siemens Networks. Thedocumentation has been prepared to be used by professional and properly trained personnel,and the customer assumes full responsibility when using it. Nokia Siemens Networks welcomescustomer comments as part of the process of continuous development and improvement of thedocumentation.

The information or statements given in this documentation concerning the suitability, capacity, orperformance of the mentioned hardware or software products are given “as is” and all liabilityarising in connection with such hardware or software products shall be defined conclusively andfinally in a separate agreement between Nokia Siemens Networks and the customer. However,Nokia Siemens Networks has made all reasonable efforts to ensure that the instructionscontained in the document are adequate and free of material errors and omissions. NokiaSiemens Networks will, if deemed necessary by Nokia Siemens Networks, explain issues whichmay not be covered by the document.

Nokia Siemens Networks will correct errors in this documentation as soon as possible. IN NOEVENT WILL NOKIA SIEMENS NETWORKS BE LIABLE FOR ERRORS IN THISDOCUMENTATION OR FOR ANY DAMAGES, INCLUDING BUT NOT LIMITED TO SPECIAL,DIRECT, INDIRECT, INCIDENTAL OR CONSEQUENTIAL OR ANY LOSSES, SUCH AS BUTNOT LIMITED TO LOSS OF PROFIT, REVENUE, BUSINESS INTERRUPTION, BUSINESSOPPORTUNITY OR DATA, THAT MAYARISE FROM THE USE OF THIS DOCUMENT OR THEINFORMATION IN IT.

This documentation and the product it describes are considered protected by copyrights andother intellectual property rights according to the applicable laws.

The wave logo is a trademark of Nokia Siemens Networks Oy. Nokia is a registered trademark ofNokia Corporation. Siemens is a registered trademark of Siemens AG.

Other product names mentioned in this document may be trademarks of their respective owners,and they are mentioned for identification purposes only.

Copyright © Nokia Siemens Networks 2008. All rights reserved.

2 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 3: 62012957 Lapd Overload

Contents

Contents 3

List of tables 4

List of figures 5

Summary of changes 7

1 Planning the cellular network 9

2 BSC nominal capacity and dimensioning 13

3 Planning of basic GSM radio network parameters 193.1 Location Area Definition and CCCH Parameters 193.2 Abis LAPD link and CCS7 link dimensioning 26

4 BSS overload protection and flow control 294.1 BSS overload protection 294.2 System throughput versus offered load 344.3 Overload protection mechanisms 354.4 How to know when the system is running above the nominal load? 434.5 AS7 overload protection implementation 464.6 The LAPD counters used to check the LAPD load statistics 464.7 Indications on overload protection mechanisms 474.8 Timers 504.9 Parameters 50

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 3 (50)

Contents

Page 4: 62012957 Lapd Overload

List of tables

Table 1. Connectivity of logical PCUs 17

Table 2. Extreme case 1: theoretical maximum (maximum paging capacity in radiointerface, TMSI4 in use, for example 4 TMSIs in each paging) 20

Table 3. Extreme case 2: theoretical minimum (minimum paging capacity in radiointerface, IMSI1 in use only, for example 1 IMSI in each paging) 21

Table 4. Parameter values for the LA size for small cells 22

Table 5. Parameter values for the LA size for medium size cells 23

Table 6. Parameter values for the LA size for large cells 23

Table 7. Parameter values for the LA size for extra large cells 24

Table 8. LA with 2 cells Combined and 8 cells Noncombined 26

4 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 5: 62012957 Lapd Overload

List of figures

Figure 1. BSS block diagram 30

Figure 2. System throughput versus offered load 34

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 5 (50)

List of figures

Page 6: 62012957 Lapd Overload

6 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 7: 62012957 Lapd Overload

Summary of changes

Changes between document issues are cumulative. Therefore, the latestdocument issue contains all changes made to previous issues.

Changes made between issues 9-0 and 8-0

In chapter Planning of basic GSM radio network parameters, moreinformation on 32 kbit/s LAPD links and CCS7 links has been added tosection Abis LAPD link and CCS7 link dimensioning.

Changes made between issues 8-0 and 7-1

Modified the name of chapter BSS overload protection to BSS overloadprotection and flow control and included the information from Flow andOverload Control document into it. Structural changes in chapter BSCnominal capacity and dimensioning.

Updated the Traffic Handling Capacity information with data for the newBSC3i variants.

Updated the PCU Capacity and Overload Protection topic with changesdue to the introduction of (E)GPRS Inactivity alarm.

Changes made between issues 7-1 and 7-0

Editorial changes.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 7 (50)

Summary of changes

Page 8: 62012957 Lapd Overload

8 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 9: 62012957 Lapd Overload

1 Planning the cellular network

This is an overview of how to plan and configure the GSM/EDGE networkimplemented by the Nokia base station subsystem (BSS) in such a waythat the performance experienced by the end-users is not limited by anoverload in the BSS.

The general procedure of radio network planning is explained, with adetailed description of CCCH parameters. Note that the network planningexamples are just generic examples and as such they do not necessarilyrepresent the optimal solution for any particular network.

Commonly cellular network (BSS) planning has the following main targets:

. to achieve the required radio coverage with the maximum time andlocation probability (more than 90%)

. to maximise the network capacity (Erl/km2) with a limited frequencyband (MHz) by reusing frequencies

. to reach a good quality of service (QoS) with a minimum level ofinterference

. to minimise the number of network elements and transmissionneeded (MSC, BSC, BTS, ...) and therefore the cost of the networkinfrastructure

The described network planning process is an optimisation process whichneeds information about network elements, system properties (GSM,GPRS, EDGE), planning of the environment, topography of the servicearea, existing facilities of the operator, distribution of the subscribers, andthe estimated future growth of subscribers.

Network dimensioning is done so that the coverage and capacityrequirements based on subscriber growth forecast are fulfilled. After thenumber of network elements and commercial aspects is available, thebusiness plan can be made.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 9 (50)

Planning the cellular network

Page 10: 62012957 Lapd Overload

An important activity of network planning is the base station subsystem(BSS) parameter planning and radio network verification and optimisation.This is based on experience gained during the trial period of the network.

The general results and documents from network planning are:

. List of sites and network elements

. Coverage predictions and measurements

. Frequency plan and interference analysis

. Capacity calculations

. BSS parameter plan

. Transmission network plan

During the network's existence, knowledge about its operation andsubscriber behaviour will improve. The network will also expand in terms ofcapacity and coverage because of continuous subscriber growth.Therefore, network planning is also a continuous process which tries tominimise the modifications to the existing network while preparingextension plans. Testing and tuning are also needed at each new stage ofthe network to ensure good quality of service.

In the planning and optimisation phase we also need to verify that all thenetwork elements and the interfaces between them have enough capacityfor the given network configuration providing good quality for allsubscribers with their estimated reference call mix.

The following planning aspects should also be checked:

1. BSC erlang capacity load (for example the BSC traffic handling(erlang)) capacity must be large enough for the planned networkconfiguration and traffic load.

2. The optimised location area (LA) and routing area (RA) size and thecommon control channel (CCCH) structure (for example combined,noncombined CCCH) is defined for the given network in order tokeep the paging load in the required physical limits in the BSC AbisLAPD links (16 kbit/s, 32 kbit/s, or 64 kbit/s) and in the radiointerface.

10 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 11: 62012957 Lapd Overload

Both aspects 1 and 2 require, of course, a certain call mix (either thedefault one or a more precise, case-specific one) to be used as the basisfor planning. If the physical limits are exceeded in the given networkconfiguration, the original plan should be reviewed by checking, forexample, LA size, CCCH channel structure or BSC configuration (amount)in question.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 11 (50)

Planning the cellular network

Page 12: 62012957 Lapd Overload

12 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 13: 62012957 Lapd Overload

2 BSC nominal capacity and dimensioning

Described here are the traffic handling capacities of the Nokia GSM/EDGEBase Station Controllers, BSC2i and BSC3i, with certain average callmixes. Additionally, the whole BSS (BSC) overload protection is described.The overload protection has been implemented to protect the equipmentand the system in exceptionally high traffic cases.

In the given network the BSC erlang (traffic handling) capacity must bechecked so that the following nominal erlang capacity is not exceeded.This is the simplest case (that is, to check only erlang figures) in which thereference call mix can be used. In many cases the call mix in real networksdiffers from the nominal one. By a separate agreement Nokia SiemensNetworks can provide a separate estimate of the BSC capacity for the callmix in question. The BSC traffic handling capacity in the case of GPRSshould be noted separately.

BSC2i and BSC3i traffic handling capacity with the following circuit-switched reference call mix is stated in Product Description of NokiaBSC2i and BSCi High Capacity Base Station Controller and ProductDescription of Nokia BSC3i High Capacity Base Station Controller underProduct Descriptions in the PDF view.

In the example below, we assume that the BSC is defined to one locationarea.

Example BSC3i 2000 Traffic Handling Capactiy

BSC3i is configured with 2000 full rate TRXs, circuit-switched:

. 11880 erlangs total traffic handling capacity

. 25 mErl per subscriber, 475 200 subscribers

. 354 000 busy hour call attempts (BHCA)

. 120 seconds mean hold time

. 70% mobile-originated calls (MO)

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 13 (50)

BSC nominal capacity and dimensioning

Page 14: 62012957 Lapd Overload

. 30% mobile-terminated calls (MT)

. 1.5 handovers (HO) per call

. 2 location updates (LU) per call

. 0.1 IMSI detach per call

. 63% no paging response

. 1 SMS call rate subs/hour

. (80% terminated SMS)

In the circuit-switched (CS) case, for example, the nominal BSC pagingload for BSC3i 2000 would be (note that both MT calls and MT SMSscreate pages):

354 000 x 0.3 (MT) + 475 200 x 0.8 (SMS MT) = 486 360 + re-paging 0.63x 486 360 = 792 767 pagings per hour.

The nominal BSC3i 2000 RACH load (MO, MT, SMS, LUs) would be, forexample:

354 000 (MO, MT) + 2 x 354 000 (LU) + 0.1 x 354 000 (IMSI detach) + 486360 (SMS) = 1 583 760 RACHs per hour.

Example BSC3i 660 Traffic Handling Capacity

BSC3i is configured with 660 full rate TRXs, circuit-switched:

. 3920 erlangs total traffic handling capacity

. 25 mErl per subscriber, 157 000 subscribers

. 117 000 busy hour call attempts (BHCA)

. 120 seconds mean hold time

. 70% mobile-originated calls (MO)

. 30% mobile-terminated calls (MT)

. 1.5 handovers (HO) per call

. 2 location updates (LU) per call

. 0.1 IMSI detach per call

. 63% no paging response

14 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 15: 62012957 Lapd Overload

. 1 SMS call rate subs/hour

. (80% terminated SMS)

In the circuit-switched (CS) case, for example, the nominal BSC pagingload for BSC3i would be (note that both MT calls and MT SMSs createpages):

117 000 x 0.3 (MT) + 157 000 x 0.8 (SMS MT) = 160 700 + re-paging 0.63x 160 700 = 261 941 pagings per hour.

The nominal BSC3i RACH load (MO, MT, SMS, LUs) would, be forexample:

117 000 (MO, MT) + 2 x 117 000 (LU) + 0.1 x 117 000 (IMSI detach) + 160700 (SMS) = 523 400 RACHs per hour.

Example BSC2i Traffic Handling Capacity

BSC2i configured with 512 full rate TRXs, circuit-switched:

. 3040 erlangs total traffic handling capacity

. 25 mErl per subscriber, 120 000 subscribers

. 91 000 busy hour call attempts (BHCA)

. 120 seconds mean hold time

. 70% mobile-originated calls (MO)

. 30% mobile-terminated calls (MT)

. 1.5 handovers (HO) per call

. 2 location updates (LU) per call

. 0.1 IMSI detach per call

. 63% no paging response

. 1 SMS call rate subs/hour

. (80% terminated SMS)

In the circuit-switched (CS) case, for example, the nominal BSC pagingload for BSC2i would be (note that both MT calls and MT SMSs createpages):

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 15 (50)

BSC nominal capacity and dimensioning

Page 16: 62012957 Lapd Overload

91 000 x 0.3 (MT) + 120 000 x 0.8 (SMS MT) = 123 300 + re-paging 0.63 x123 300 = 200 979 pagings per hour, provided that all cells are paged inthat BSC and there is only one LA defined per BSC. Here the load is oneach Abis link of a broadcast control channel (BCCH) TRX on the sameLA.

The nominal BSC2i random access channel (RACH) load (MO, MT, SMS,LUs) would be, for example:

91 000 (MO, MT) + 2 x 91 000 (LU) + 0.1 x 91 000 (IMSI detach) + 123 300(SMS) = 405 400 RACHs per hour. This is the total load per BSC whichcan be divided by the number of cells when calculating the number ofRACHs per BCCH-TRX Abis.

With this reference call mix the BSC processor load still remains in thesafe area. The maximum 60% CPU load is the target for dimensioning.This gives enough margin for peak load situations as well as for newsoftware releases.

Some call mix is needed in order to get the main performance figures(erlangs, BHCAs) with maximum allowed processor load. With a differentcall mix, the BHCA value, for example, varies a lot because in a complexsystem such as GSM there are many other transactions, in addition tocalls, which load the system.

Roughly it can be said that the erlangs per air channel are the largestcontribution to the BSC processor load; the next largest ones are thenumber of call procedures (call set-up, clearing), SMSs and locationupdates (LUR, IMSI Attach/Detach are similar) and, lastly, all differenttypes of handovers. By saying that erlangs as such are significant wemean that there is a load in the system even though there is one call withindefinite length on each channel without having any HOs, LURs, and soon.

Packet-switched capacity

. BSC3i 2000: max. 100 PCU2 (50 physical PCU2 units per BSC, 2logical PCUs in one PCU2 plug-in unit)

. BSC3i 1000: max. 50 PCU2 (25 physical PCU units per BSC, 2logical PCUs in one PCU2 plug-in unit)

. BSC3i 660: max. 24 PCU1/PCU2 (12 physical PCU units per BSC, 2logical PCUs in one PCU1/PCU2 plug-in unit)

. BSC2i: max. 16 PCU1/PCU2 (16 physical PCU units per BSC, 1logical PCU in one PCU1/PCU2 plug-in unit)

16 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 17: 62012957 Lapd Overload

Table 1. Connectivity of logical PCUs

16kbit/s AbisTSL

TRX Cell/ Segments BTS

Logical PCU2

(PCU2-D/PCU2-U)

256 256 64 128

Logical PCU1

(PCU-B/PCU-T)

256 128 64 64

Logical PCU1

(PCU/PCU-S)

256

(128 RTSL)

128 64 64

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 17 (50)

BSC nominal capacity and dimensioning

Page 18: 62012957 Lapd Overload

18 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 19: 62012957 Lapd Overload

3 Planning of basic GSM radio networkparameters

The following sections provide some recommendations and practicalparameter values for network planning which help you to plan anddimension the system to perform in a nominal load area and to ensure thebest system performance. To illustrate the idea of calculations, theconfigurations are simplified to four configurations:

. 2 + 2 + 2

. 4 + 4 + 4

. 6 + 6 + 6

. 12 + 12 + 12

Note that all the theoretical examples presented here are very simplified.They are calculated just to give some ideas which parameters the networkplanners should take into account for example when dealing with thepaging process.

3.1 Location Area Definition and CCCH Parameters

Paging channel (PCH) signalling will be sent over the whole location area(LA). This means that one paging message over the A interface is 'copied'to all Abis links going to the common control channel (CCCH) TRX of cellsin the same location area. An optimal LA size is a balance between PCHload and location updates (LU). If the LA size is too large, paging channelsand capacity will be saturated because of limited LAPD Abis or radiointerface CCCH paging capacity. On the other hand, with large locationareas there will be a smaller number of location updates (LU) performedand vice versa.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 19 (50)

Planning of basic GSM radio network parameters

Page 20: 62012957 Lapd Overload

The same applies to paging coming via the Gs and Gb interfaces: the MSCsends the paging message to the SGSN with the LA info and the SGSNdefines it to a more accurate area: cell, routing area (RA), LA or BSS. Ifwithin the SGSN area there are cells that do not support GPRS services,the SGSN will group these cells under a 'null RA'. The SGSN will performthe paging procedure described above within both the RA(s) derived fromthe location information and the 'null RA'.

The number of CCCHs depends on the channel structure as follows:

. COMBINED: for a small cell, ≤ 2 TRXs/cell, 3 CCCHs in everysignalling multiframe (51 TDMA, 235 ms)

. NONCOMBINED: for a large cell, ≥ 3 TRXs/cell, 9 CCCHs in everysignalling multiframe (51 TDMA, 235 ms), used if GPRS is enabledin the cell.

Note that this is a kind of 'rule of thumb' of today, assuming not very heavySMS traffic.

The parameters that affect the CCCH capacity on a cell basis are thefollowing:

. Number of blocks reserved to AGCH (BS_AG_BLKS_RES);once this parameter is specified, the PCH is calculated; theparameter range is 0 to 7 and value zero is not recommended.

. number of multiframes (BS_PA_MFRMS); this specifies howmany multiframes will go until the given paging group is re-paged;the parameter range is 2 to 9 and the recommended value is 5.

The paging method is also set in MSC TMSI or IMSI. TMSI is morecommonly used, because of bigger capacity (4/page group). Here weassume that all the radio interface capacity is used, thus all extra pagingwill be ignored.

Below there are two extreme cases in terms of how high or low the pagingcapacity is over the radio interface. These examples are theoretical onesand the intention is to show the range of variation caused by differentCCCH parameterisations.

Table 2. Extreme case 1: theoretical maximum (maximum paging capacity inradio interface, TMSI4 in use, for example 4 TMSIs in each paging)

COMBINED NONCOMBINED

MAX (≤ 2TRXs/cell) (≥ 3 TRXs/cell)

20 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 21: 62012957 Lapd Overload

Table 2. Extreme case 1: theoretical maximum (maximum paging capacity inradio interface, TMSI4 in use, for example 4 TMSIs in each paging)(cont.)

total CCCH 3 9

PCH 2 8

AGCH 1 1

Pages per hour 122 553 490 212

Table 3. Extreme case 2: theoretical minimum (minimum paging capacity inradio interface, IMSI1 in use only, for example 1 IMSI in each paging)

COMBINED NONCOMBINED

MIN (≤2 TRXs/cell) (≥ 3 TRXs/cell)

total CCCH 3 9

PCH 1 2

AGCH 2 7

Pages per hour 15 319 30 638

The following gives some recommended parameters and LA sizes whichmatch the BSC nominal call model as specified in BSC Nominal Capacityand Dimensioning. The paging capacity is presented for four different cellconfigurations:

. 2 + 2 + 2

. 4 + 4 + 4

. 6 + 6 + 6

. 12 + 12 + 12

Example LA size for small cells (2 + 2 + 2 configuration)

In this example it is assumed that we have a configuration with 2 TRXs percell. If we use 2% blocking in the radio interface, we can see from theerlang (the unit of measure of carried traffic intensity) B-table that some 9erlangs will be served on a cell basis. This can be converted to some 360subscribers per cell (0.025 erlang per subscriber). If one site consists of 2+ 2 + 2 as a configuration, some 56 sites together will serve some 1 500erlang or 60 480 subscribers.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 21 (50)

Planning of basic GSM radio network parameters

Page 22: 62012957 Lapd Overload

The LA size for small cells (2 + 2 + 2 configuration) could then be 336TRXs.

Table 4. Parameter values for the LA size for small cells

Total number of subscribers 60 480 (in 168 cells, each 360 subs.)

TRXs in LA 336 (in 168 cells, each 2 TRX)

Cell configuration 2 + 2 + 2

CCCH channel structure COMBINED (for example small cell)

total CCCH 3

typical PCH 2

typical AGCH 1

number of multiframes 5 (does not affect PCH capacity, but MS batterylifetime)

Max. Pages per hour (in Air) theoretically 110 297 (TMSI4 80%, IMSI2 20%)

Pages per hour with BSC nominal call mix 29 829

Here we assume the BSC nominal call model with only 0.1 SMS call rateper subs/hour.

The parameter number of multiframes value is 5 here. It means thatthe same paging group will be re-paged after 5 x 235 ms = 1.175 sec. Thiswill ensure longer MS battery lifetime, because the MS has to listen quiteseldom to a CCCH channel in a serving cell. You must ensure that thepaging load does not exceed the physical limits in radio/Abis interfaces.These could be practical values provided that the SMS paging amount inthe BSC call model would be less - for example 0.1 SMS call ratesubscribers per hour, which would reduce the paging load.

Example LA size for medium size cells (4 + 4 + 4 configuration)

In this example it is assumed that we have a configuration with 4 TRXs percell. If we use 2% blocking in the radio interface, we can see from theerlang B-table that some 21.9 erlangs will be served on cell basis. This canbe converted to some 876 subscribers per cell (0.025 erlang persubscriber). If one site consists of 4 + 4 + 4 as a configuration, some 23sites together will serve some 1 500 erlang or 60 444 subscribers.

The LA size for medium size cells (4 + 4 + 4) configuration could then be276 TRXs.

22 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 23: 62012957 Lapd Overload

Table 5. Parameter values for the LA size for medium size cells

Total number of subscribers 60 444 (in 69 cells, each 876 subs.)

TRXs in LA 276 (in 69 cells, each 4 TRX)

Cell configuration 4 + 4 + 4

CCCH channel structure NONCOMBINED (for example large cell)

total CCCH 9

typical PCH 6

typical AGCH 3

number of multiframes 5 (does not affect PCH capacity, but MS batterylifetime)

Max. Pages per hour (in Air) 147 063 (TMSI2 60%, IMSI1 40%)

Pages per hour with BSC nominal call mix 29 829

Here we assume the BSC nominal call model with only 0.1 SMS call rateper subs/hour.

Example LA size for large cells (6 + 6 + 6 configuration)

In this example it is assumed that we have a configuration with 6 TRXs percell. If we use 2% blocking in the radio interface, we can see from theerlang B-table that some 34.6 erlangs will be served on cell basis. This canbe converted to 1384 subscribers per cell (0.025 erlang per subscriber). Ifone site consists of 6 + 6 + 6 as a configuration, some 43 cells together willserve some 1487 erlang or 59 512 subscribers.

The LA size for large cells (6 + 6 + 6 configuration) could then be 258TRXs.

Table 6. Parameter values for the LA size for large cells

Total number of subscribers 59 512 (43 cells, each 1384 subs.)

TRXs in LA 258 (43 cells, each 6 TRX)

Cell configuration 6 + 6 + 6

CCCH channel structure NONCOMBINED (for example large cell)

total CCCH 9

typical PCH 8

typical AGCH 1

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 23 (50)

Planning of basic GSM radio network parameters

Page 24: 62012957 Lapd Overload

Table 6. Parameter values for the LA size for large cells (cont.)

number of multiframes 5 (does not affect PCH capacity, but MS batterylifetime)

Max. Pages per hour (in Air) theoretically 490 212 (TMSI4)

Pages per hour with BSC nominal call mix 29 586

Example LA size for extra large cells (12 + 12 + 12 configuration)

In this example it is assumed that we have a configuration with 12 TRXsper cell. If we use 2% blocking in the radio interface, we can see from theerlang B-table that some 77.3 erlangs will be served on cell basis. This canbe converted to some 3092 subscribers per cell (0.025 erlang persubscriber). If one site consists of 12 + 12 + 12 as a configuration, some19 cells together will serve some 1469 erlang or 58 748 subscribers.

The LA size for extra large cells (12 + 12 + 12 configuration) could then be228 TRXs.

Table 7. Parameter values for the LA size for extra large cells

Total number of subscribers 58 748 (in 19 cells, each 3092 subs.)

TRXs in LA 228 (in 19 cells, each 12 TRX)

Cell configuration 12 + 12 + 12

CCCH channel structure NONCOMBINED (for example large cell)

total CCCH 9

typical PCH 8

typical AGCH 1

number of multiframes 5 (does not effect PCH capacity, but MS batterylifetime)

Max. Pages per hour (in Air) theoretically 490 212 (TMSI4)

Pages per hour with BSC nominal call mix 29 207

The parameter number of multiframes value is 5 here. It means thatthe same paging group will be re-paged after 5 x 235 ms = 1.175 sec. Thiswill ensure longer MS battery lifetime, because the MS has to listen quiteseldom to a CCCH channel in a serving cell. In this case you must also

24 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 25: 62012957 Lapd Overload

ensure from the estimated call mix or from live network statistics andmeasurement values that you operate in the nominal BSC load area andthat the Abis paging load does not exceed the limits of LAPD (16 kbit/s, 32kbit/s, or 64 kbit/s) link capacity nor the radio interface paging capacity.

The recommendation concerning MSC paging parameters (see The LAPDcounters used to check the LAPD load statistics) is to use the 'LA' pagingmethod (LAC or LAI), which prevents the unnecessary cell level CIinformation from being sent to the BSC via the A interface. If the CIinformation is included, it is sent for each cell in the LA. Paging via the Ainterface is always performed on LA level. Inclusion of the CI informationdoes not provide any benefit and loads the MSC, BSC and the A interfaceunnecessarily.

In the MSC there are also parameters related to CCCH (actually PCH)capacity, which are on a LA basis. To ensure that the paging messagereaches the MS, the paging message is sent several times. The repetitionprocedure is defined in the MSC. These MSC parameters areRepaging_Interval (time between paging attempts) andNumber_of_Repaging_Attempts, which can be modified in the (Nokia)MSC.

The recommended values are: Number_of_Repaging_Attempts = 0,Repaging_Interval = 3.5s. This works better if TMSI is in use. Thismeans that the first paging goes with TMSI, and then after 3.5 secondswith IMSI, if the subscriber does not respond to TMSI.

The conclusion is that paging load is highly dependent on parameters. Inthe same LA, the paging load should be monitored. Note that if there isonly one small cell in a given LA, where combined channel structure is inuse, this will be the bottleneck if paging blocking criteria are strictlyfollowed. In other words: the smallest cell in the LA will set the PCH limit.Note also that some maximum configurations would not be possiblebecause of other limiting factors such as the 16 kbit/s Abis or radiointerface, which would start to limit the message traffic, thus it would beuseless to define such parameter settings (for example too large locationarea size).

If there are only one or two cells with combined channel structure in an LA,you can choose to live with a high paging blocking rate in this cell becausethe probability of MS location in this cell is very low. Therefore, the Pagingblocking rate as seen from the MSC is not modified much by too few PCHson this cell.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 25 (50)

Planning of basic GSM radio network parameters

Page 26: 62012957 Lapd Overload

Table 8. LA with 2 cells Combined and 8 cells Noncombined

Blocking rate MS locationprobability

Cell 1 30% 2%

Cell 2 30% 2%

Cell 3 1% 6%

Cell 4 1% 10%

Cell 5 1% 10%

Cell 6 1% 10%

Cell 7 1% 10%

Cell 8 1% 10%

Cell 9 1% 10%

Cell 10 1% 10%

The final Blocking rate is 30 x 4/100 + (1 x 96/100) = 2.16%.

Moreover, if the MSC repeats the Paging messages, the end user blockingrate can be considerably reduced if the PCH is not overloaded too much:10% x 10% ≈ 1%.

3.2 Abis LAPD link and CCS7 link dimensioning

Abis LAPD

You should also take into account the following factors when estimatingthe capacity of the Abis LAPD signalling, especially the 16 kbit/s link:

. There can be a maximum of 2 signalling messages unacknowledgedat any time, all the subsequent messages must wait for theacknowledgement of at least one of these messages; all themessages are stored in an AS7 buffer until responded to by theopposite end. The LAPD window size 2 is recommended by 3GPPTS 48.056.

. The acknowledgement delay varies from a few milliseconds to tensof milliseconds because of the characteristics of the LAPD protocol,especially if there is a lot of signalling traffic coming in from theopposite direction (measurement reports and so on). All this lowersthe maximum capacity much below 16 kbit/s.

26 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 27: 62012957 Lapd Overload

. Disturbances on the physical 2 Mbit/s line may cause more delays,which lowers the capacity.

. The average AS7 transmit buffer occupancy should be close to 0 orat most, for optimal use of the link capacity, just a few messages (sothat there is always one message waiting for transmission), that is,the buffer is used mainly for temporary storage of the transmittedmessages waiting for acknowledgement and for occasional bursts ofmessages. If, instead of this, we assumed a high average bufferoccupancy, it would also mean that signalling messages wouldgenerally experience long delays while they wait for transmission inthe buffer. Generally, the transmit buffer size of the AS7 and itsoccupancy level need not be considered in dimensioning, as thecapacity of the buffer is sufficient to handle any burst of messagesthat is still within the capacity of the Abis signalling link andmaximum delays considered.

. Based on these previous factors and measurements made on theAbis link, the maximum average signalling traffic load should notexceed 8 kbit/s (1000 bytes/sec). There is a risk of AS7 overflow ifthe load is more than 1200 bytes/sec.

One of the most common messages sent on the highest loaded Abis link(that is, the BCCH TRX link) is the paging message.

The length of the paging message (including FCS and flags) is about 21bytes. According to the BSC nominal load and call mix, about 60% of allcapacity can be given to the paging messages; the average pagingmessage count/sec/link is thus 0.6 x maximum signalling traffic load / 21. =29, which roughly equals 100 000 pages per hour (16 kbit/s). This issufficient, for example, for the nominal BSC call model.

The same general principles apply for 32 kbit/s and 64 kbit/s links. Themaximum recommended average signalling traffic is 2000 bytes/sec (32kbit/s) or 4000 bytes/sec (64 kbit/s), which roughly equals to 200 000 (32kbit/s) and 400 000 (64 kbit/s) pages per hour.

The number of paging messages is different depending on the call mix andconfiguration. If the reference nominal call mix is not suitable, the limits tobe considered are 1000 bytes/s (16 kbit/s LAPD), 2000 bytes/s (32 kbit/s),and 4000 bytes/s (64 kbit/s LAPD) per Abis link, already mentioned above.

Note that these values are for individual links only and they should not beused to estimate the total or even the BCSU capacity without taking otherdimensioning rules into account.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 27 (50)

Planning of basic GSM radio network parameters

Page 28: 62012957 Lapd Overload

CCS7

Extended SS7 signalling capacity is needed especially for high SMSusage, short calls, and signalling needs of a high BSC TRX capacity. TheBSC3i 2000 offers connectivity and capacity with 2Mbit/s, 512 kbit/s, 256kbit/s, 128 kbit/s, and 64 kbit/s links as well as SIGTRAN links. SIGTRANallows flexible capacity allocation which is not tied to a fixed bandwidthsuch as 64 kbit/s or 2 Mbit/s, for example.

When estimating the need of signalling links, it is recommended that onesignalling link load should not overrun 0.2 Erl. Additionally, redundancyshould be considered in the case of failure in one of the links. Additionallinks should be provided so that the overall SS7 signalling traffic maintainsnormal operations during the failure of a link. In satellite links, the signallinglink load should be under 0.06 erlang.

The operator should routinely monitor growing traffic and link conditions toensure appropriate dimensioning.

28 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 29: 62012957 Lapd Overload

4 BSS overload protection and flowcontrol

The main purpose of overload protection and flow control is to avoid thenetwork becoming heavily loaded, which in the worst case scenario couldlead to a total collapse of service in a network element. Overload situationscan lead to, for instance, call clearing or reset procedures.

In the case of an overload, overload protection and flow control aims tomaintain the service level of established calls as normal as possible. Thisalso pertains to handover attempts caused by the radio environment.

The principle of flow control is to set certain restrictions for new callattempts within the network. In the GSM system, flow control can beperformed either in the radio interface or inside the network. Flow control inthe radio interface is performed by limiting access from mobile stations(MSs) according to specified rules. If required, new call attempts may beabandoned in the network, nevertheless allowing for priorisation, as, forexample, in the case of emergency calls.

4.1 BSS overload protection

The main building blocks of the BSS system, presented below, help inunderstanding the BSS traffic dimensioning and BSS performing inoverload situations. Figure BSS block diagram illustrates the componentsthat are relevant from the BSS capacity point of view.

There are two types of blocks: channels or buses with a certain fixedbandwidth, and processors. There are several different types ofprocessors in the system:

. digital signal processors (DSP)

. BTS TRX controllers

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 29 (50)

BSS overload protection and flow control

Page 30: 62012957 Lapd Overload

. BSC main CPUs (BCSU, MCMU)

. BSC pre-processors (AS7)

Figure 1. BSS block diagram

. The Radio interface channel has a certain maximum capacity asspecified by GSM specifications. From the capacity planning point ofview, there are several different channel types with differentthroughputs to be selected: common control channel (CCCH) and acombined SDCCH/4 & CCCH.

Within the CCCH, there is the parameter Number of blocksreserved to Access Grant to adjust CCCH downlink partition.There are also some other CCCH-related parameters, namelypaging related: BS_PA_MFRMS and random access-related (RACH)TX_INTEGER and MAX_RETRANS which have an indirect impact onchannel capacity.

TRXBTS

Airinterfacechannel

BTS

CCCH,TCH,SDCCH

AbisLAPDlink

A interface#7 link

Gb interface

BCSUBSC

Message bus

CPU

MCMU

TRX

DSP

CPU

BCSU

AS7,LAPD

AS7,#7

PCU

30 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 31: 62012957 Lapd Overload

. BTS DSP is a dedicated block signal processing power allocatedpermanently for one radio time slot. The actual number of radio timeslots per a single DSP chip varies between BTS generations. Theprinciple from the loading analysis point of view is the same: theDSP runs in a loop to execute tasks related to a radio time slot(RTSL) and if it can handle all the frames coming from the radiointerface. The processing power has to be there for the 'worst case'that a single RTSL provides and there is no need to think about callmixes, and so on.

. BTS TRX represents here the processing power reserved to carryTRX-level call control-related tasks such as LAPD termination, BTScall control protocol, DSP interface, and so forth. The actualprocessor varies in different BTS generations, but in all of them thereis a single processor dedicated to a single TRX. At the BTS TRXlevel, the loading of the processor depends on traffic andconfiguration.

The biggest load is caused by such channel configurations in whichthe actual number of logical channels is the highest. There is alimitation of a certain number of SDCCHs per TRX. For detailedinformation, see Dynamic SDCCH in Radio Channel Allocationunder Functional Descriptions/Radio network performance in thePDF view. TRX processing is dimensioned – according to the worstcase – to handle all the traffic coming from Air/DSP, and no otherattention needs to be paid to it when dimensioning the network.

The traffic dimensioning can be done just by taking the TRXs asseparate and isolated blocks, that is, the high load on one TRXcannot harm the other one within the same BTS. (Frequencyhopping is an exception of sharing some common processing powerover sectors/BTSs. Possible limitations coming from frequencyhopping are hopping configuration-related, for example the numberof hopping groups, and they do not affect call handling capabilities.)

. Abis LAPD link is on a TRX basis and it can be configured to one ofthe following speeds:. 16 kbit/s. 32 kbit/s. 64 kbit/s

The link can be a bottleneck because both the BSC and BTSprocessors can offer more messages than it can transfer in somecases. Roughly we can say that a 16 kbit/s link can handle all full-rate configurations of a TRX if the paging load is reasonable. A 32kbit/s link is needed with half rate when the number of logicalchannels per TRX increases. 64 kbit/s exists for historical reasonsas the first, non-optimised LAPD link implementation.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 31 (50)

BSS overload protection and flow control

Page 32: 62012957 Lapd Overload

. AS7 LAPD is the BSC's preprocessor for performing the LAPDprotocol layer 2 level tasks such as link establishment on the Abisinterface, error detection, message segmentation, andretransmissions. AS7 handles LAPD links. The links can be eithertelecom (TRX) links or O&M (BCF) links. The number of linkshandled by one processor depends on the plug-in unit variant in use.

The basic assumption on dimensioning has been that AS7 canhandle all the traffic that both the BSC/BCSU CPU and Abis LAPDlinks can in practice offer to it. This is true when we talk about calls,HOs and so on (that is, the bottleneck is not the AS7 but the CPU).Because of the AS7 unit, we have to pay some attention to thepaging load in LAPD links. Paging comes through the BSC CPU,requiring very small processing, and it is also very easy to generatea high paging load by radio network planning. That is why it ispossible in real networks to run the AS7 to overload protection state.To avoid this high paging load we have to follow certainrecommendations as far as the maximum paging load is concerned(see details in Planning of basic GSM radio network parameters).

A preprocessing unit AS7 is equipped to the BCSU for both LAPDand SS7 signalling. The SS7 interface is for the A interface. Itcontains a preprocessor, which is capable of handling a maximum offour signalling channels. The regular bit rate in SS7 links is 64 kbit/s.Additionally, wider 128 kbit/s, 256 kbit/s, 512 kbit/s or 2 Mbit/s linkscan be used. The signalling terminal is semipermanently connectedto the time slots used for signalling.

The basic assumption on dimensioning has been that the A interfaceAS7 can handle all the traffic that both the BSC/BCSU CPU and theMSC's A interface #7 links can in practice offer. Additionally, the #7links are normally redundant. The paging load, however, may alsobe a bottleneck for the #7 link and its performance.

BCSU, CPU

The BSC Signalling Unit (BCSU) performs those BSC functions that arehighly dependent on the volume of traffic. It consists of two partscorresponding to the A and Abis interfaces.

In the BSC design phase, the BCSU was thought to be the heaviestloaded functional unit of the BSC even though the MCMU is centralisedand the BCSU distributed. A very large number of radio measurementscoming from the BTS/MS are processed by the BCSU.

This assumption was confirmed by performance testing and it has alsobeen the case in live networks.

32 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 33: 62012957 Lapd Overload

That is why the main focus is to keep the BCSU CPU load in the safe areawhen loading the BSC. However, normally the BCSU CPU load is not thelimiting factor and even in high traffic cases the CPUs have sufficientoverload protection.

Packet-switched traffic has only a marginal effect on BCSU load, since thePCU handles it.

MCMU, CPU

The Marker and Cellular Management Unit (MCMU) performs the controlfunctions of a switching matrix and the BSC-specific managementfunctions of the radio resources.

The cellular management functions of the MCMU assume responsibility forcells and radio channels that are controlled by the BSC. This responsibilityis centralised in the MCMU. The MCMU reserves and keeps track of theradio resources requested by the MSC or the handover procedures of theBSC. Thus the BSC's MCMU load might rise in high traffic cases, but theCPU has been protected against exceptionally high traffic cases.

Message Bus (MB)

A duplicated high-speed Message Bus (MB) is used for data transferbetween the OMU and the call control computers of the BSC3i 660.

The length of each message is determined individually by a messagelength parameter at the beginning of the message. The sender and thereceiver of the message are indicated in the address field of the message.The receiver can be a single microcomputer, or it can be a group ofmicrocomputers specified by the broadcast address.

The hardware of the Message Bus consists of several parallel twistedpairs, which carry the actual data and also control the information requiredfor the message transfer. In the event of a failure, the hot standby MessageBus takes over the functions of the active bus without interfering with theongoing calls.

BSC3i 1000/2000 configurations use new Ethernet-based Message Bus(EMB). Today the message bus capacity is not a bottleneck in the BSCeven in very high traffic cases, mainly because of the high bus speed andalso because the length of the bus is very short – the BSC has only amaximum of two racks, which enables high speed and fast responsetimes.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 33 (50)

BSS overload protection and flow control

Page 34: 62012957 Lapd Overload

Packet Control Unit (PCU)

The Packet Control Unit (PCU) is used to control the (E)GPRS radioresources. Its procedures include radio resource allocation andmanagement, connection establishment and management, scheduling,data transfer and Gb load sharing and flow control. Reliability is achievedby N+1 redundancy.

The PCU is protected against exceptionally high traffic cases by ignoringthe overflowing data. The data will not be lost since other networkelements will resend it in the acknowledged mode. There are also othersoftware-specific overload protection mechanisms.

With the full PCU configuration, you can share the BSC-controlled BTSdomain for multiple PCUs. One BTS is, however, always controlled by onePCU. This means that the packet-switched traffic load can be sharedamong BCSUs. The equal sharing of the BTSs is also efficient from theO&M point of view. This is because no BTS switchovers from one PCU toanother are needed when the network is growing.

4.2 System throughput versus offered load

Figure 2. System throughput versus offered load

Curve 1) in the figure above represents the system design goal; thesystem will remain stable even though it is loaded above the 100%possible load (here 100% means typically a CPU with full processorusage). In a complex multiprocessor/multitask system, it is quitechallenging to keep the system stable at 100% load.

SYSTEMTHROUGHPUT

100 % LOAD

NOMINALLOAD

OFFEREDLOAD

1)

2)

34 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 35: 62012957 Lapd Overload

Curve 2) describes the situation in which the system is quite sensitive togoing above the nominal load used as a design criterion. Normally thismeans that the processor(s) go to 'reset' state and do not start up properlybecause internal buffers will overflow in an uncontrolled manner. Aprocessor might also close all the tasks it is handling and/or it may blockthe channels it is handling. In this case, the system is protected from goingunstable, but the performance goes down.

The challenge in maintaining the system on curve 1) is mainly to detect theemerging overload situation and to throw away tasks/messages in acontrolled way at the 'border' of a processor. This requires carefulconsideration of all the possible interfaces from where messages cancome in to a processor and to analyse which are the initial/first messagesto trigger actions inside the processor.

In Nokia BSS, mechanisms have been implemented to keep the systemon curve 1) shown in Figure System throughput versus offered load. Thesemechanisms cover all the major components of the system (Figure BSSblock diagram in BSS overload protection). These mechanisms aredescribed in Overload protection mechanisms.

4.3 Overload protection mechanisms

Overload cases in BSS can be divided into the following two types:

. External overload: overload in the network as a whole

. Internal overload: BSC's own processor overload.

External overload

Overload in the network is indicated by messages sent from one networkelement to another. The message transfer channels can also becongested.

In the BSC, external overload is manifested in the following cases:

. overload at MSC, indicated by OVERLOAD messages

. overload at BTS, indicated by OVERLOAD and LOAD INDICATIONmessages

. MTP congestion

. LAPD overload.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 35 (50)

BSS overload protection and flow control

Page 36: 62012957 Lapd Overload

Overload at MSC

Overload at the MSC is reduced by a stepwise procedure. When the BSCreceives an OVERLOAD message from the MSC, traffic is reduced stepby step by barring mobile access classes. The number of access classesthat are barred during one step is specified with a UTPFIL parameter.Possible values include 1, 2 and 5; as a default two access classes arebarred at a time. The access classes are updated by sending a newsystem information message via the Abis interface to the MSs.

The traffic reduction procedure is supervised by two timers, T17 and T18,which are started when the OVERLOAD message is received. Possibleother incoming OVERLOAD messages are ignored until T17 expires. If anOVERLOAD message is received after the expiry of T17 while T18, whichis always set further than T17, is still running, traffic is reduced by onemore step and both timers are restarted. The reduction procedure isrepeated as long as there are incoming OVERLOAD messages, and stepsleft in the procedure to reduce traffic. The same step-by-step process isalso used to increase traffic, if no OVERLOAD messages are receivedbefore the expiry of T18.

OVERLOAD messages come from the MSC as connectionless SCCPmessages. The SCCP sends them to the BSC and the hand of the BSCsets the timers T17 and T18. For more details on the timers, see PAFILETimer and Parameter List under Reference/Parameters in the PDF viewand 3GPP TS 48.008 Mobile-services Switching Centre – Base StationSystem (MSC – BSS) interface; Layer 3 specification).

The BSC sends an access class updating message to all distributedmasters. This message contains the new access class octets for theRACH control parameter information element. A master distributes themessage to its TRX handling hands that update the system information ofall the BTSs. If the new RACH control has a user-set value different fromthe previous one, the master combines both values so that the user-setvalue remains valid. As a result, there will not be fewer restrictions thanwith the predefined value.

The BSC controls the access class barring procedure by choosing adifferent class every time: the barring is done in a logical order, taking thepreviously barred access class(es) into consideration. Similarly, barring isremoved in the same order as it was set, so each access class shouldhave equally long barring. This applies to the access classes from AC C00to C09. The user can bar other classes with the EQF MML command.

36 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 37: 62012957 Lapd Overload

The BSC sets the alarm 2478 MOBILE ACCESS CLASS ABNORMALwhen mobile access classes are restricted by the system, and,correspondingly, cancels the alarm when the access classes are notrestricted by the system.

For more information on handling overload at MSC when Multipoint Ainterface application software is in use, see Multipoint A Interface in BSCunder Feature descriptions/Radio network performance in the PDF view.

Overload at BTS

The BSC receives CCCH LOAD INDICATION messages from the BTS ona regular basis, which indicate the load situation on paging (PCH) andrandom access (RACH) channels. The CCCH LOAD INDICATIONmessage contains either load on an RACH or load on a PCH if BTS level isphase 2 or GPRS PH 1 (General Packet Radio Service Phase 1). TheCCCH LOAD INDICATION message is used mainly for statistics: itsreception increments the Resource Access Measurement counters.

Every few seconds, the BTS sends information to the BSC about theresources and buffer space available for PAGING messages coming fromthe BSC. When the PAGING message buffer is full, the BSC sends anOVERLOAD message to the MSC in order to reduce paging in a particularcell. Because the BSC does not have the means to reduce paging traffic ina sophisticated way, the BTS discards the PAGING COMMANDmessagesthat it cannot send forward to the MSs. The deletion of a PAGINGCOMMAND message causes the counter 003038 AVE PCH LOAD ONCCCH to be incremented.

An access grant channel (AGCH) message overflow is indicated by thereception of a DELETE INDICATION message from the BTS. When aDELETE INDICATION message is received, counter 003005 DELETE INDMESSAGES RECEIVED FROM BTS is incremented.

The CCCH LOAD INDICATION and DELETE INDICATION messages arereceived by the CCCH hand of the BSCU. The CCCH hand alsoincrements the counters.

For more details on counters related to BTS overload, see 3 ResourceAccess Measurement.

MTP congestion

The procedure used in a congested signalling system is similar to the onedescribed in Overload at MSC except that here the indication of thesituation comes from the SCCP that sends MTP congestion messages tothe BSC.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 37 (50)

BSS overload protection and flow control

Page 38: 62012957 Lapd Overload

LAPD overload

LAPD overload may occur in both uplink and downlink direction. In uplinkdirection, measurement reports are transferred from the BTS (and MS) tothe BSC. The control of this overload is a matter of network configurationand the BTS.

Overload may occur in downlink direction because of heavy paging andother call signalling procedures. The transmit buffers of the LAPD linkshandling the Abis signalling may overflow especially with the 16 kbit/s Abislinks. BCSU and PCU therefore mark the PAGING and IMMEDIATEASSIGNMENT REJECT messages to enable them to be discarded duringvery high load on the LAPD signalling link. This ensures that normal callsignalling, for example handovers, proceeds. If the congestion level of thetransmit buffers still becomes very high, all signalling messages on thechannel(s) causing the congestion may be discarded.

Internal overload

In the BSC, there are centralised and distributed functions:

. centralised functions are situated in the Marker and CellularManagement Unit (MCMU)

. distributed functions are situated in several BSC Signalling Units(BCSU) and in Packet Control Units (PCU).

Radio resource management and administration are an example ofcentralised functions and interface signalling procedures an example ofdistributed functions.

Internal processor overload control mechanisms have been implementedexclusively in the distributed unit. See Centralised unit overload for details.

Centralised unit overload

Overload control of the centralised unit is handled by the distributed unit.

When the centralised unit handles the circuit switched random accessesslowly, congestion of the CCCH hand can also occur. The CCCH channelhand process has a certain number of places for the pending randomaccesses. If the response from the MCMU takes a long time and the queuefor the pending random accesses is full, the BCSU discards new randomaccesses coming from MSs. For every pending random access there isalso a timer that supervises the MCMU response. In the case of a timerexpiry, the random access is removed from the queue and the place isreleased for a new one.

38 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 39: 62012957 Lapd Overload

1. BSC/BCSU/CPU protection against excessive number of pagingmessages coming from the A interface

The PAGING messages sent by the MSC can cause a heavyprocessor load. The load from the A interface can be increasedfurther because of large amount of paging repetitions.

This type of overload situation takes place, for example, when all theA interface links have been down for a while. That has caused theBSC to put all the cells to 'cell barred' status. When the A interfacebecomes available, the radio network will still remain 'closed' for awhile before the BSC has updated all the cell barring away and allthe access classes have been opened for traffic.

The paging load from the A interface is high and there are a lot ofrepetitions of paging. To protect the BCSU from this type of overload,the A interface application software constantly checks the number ofunhandled messages in the BCSU's CPU before distributing thepaging message to the other BCSUs, and if the load limit isexceeded, paging messages are omitted (paging represents thebiggest unpredictable message load here, because traffic from theradio network side is starting smoothly because of phased openingof access classes).

For every omitted paging message, counter 500627 AIV PAGINGREFUSED BIG LOAD is pegged. See 50 BSC Level Clear CodeMeasurement for details.

After the distribution of the paging message, the Abis interfaceapplication software again makes the same checks with a lower loadlimit, and if it is exceeded, the paging message is omitted. For everyomitted paging message the counter PAGING FROM A INTERFACEREFUSED DUE BCSUX BIG LOAD is pegged. The x stands for themessage bus address of the BCSU (30 - 38). The correspondingnumbers for the counters are 51177 to 51185.

If a predefined threshold of omitted paging messages has beenexceeded in the BCSU, the BSC sends an Overload message to theMSC (see 3GPP TS 48.008). At the arrival of this message, the MSCdecreases the amount of paging by a certain percentage.

The BSC sets the alarm 1302 PAGING OVERLOAD when the totalnumber of omitted paging messages exceeds a certain threshold inthe BCSU.

2. BSC/BCSU/CPU general protection

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 39 (50)

BSS overload protection and flow control

Page 40: 62012957 Lapd Overload

It is essential to know that about half of the processing power is usedto the handling of radio measurement results, the other half to theactual call control signalling. Radio measurements can be sacrificedfirst when traffic increases. This makes it possible to have a quitelarge area of the capacity buffer between the nominal load and 100%level.

The BCSU is protected against running out of message bufferswhich normally causes the BCSU restart. Radio measurementshedding starts when a certain limit of unhandled messages in theBCSU's CPU has been exceeded.

The BSC's handover and power control algorithm supervises theflow of radio measurements based on the running number in them.Radio measurement reports can also be lost because the LAPD linkis overloaded and the BTSs have to throw some of them away.

This supervision is useful, for example, when introducing half rateinto the network in order to know when it is time to upgrade the CPU/LAPD link speed. The BSC sets an alarm on when an overloadsituation occurs. The detection of the overload is based on thethreshold ratio of the rejected measurement results to allmeasurement results. If the threshold is exceeded, the BCSU unit/LAPD link is considered as being overloaded.

3. BSC BCSU CPU overload against high random access channel(RACH) load also provides protection for BSC/MCMU/CPU

A typical example of this type of overload is a special event which iscausing a large number of subscribers in an area to try to call at thesame time. In this case there are many access attempts to thesystem. Many of the access attempts will face congestion in a stand-alone dedicated control channel (SDCCH) and traffic channel (TCH),which in turn leads users to try again and again. Note that this kind ofbehaviour is exceptional compared to the average busy hour. Thecase requires special protection.

Each time a new CHANNEL_REQUIRED message comes from theBTS through the Abis interface, a software application in the BCSUchecks the number of unhandled messages in the BCSU's CPU.Two limits are defined. When the lower limit is exceeded, the causesmoc_data, location_update, other_cases, and re_establishment willbe omitted and counter 003039 BCSU OVERLOAD LOWER LIMITis incremented.

For more details on the counters, see 3 Resource AccessMeasurement.

40 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 41: 62012957 Lapd Overload

The procedure for handling IMMEDIATE_ASSIGNMENT_REJECTmessages can be considered to belong to this category. Themessage is sent to the MS in two cases:

a. no free SDCCH

b. the Abis interface process has no internal resources to handlethe request

Inside the message there is the field 'Wait indication', which definesthe wait period for the MS. You can set the value of the timer by theMMI from 2 s to 10 s. The default value is 10 s.

4. AS7 LAPD protection

When the send buffer on a plug-in unit is full, the messages aredeleted until the situation has been corrected. The situation may becaused by an overload or a failure in the plug-in unit. The alarm isacknowledged when there is a certain number of space in the buffer.See the details of the alarm 2133 in Indications on overloadprotection mechanisms.

5. BSC/BCSU protection against excessive number of pagingmessages coming from the Gb interface

The BCSU has an overload control to protect itself against theprocessor overloading and the TRXSIG link overloading. The BCSUcuts down the load by rejecting particular messages when theprocessor load or the link load exceeds the defined load limit.Application software rejects messages which are sent in thedownlink direction to the TRXSIG, if needed. Each message that issent to the TRXSIG is given a certain message group value. In casethe message buffers of an AS7 are going to fill up, the applicationsoftware starts to discard messages based on the message groupvalue.

Application software in the BCSU cuts down the load caused bypaging messages sent by the SGSN. The load control is based onthe number of unhandled messages in the BCSU's message queue.The SGSN sends paging messages to the PCUs via the Gbinterface and the PCUs forward them to the host BCSU. Applicationsoftware checks the count of unhandled messages in the BCSU'smessage queue every time a new paging message is received afterbroadcasting the message to the BCSUs. If the load limit isexceeded, the message in the separate BCSUs is deleted.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 41 (50)

BSS overload protection and flow control

Page 42: 62012957 Lapd Overload

For every omitted paging message, a counter is pegged. For everydeleted CS paging message, the counter CS PAGING REFUSEDDUE BCSUX BIG LOAD (counters 51168 to 51176) is pegged andfor every deleted PS paging message the counter PS PAGINGREFUSED DUE BCSUX BIG LOAD (counters 51159 to 51167) ispegged. The x stands for the message bus address of the BCSU (30- 38).

The BSC will set the alarm 1302 PAGING OVERLOAD when thetotal number of omitted paging messages exceeds a certainthreshold in the BCSU unit.

6. BSC/BCSU protection against high PS RACH load

In the uplink direction BCSU software cuts down the load caused byPCUs. The software rejects P_CHANNEL_REQUIRED messagesreceived from the TRXSIG if the processor load exceeds the definedload limit. The load control of the software is based on the number ofunhandled messages in the BCSU's message queue. Softwarechecks the count of unhandled messages in the BCSU's messagequeue every time a new P_CHANNEL_REQUIRED message isreceived.

7. Overload due to excessive number of radio measurement reports

Most of the messages received by the BCSU are radiomeasurement reports coming from the BTS and the MS. Thesemessages come from the LAPD link. Before the distribution, the loadstate of the BCSU is checked:. If the load exceeds a certain predefined limit, the message is

discarded. The message buffer and processing capacitydemand in the unit is thus decreased. The load limit amountsto the number of unhandled messages in the unit.

. If the load does not exceed the limit, messages are distributedand handled normally.

The loss of the radio measurement report does not affect the servicequality significantly; some reports are lost anyway because of theload on the LAPD link. The method of discarding messages is alsorandom, so the loss of messages for one particular connection stayswithin reasonable limits.

The frames incoming from the LAPD link cause an interruption forthe BCSU processor. The BCSU reads the frames from the memoryof the preprocessor AS7-U, and distributes them to other units thathandle message information. The frames are distributed accordingto the frame type and the channel or TRX number given in the frameinformation.

42 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 43: 62012957 Lapd Overload

Because incoming UI frames only carry measurement reportinformation, these frames acquire special treatment. Every time a UIframe is received, the BCSU reads the number of free messagebuffers. If this number is below the predefined limit, a message withthe frame information is sent forward. On the other hand, if this limitis exceeded, the BCSU discards the UI frame. This method enablesan extremely fast response to overload conditions, provided that theparameter limit for discarding UI frames has been set correctly. Thesignalling in CM/MM/RR layers is thus not affected by an overloadsituation in the BCSU.

8. PCU overload protection

The PCU protects itself from a processor overload with its owninternal overload protection mechanism.

The operator can enable a notice (0125 PCU PROCESSOR LOADHIGH), which is set by PCU if the processor load is approacing theoverload levels. This is only an early warning of PCU processor loadbeing high and no protective actions are taken at this stage. Thisnotice does not require cancellation and is suppressed as set in thedefault options.

When the PCU processor load becomes exceptionally high, thePCU raises an alarm (3164 PCU PROCESSOR OVERLOAD**). Atthis stage the PCU overload protection actions are initiated. ThePCU first starts to reduce load by limiting the load caused byfunctionalities such as Network-Controlled Cell Re-selection, NCCR(if activated). In the first phase, the accuracy of the NCCR algorithmis reduced by ignoring every second measurement report sent byMSs. If the load still rises, the PCU starts to reduce the schedulingrate proportionally as the load rises, which limits the total throughputon PCU level and, if NCCR is activated, new MSs are not acceptedto network-controlled cell-reselection, but the MSs are graduallymoved to autonomous cell re-selection. If the load still continues torise, the scheduling rate is further reduced. In extreme load cases,the PCU starts to tear down connections which do not meet thequality targets defined for their traffic class.

4.4 How to know when the system is running above thenominal load?

The following may be seen as indicators of the system running above thenominal load, depending on the actual overload case.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 43 (50)

BSS overload protection and flow control

Page 44: 62012957 Lapd Overload

Utilisation of call processing CPUs in the BSC (MCMU, BCSU) is onthe average above 60%

This can be verified from BSC performance measurement counters.

For more information, see 6 Load Measurement under Reference/Counter/performance indicators/Call control measurements (CS) in the PDF view.

Paging load is above 100 000 pages/h (16 kbit/s LAPD link) or above410 000 pages/h (64 kbit/s LAPD link)

You can verify this from the Resource Access Measurement counter003000 PAGING MESSAGES SENT TO BTS. This BSC counter tells howmany pagings have been sent via one LAPD link (and location area, LA).As paging messages are received on the Abis interface, the counter isincreased. Here one message corresponds to one MS to be paged, whichmeans no packing of several IMSI/TMSIs to the same paging message. Tofind out the paging load over BTSs at the same location area, it is enoughto check one BTS per LA.

In 4.6 The LAPD counters used to check the LAPD load statistics there is adescription of layer 2 level counters on Abis LAPD utilisation. Themeasurement shows all the bytes sent in LAPD paging and othermessages.

Note

This measurement is actually a more accurate source of Abis link orAS7 overload detection. This is because it measures all the data overthe link.

However, we recommend the paging counters described above as themain pointer of high load for two reasons:

. The paging load is the component of system load which will increaseindirectly (that is, you add load to all cells at the same LA when anew subscriber or a new cell comes to the network).

. BSC resource access measurements are more user-friendly tofollow.

Degradation of end-user-experienced network quality

Decreased quality is also shown by key performance indicators:

44 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 45: 62012957 Lapd Overload

. congestion rate

. paging success rate

. call success rate

Handover success rate may also indicate equipment overload. However,in most cases the counters do not indicate problems until the system isoverloaded. Also, the paging problems are not as well visible as the mainindicators.

The paging success rate can be calculated from MSC counters as follows:

PAGING SUCCESS RATE, only for Nokia MSC, M056: Paging Success %(M7)

The following formula indicates the success rate of paging (call & SMS):

100 x sum(succ_page)/sum(fail_page + succ_page)

Overload protection mechanism(s) is continuously triggered

In The LAPD counters used to check the LAPD load statistics there arepointers for all the overload mechanisms that are described earlier. If theissues that are mentioned earlier (CPU load, paging load, networkperformance) are on the safe side, it is improbable that any overloadmechanism triggers. In high traffic BSCs, you should follow the alarm 2720TELECOM LINK OVERLOAD for BCSU/CPU general protection orstatistics of BCSU/CPU RACH load protection.

The alarm 2133 SEND BUFFER OVERFLOW IN SIGNALLINGTERMINAL also indicates that the Abis LAPD interface capacity isoverloaded.

See the following for details:

. CPU load counters: 6 Load Measurement

. paging, RACH counters and measurements: 3 Resource AccessMeasurement

. TCH/SDCCH counters and measurements: 1 Traffic Measurement

under Reference/Counter/performance indicators/Call controlmeasurements (CS) in the PDF view.

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 45 (50)

BSS overload protection and flow control

Page 46: 62012957 Lapd Overload

4.5 AS7 overload protection implementation

An overload protection mechanism has been implemented to the BSCadditionally to protect the equipment from exceptionally high messagetraffic, for example because of high peak paging load.

The transmit buffers of the AS7 handling the Abis signalling links mayoverflow during excessive signalling traffic load, especially with the 16 kbit/s Abis links.

To avoid the overflow, a mechanism has been implemented for discardingmessages of less significance (requiring no immediate handling oracknowledgment) when AS7 detects congestion on the transmit buffers ofthe Abis signalling link. IMMEDIATE_ASSIGNMENT_REJECT andPAGING messages are 'labelled' so that they may be discarded duringvery high load on the LAPD signalling link to ensure that normal callsignalling proceeds.

Additionally, if the congestion level of the transmit buffer still becomes veryhigh, all the signalling messages on the channel(s) causing the congestionmay be discarded.

The storage space for incoming channel requests is also decreased toavoid an overload situation in the BSC.

4.6 The LAPD counters used to check the LAPD loadstatistics

Abis LAPD layer 2 counters can be checked in measurement periods ofhalf an hour.

Use the DMFMML command. The following command, for example, showsall counters:

ZDMF:P,,P:A,P:D:;

The observed counter is number 5 TRANSMITTED TOTAL OCTETCOUNT. If it exceeds in a half-hour period more than 1 800 000 (in case of16 kbit/s) or 7 000 000 (in case of 64 kbit/s), the LAPD overload problemmost probably then already exists and the alarm 2133 is activated.

The safe values could be as follows:

46 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 47: 62012957 Lapd Overload

. 1 000 000 for 16 kbit/s

. 3 500 000 for 64 kbit/s

An example printout of the counters from the BSC looks as follows:

LOADING PROGRAM VERSION 2.6-0 DX 200 BSC03

1997-11-16 14:55:59 TRANSMITTED AND RECEIVED FRAMES AND OCTETS PRIMARY

RATE LINK SET ACTIVE UNIT METERS OF LAST PERIOD: 14:00:00 - 14:30:00

1 2 3 4 5 D-CHA TYPE

6 7 8 9 10 ===== ==== ========== ==========

========== ========== ========== 0 PRI 0000000059 0000000000 0000001829

0000000000 0000005425 0000000059 0000000000 0000001829 0000000000

0000005425 1 PRI 0000000059 0000000000 0000001829 0000000000 0000005421

0000000059 0000000000 0000001829 0000000000 0000005421 2 PRI

0000000000 0000000000 0000000000 0000000000 0000000720 0000000000

0000000000 0000000000 0000000000 0000000720 3 PRI 0000013170 0000000000

0000248086 0000000000 0000264586 0000004125 0000010900 0000066502

0000428455 0000547637 MEASUREMENT NAMES 1 ... TRANSMITTED I FRAMES 2

... TRANSMITTED UI FRAMES 3 ... TRANSMITTED I FRAME OCTETS 4 ... TRANSMITTED

UI FRAME OCTETS 5 ... TRANSMITTED TOTAL OCTET COUNT 6 ... RECEIVED I FRAMES

7 ... RECEIVED UI FRAMES 8 ... RECEIVED I FRAME OCTETS 9 ... RECEIVED

UI FRAME OCTETS 10 ... RECEIVED TOTAL OCTET COUNT COMMAND EXECUTED

4.7 Indications on overload protection mechanisms

1. BSC/BCSU/CPU protection against excessive number of pagingmessages coming from the A interface. See counter 500627 in 50 BSC Level Clear Code

Measurement under Reference/Counter/performanceindicators/Call control measurements (CS) in the PDF view ondeleted paging messages.

. See counters 51177-51185 in 51 BSC Level Clear Code (PM)Measurement under Reference/Counter/performanceindicators/Call control measurements (CS) in the PDF view.

. Alarm 1302 PAGING OVERLOAD (disturbance)

Meaning: The BSC sends an overload message to the MSC inoverload situations inside the BSC. At the arrival of thismessage, the MSC decreases the amount of paging by acertain percentage. That helps the BSC to cope with theunusually high number of paging messages.

The BSC sets the alarm when a predefined threshold ofdiscarded paging messages has been exceeded.

For more information, see Disturbance Printouts (1000-1999)under Reference/Alarms in the PDF view.

2. BSC/BCSU/CPU general protection

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 47 (50)

BSS overload protection and flow control

Page 48: 62012957 Lapd Overload

Alarm 2720 TELECOM LINK OVERLOAD

Meaning: The ratio between the number of rejected measurementresults and the number of all measurement results exceeds theacceptable limit set by the user. The alarm is used to supervise thetraffic load of LAPD links and BCSU units and to detect the possibleoverload situations.

SUPPLEMENTARY INFORMATION FIELDS

a. Unit. 00: BCSU unit overloaded. 01: LAPD link overloaded

b. BTS identification of the LAPD link...

c. etc.

For more information, see Failure Printouts (2000-3999).

3. BSC BCSU CPU overload against high RACH load

See the following counters in 3 Resource Access Measurementunder Reference/Counter/performance indicators/Call controlmeasurements (CS) in the PDF view:. 003039 BCSU overload lower limit. 003040 BCSU overload upper limit. 003041 BCSU overload deleted rach

4. BSC/MCMU/CPU protection

MCMU CPU load is high, the ratio of channel requests and givenchannels slightly decreases.

5. AS7 LAPD protection

Alarm 2133 SEND BUFFER OVERFLOW IN SIGNALLINGTERMINAL

Meaning: The send buffer on a plug-in unit is full. Messages aredeleted until the situation has been corrected. The situation may becaused by an overload or a failure in the plug-in unit. The alarm isacknowledged when there is a certain amount of space in the buffer.

See also Failure Printouts (2000-3999).

6. GSM-specified overload protection implemented by Nokia

No statistics available.

7. Message overflow at BTS

48 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection

Page 49: 62012957 Lapd Overload

ACCESS GRANT CHANNEL BLOCKING can be calculated byusing BSC measurements. The BSC sends 'immediate assignment'or 'immediate assignment rejected' commands to the BTS. If the AGbuffer in the BTS is full, it responds with a delete indication. Thus theratio of delete indications to the sum of imm.assign. and imm.assgn.rej. describes the blocking of AGCH.

100 x sum(del_ind_msg_rec)/ sum(imm_assgn_rej+imm_assgn_sent)

all counters from p_nbsc_res_acc

8. BSC/BCSU protection against excessive number of pagingmessages coming from the Gb interface

See counters 51159-51176 in 51 BSC Level Clear Code (PM)Measurement under Reference/Counter/performance indicators/Callcontrol measurements (CS) in the PDF view.

9. PCU processor load indicators. Alarm 0125 PCU PROCESSOR LOAD HIGH (notice)

Meaning: This notice indicates that the PCU processor load isapproaching the overload level, but no overload protectionactions are initiated. This notification provides an early warningof a probable need for additional PCU capacity. This notice issuppressed by default, that is, the operator must enable it. Thenotice identifies the PCU plug-in unit in which the high loadhas occured. In PCU2, the notice also identifies with aprocessor index whether the situation has occured on mainprocessor PQII (0xFF) or on one of the DSPs (0x0 .. 0x7).

See also Notices (0-999).. Alarm 3164 PCU PROCESSOR OVERLOAD (**)

Meaning: This alarm indicates that the PCU processor loadhas risen to the level at which the overload protection actionsare initiated. If the load still continues to rise, further steps toreduce the load are taken by reducing the scheduling rategradually according to the PCU processor load. Lowering thescheduling rate lowers the processor load at the expense ofdata throughput. Other software-specific overload protectionactions are also taken. For example, the PCU starts to moveMSs from network-controlled cell re-selection to autonomouscell re-selection, if NCCR is activated. The alarm identifies thePCU plug-in unit in which the overload situation has occured.In PCU2, the alarm also identifies with a processor indexwhether the overload situation has occured on the mainprocessor PQII (0xFF) or on one of the DSPs (0x0 .. 0x7).

DN9812388Issue 9-0 en29/02/2008

# Nokia Siemens Networks 49 (50)

BSS overload protection and flow control

Page 50: 62012957 Lapd Overload

4.8 Timers

The following timers are related to Flow and Overload Control. For moredetails on the timers, see PAFILE Timer and Parameter List underReference/Parameters in the PDF view.

. T17 Overload procedure

. T18 Overload procedure

4.9 Parameters

The following parameter is related to Flow and Overload Control:

. LAPD load threshold (LAPDL)

For more details, see BSS Radio Network Parameter Dictionary underReference/Parameters in the PDF view.

See also Failure Printouts (2000 - 3999) under Reference/Alarms in thePDF view.

50 (50) # Nokia Siemens Networks DN9812388Issue 9-0 en29/02/2008

BSS (BSC) Traffic Handling Capacity, Network Planning andOverload Protection