Cell ReselectionFirst step in resource allocation procedure is the cell selection (and reselection inmobility). The cell selection and reselection is based on C1, C2, C31/C32, NCCR andNACC.ETSI define three network control order parameters, which determine themeasurement reporting and network control on the MS.•NC0: MS controlled cell reselection, no measurement reporting;•NC1: MS controlled cell reselection, MS sends measurement reports;•NC2: Network controlled cell reselection, MS sends measurement.6.1.1 C1 and C2The Nokia BSS supports network control order NC0, and therefore there are nonetwork controlled handovers in GPRS, the cell is selected autonomously by themobile using the existing path loss criteria C1 and cell reselection parameter C2.The network broadcasts on the BCCH the Modified system info 3 and System info 13parameters related to mobility management, which the (E)GPRS mobiles utilize toensure that they are camped on the cell offering best service in each area (thePBCCH functionality will be described later on in this document). The process for thispurpose is called Cell Selection and is based on C1 and C2 comparison.The MS calculates the value of C1 and C2 for the serving cell and will re-calculate C1and C2 values for the neighbouring cells every 5 seconds. The MS will then checkwhether:•The path loss criterion (C1) for current serving cell falls below zero for a periodof 5 seconds. This indicates that the path loss to the cell has become too high.•The calculated value of C2 for a non-serving suitable cell exceeds the value ofC2 for the serving cell for a period of 5 seconds.If, however, in the case of the new cell being in a different location area or, for aGPRS MS, in a different routing area or always for a GPRS MS in ready state, the C2value for the new cell shall exceed the C2 value of the serving cell by at leastCELL_RESELECT_HYSTERESIS dB.The idea is that the MS compares field strength levels of different cells defined in theidle mode BA list and selects the most appropriate using the C1 criteria:C1 =(a-Max(B,0))A= received level Average – p1B= p2-maximum RF Power of the Mobile Stationp1= Rxlevel access min(gprsRxLevAccessMin)p2= MS TXPower MAX CCH(gprsMsTxpwrMaxCCH)All values are expressed in dBm. POWER_OFFSET is not used

The C2 parameter can be utilized together with the C1 parameter to provide theoperator with greater traffic management capability. The C2 parameter wasintroduced in GSM phase two and designed for use in layered-architecture networks(micro/macro cell/Dual Band).The C2 feature brings associated parameters that are related to microcellularplanning.•penaltyTime(20 ... 640 s) describes the time delay before the finalcomparison is made between two cells.•

temporaryOffset(0 ... 70 dB) describes how much field strength could havebeen dropped during this penalty time,•

cellReselectOffset(0 ... 126 dB) describes an offset to cell reselection. C2cell reselection is calculated by equation

C2 = C1 + cellReselectOffset - temporaryOffset x H(penaltyTime-T) whenpenaltyTime < 640OrC2 = C1 - cellReselectOffset when penaltyTime=640

C31/C32The C31/C32 parameters will give the possibility to optimize the cell reselection for(E)GPRS without affecting the circuit switched cell reselection behavior. This willallow more flexible use of cell resources, allowing, for example, some cells to bepacket free if this is the intention.The C31/32 functionality will only be applicable if the PBCCH is allocated, otherwisethe circuit switch signaling channels will be used and consequently C1 and C2.In a multi-vendor environment one requirement is that all the vendors should supportbroadcasting of the C31/C32 parameters.C31 parameterSignal strength threshold criterion (C31) for hierarchical cell structures (HCS) is usedto decide whether the cell is qualified for prioritized hierarchical cell selection.C31(s) = RLA(s) -hcsThreshold (s)(serving cell)C31(n) = RLA(n) – hcsThreshold (n) -TO(n)’ L(n) (neighbor cell)WhereHCS_THR = signal threshold for applying HCS reselectionTO(n) =gprsTemporaryOffset (n)’ H(gprsPenaltyTime(n) – T(n))L(n) = 0, if hcsPriorityClass (n) = PRIORITY_CLASS(s)1, if hcsPriorityClass(n) ‡hcsPriorityClass(s)H(x) = 0, if x < 01, if x >= 0gprsTemporaryOffset applies a negative offset to C31/C32 for the duration of gprsPenaltyTimeafter the timer T has started for that cell.T is a timer implemented for each cell in the list of strongest carriers. T shall bestarted from zero at the time the cell is placed by the MS on the list of strongestcarriers, except when the previous serving cell is placed on the list of strongestcarriers at cell reselection. In this case, T shall be set to the value of PENALTY_TIME(i.e. expired).C32 parameterThe cell ranking criterion (C32) is used to select cells among those with the samepriorityC32(s) = C1(s) (serving cell)C32(n) = C1(n) +gprsReselectOffset(n) –TO(n)’ (1 – L(n))(neighbour cell)WheregprsReselectOffset applies an offset and hysteresis value to each cell.TO and L as in C31.gprsReselectOffset applies an offset and hysteresis value to each cell.The MS must select the cell having the highest C32 value among those that have thehighest priority class among those that fulfill the criterion C31 >= 0. The priorityclasses may correspond to different HCS layers. If no cells fulfill the C31>=0 criterion,the MS must select the cell having the highest C32 value.If PBCCH is not allocated to the cell, criterions C1 and C2 are used as they are usedin current CSW services

 .

 MS cell reselection algorithm

MS makes cell reselection if path loss criterion (C1) for the serving cell falls belowzero.MS can make cell reselection also when it founds a non-serving suitable cell betterthan the serving cell.

The best cell is the cell with the highest value of C32 among thecells with the highest PRIORITY_CLASS and fulfill the criterion C31 >= 0 or all cells ifthere no cell fulfilling the C31 >= 0 criterion.

Ifc32Qual parameter is set, positive gprsReselectOffset values shall only beapplied to the neighbour cell with the highest RLA_P value of those cells for whichC32

is compared above.

When the MS is inready state

, time defined bygprsCellReselHysteresis 

value issubtracted from C32 value for neighbour cells. If parameterc31Hysteresis 

is set theGPRS_CELL_RESELECT_HYSTERESIS is subtracted also from C31 neighbourcells. When the new cell is from different routing area

raReselectHysteresis parameter value is subtracted from C32 for neighbour cells. In case of a cellreselection occurred within

the previous 15 seconds, 5 dBs are subtracted from C32for neighbour cells.

Abnormal Cell reselection

Whenever the MS receives PACKET UL ACK/NACK (Packet Ack/Nack is PAN) thatallows the advancement of data transmit, the mobile station shall increment

N3102 bythe broadcast value

PAN_INC , however N3102 shall never exceed the value

PAN_MAX . Each time T3182 expires the mobile station shall decrement N3102 bythe broadcast value

PAN_DEC . When N3102 <= 0 is reached, the mobile stationshall perform an abnormal release with cell reselection.

BSC parameters:•

An other reason for abnormal cell reselection is MS not being able to readPSI1 in 60 sec. (both in packet idle or in packet transfer mode)

•Abnormal cell reselection will happen if

randomAccessRetry = 1•

Going back to the original cell is prohibited by parametertResel sec if anothersuitable cell is available.

More information about cell reselection parameters and its optimization can be foundin (E)GPRS Radio Networks – Optimization Guidelines

Network Controlled Cell ReselectionTarget cell to which the cell reselection is done, can be selected by the MS itself or bythe network.In earlier releases Nokia implemented only MS controlled cell reselection withoutmeasurement reports,

which is basically commanded by Network Control Order 0(NC0). In NC0, cell reselection is controlled by MS alone in both MM Ready and MMStandby states whether MS is in Packet Idle Mode or Packet

Transfer Mode. Whenthere is NCCR in the network cell reselection for MSs in MM Ready state arecontrolled by the network. When MSs go back to MM Stand By state, cell reselection

is done by MS as in NC0. NCCR support is indicated by setting the Network ControlOrder to NC2.PBCCH is not needed for NCCR.NCCR can be enabled for Release97 mobiles onward.Handover procedure, where cell resources are reserved in the target cell beforeordering MS cell change is not provided for packet switched services in 3GPP release 4

NCCR BenefitsBenefits that S11.5 Network Controlled Cell Reselection introduces:o

Efficient allocation of EGPRS resources. Some operators introduce EGPRSTRXs gradually in GSM networks. Some cells have EGPRS TRXs and somewill not. EGPRS resources will be scarce and will

need to be allocatedefficiently. PCU will push EGPRS capable MSs to EGPRS cells and GPRScapable MSs to non-EGPRS capable cells by power budget NCCR criterion.Cell attractiveness can be defined neighbour cell specifically also taking intoaccount each neighbour cell’s capacities (e.g. CS-3/CS-4 or

EQoS support).o

Quality criterion allows NCCR when the serving cell quality drops even if thesignal level is good.o

Quality Control may trigger NCCR. It means that EQoS can trigger NCCR tomake cell selection.o

Service based NCCR is possible (SGSN UTRAN CCO BSSGP procedure)o

Possibility to select WCDMA network as soon as it is available or when GSMcoverage ends, depending on operator choice.NCCR is an optional feature. Operator can set the feature on/off on BSC level,

anddecide whether NCCR to WCDMA FDD cells is allowed.NCCR is a standard feature for MS and SGSN. However, there may be MSs, whichdo not support NCCR and the PCU has to be prepared for that. PCU will monitor onlyMSs, which send neighbour cell measurement reports. Further there is a

possibility toswitch the NCCR off on 3GPP release basis (Release 97, 99, 04).6.1.3.2 NCCR FunctionalityThe operator has to set cell adjacencies, NCCR algorithm parameters, NetworkControl Order (NCO) mode and MS reporting period parameters.Cell adjacencies and NCO mode

are broadcast to MS. Depending on the operatorparameter the MS may be commanded to send neighbour cell measurements bybroadcasting the command to all MSs or by commanding individual MSs during TBF.Once commanded to report neighbour cell measurements MS will send neighbour cellmeasurements to PCU in a frequency defined by the reporting period parameters. MSsends neighbour cell measurements in MM

ready state, i.e. RR packet transfer andpacket idle modes.PCU sets MS NCCR context for each MS, which has been commanded, to NCOwhen first TBF is set for such MS or when first measurement report from such MS isreceived. PCU performs averaging for the measurements and applies NCCRalgorithm to averaged measurements. The NCCR algorithm is based on operator setthreshold values, so when certain

threshold triggers NCCR is started.

S11.5 includes following NCCR criteria:•

Power budget NCCR(NCCR EGPRS PBGT margin, NCCR GPRS PBGT margin, NCCR streamingTBF offset, NCCR other PCU cell offset)

•Quality Controlled NCCR

•Coverage reason ISNCCRThe later BSS releases will introduce the following NCCR types:

•EQoS Quality ControlWhen serving cell cannot provide the guaranteed throughput or thetransmission

quality is below operator set threshold, NCCR may be tried tooffer better service.The Quality Control (QC) NCCR triggering is described in EQoS planningmaterials. The radio link quality based NCCR is required even irrespective ofS11.5 EQoS feature implementation. Target cell selection is performed whenthe QC

NCCR trigger comes and always when new PACKET (ENHANCED)MEASUREMENT REPORT message is received until:

•MS NCCR context is deleted

•TBF is released, or

•QC cancels the NCCR trigger.

•Service based ISNCCR

6.1.3.3 Target cell selectionThe target cell evaluation is based on an RXLEV threshold algorithm, depicted inFigure 43. In this figure it is shown the algorithm that the BSC would apply for anEGPRS MS.

For GPRS MSs the algorithm is the same except for the Rx level margincomparison, which is:AV_RXLEV_NCELL(N) > AV_RXLEV_SERV + CellReselMarginQualforGPRSMS(n)for GPRS MS.Due to

separate thresholds for EGPRS capable and non-capable MSs, this criterioncannot be used before the MS EGPRS capability is known

Signaling FlowThe signaling flow of NCCR can be seen below

BLER Limits are Needed for the Quality Control Function in PCU2The maximum block error rate (BLER) limit is set with different parameters in PCU1and PCU2.For PCU1:

• 

MAXIMUM BLER IN ACKNOWLEDGED MODE (BLA)• 

MAXIMUM BLER IN UNACKNOWLEDGED MODE (BLU)For PCU2:

• 

PFC ACK BLER LIMIT FOR TRANSFER DELAY 1 (ABL1)• 

PFC UNACK BLER LIMIT FOR SDU ERROR RATIO 1 (UBL1)The EQoS specific packet flow context (PFC) feature is not applicable with BSC SWrelease S11.5 and PCU2 Only the ABL1 and UBL1 parameters are used, although allABL1-5 and UBL 1-6 parameters are

visible. All BLER limit parameters are visibleregardless of the EQoS feature’s state.In Quality Control function the above corresponding parameters are used similarly inboth PCU1 and PCU2.The BLER

parameter values are not directly comparablethough, so they are not converted in the upgradeBTS Selection

The Common BCCH/Multi BCF features are bringing the new Segment (SEG)concept into planning. The affect on GPRS is that there may be more than one BTSunder one segment, which supports GPRS. Therefore at the TBF establishment andalso later for reallocations of the TBF (if necessary) a BTS

selection procedure will beutilized.For example there is one Talk family BTS, which supports GPRS, and EGPRScapable BTS, which can also support GPRS, under the same segment. Also theoperating

frequency of the BTSs under one segment can be different for example theBTS which carries BCCH/PBCCH operates in 900 MHz and the other BTS(s)operates in 1800 MHz.The main principle of BTS selection is primarily to allocate GPRS TBF to a GPRSBTS and EGPRS TBF to an EGPRS BTS.Multi BCF feature can be used in single band environment and this feature will allowalso having only one

BCCH in the segment. For dual band solution common BCCHhas to be applied. Common BCCH is an optional feature while multi BCF is standard.

6.2.1 Initial BTS SelectionInitial BTS from SEG is selected in CHM (Channel Management) when new TBF iscreated.The main

steps of initial BTS selection is listed below:•

BTSs supporting the frequency bands, which are indicated in Radio AccessCapability (RAC) of the (E) GPRS MS, are selected.RAC may not be known at the time of TBF initiation (RAC information isdelivered to the SGSN during the GPRS attach). Therefore for a DL TBF theSGSN most probably has the RAC of the (E)GPRS MS. It is more likely thatfor the UL TBF, the RAC is not known at the establishment. (For more on RACrefer to 3GPP 04.60). If the RAC is not known then BTSs supporting the samefrequency

band as BCCH BTS are selected. Therefore there must be GPRSterritory in the BCCH band.•

The signal level must be good enough on the selected BTS:- If RX_level (C-value) is known then the BTS selected for allocation hasto satisfy the following: RX_Lev - BTS's non_bcch_layer_offset

>BTS's GPRS_non_BCCH_layer_rxlev_upper_limit.-

If RX_level is not known then Direct access BTS is selected.Direct_GPRS_access_threshold parameter is used to compare BTSobjects relative preference: when the value ofDirect_GPRS_access_threshold

parameter is higher than the value ofthe parameter non BCCH layer offset then the BTS is valid forallocation.At the TBF establishment phase there may not be any Rx-lev measurementresults yet

because of that RX-Lev criteria cannot be used. TBF is initiallyallocated to a BTS whereDirectGPRSAccessBTS

is set on (in practice itmeans thatDirectGPRSaccessBTS

>nonBCCHlayerOffset)*

.•

An EGPRS capable BTS will be selected for a GPRS TBF only if:- The segment doesn’t have GPRS capable BTS, or-

TBF/TSL > MaxTBFinTSLin every TSL in every GPRS capable BTS(i.e. the GPRS territory is totally full)

ANDaverage

TBF/TSL <MaxTBFinTSLin every EGPRS capable BTS.

•If there are several possible BTSs then the BTS in segment with minimumdownlink TBF/TSL QoS load is

selected.• 

If there is no possible BTS then BTS is not selected and TBF is not created. 

*DirectGPRSaccessBTSconcept has basically been developed for selecting theappropriate BTS at the initial BTS selection. It is

possible to indicate the preferredBTSs for allocation if the Rx_lev is not known at the TBF establishment. PreferredGPRS BTS

 has

nonBCCHlayerOffset  

parameter set so that it is smaller than 

DirectGPRSaccessBTS 

.This parameter

(nonBCCHlayerOffset) 

 indicates coveragearea of a BTS in the segment compared to the BCCH BTS.

 The smaller the value is,the closer the coverage of the BTS is to the BCCH BTS.

DirectGPRSaccessBTS  indicates the risk that can be allowed when allocating a non BCCH BTS in case of noRx_Lev

measurement.So if theDirectGPRSaccessBTS 

= 2in this segment this means that BTSs with 2dB less coverage than the BCCH BTS to be allocated as an

initial BTS for a TBF. Andif compare the 

nonBCCHlayerOffset with this parameter it is checked whether theBTS fulfils this requirement. Then list the BTSs that are

appropriate for initial BTSselection, of course BCCH BTS is also included. Operator may want to direct all theGPRS traffic to the non BCCH BTS(s). In order to do that the

GPRSenabledTRX parameter on all the TRXs in the BCCH BTS must be set to

OFF.However it ispossible that the RAC of the MS is unknown during initial TBF selection. In that casethe non-

BCCH BTS must be from the same band as BCCH band.Rx-Lev Measurements

are used as BTS selection criteria, when available;The DL Rx_Lev measurements (C_value)

are sent to the PCU in the DLACK/NACK messages. The receiving frequency of these measurements depends onthe polling frequency of the TBF. Actually the Rx-lev reported is averaged by the MSduring the

polling period.The UL Rx- Lev measurement

is included to each PCU frame by the BTS. Thereceiving frequency of these measurements depends on how often the uplink TBFgets a transmission turn. These measurements are averaged in PCU and the averageis used by the allocation algorithm.For more information on measurements done by MS please

check 3GPP 05.08. 

The following figure shows the block diagram of initial BTS selection.

Figure 45 Initial BTS selectionNote1

: If MS RAC band is not known then BCCH band BTSs are selected Note2

: BTSs that don’t have PSW territory or channels for PSW are not selected.Note3

: UL: BTSs whose average TBF/TSL is not less thanMaximumNumberOfULTBF are not selectedNote4

: DL: BTSs whose average TBF/TSL is not less thanMaximumNumberOfDLTBF are not selectedBTS Selection for Reallocating TBF

TBF reallocation processing can take place if better quality data transfer is expected(related to Rx_level) or BTS packet traffic load is unevenly spread in a segment orbetween supported frequency bands in a

segment.Procedures used to check TBF reallocation activation need in Channel Management(CHM) are in order below. Reallocation request to MAC shall be activated in any ofthe procedures at once (1-

4).1. BTS Load reallocation2. Uplink Rx level reallocation3. Downlink Rx level reallocation4. Downlink RX level received first time reallocationPeriodical checks are done every time

TBF_LOAD_GUARD_THRSHLD amount ofblock periods are used by the TBF after the last reallocation.Periodic reallocation check in PCU2 is tied to the amount of transmitted data, not totime as in PCU1.PCU2 does periodic reallocation check for all non-streaming TBFs to check andreallocate if there are better resources that could be allocated to an MS. Reallocationcheck is triggered on transmission of

TBF_LOAD_GUARD_THRSHLD RLC datablocks to the MS.PCU1 triggers periodic reallocation check after TBF_LOAD_GUARD_THRSHLD RLCblock periods. With TBF_LOAD_GUARD_THRSHLD

parameter’s default value,PCU1 does periodic reallocation check for an MS once in second.For example, if there are heavy traffic and an MS get transmission turns not so often,time between periodic reallocation

checks for the MS in PCU2 is longer. A certainamount of data blocks are transmitted before PCU2 triggers periodic reallocationcheck.If the result of checking is that reallocation is needed then the CHM

requests for areallocation from MAC. It can happen that there are several simultaneous reasons forreallocation. The CHM should take care that when it has requested a reallocation fora TBF, it will not anymore request reallocations for the same TBF.Selection algorithm for BTS reallocation is quite similar

to the case of BTS initialselection but more information is available: Rx-level measurement data is normallyavailable as well as MS RAC information. The same algorithm is used for both uplinkand

downlink reallocation.The mode of the TBF is not changed during reallocation. It means that EGPRS TBFcannot be reallocated to GPRS BTS.GPRS TBF is reallocated primarily to GPRS BTS. If there is not suitable GPRS BTSthen EGPRS BTS can be used. When better acceptable BTS is not found, the TBF

isnot reallocated.