USER DESCRIPTION
78/1553-HSC 103 12/4 Uen B
Us e rDe s c r i p t i o n,Dy n ami cBTSPo we rCo nt r o l Copyri!" © Ericsson AB 2002. All rights reserved. Di#$%&i'er The contents of this document are subject to revision without notice due to continued progress in methodology design and manufacturing. Ericsson shall have no liability for any error or damages of any !ind resulting from the use of this document.
Cont ent s 1
In"ro()$"ion
2 2." 2.2
*%o##&ry #oncepts Abbreviations and Acronyms
3 $." $.2 $.$ $.'
C&p&+i%i"ie# %nterference Battery bac!up po power consumption &eceiver saturation (uality and signal strength impact
4 '." '.2 '.$ '.' '.'. '.3
Te$!ni$&% (e#$rip"ion )eneral Algorithm *andover power boost +ower regulation e,ample )+&/E)+& A1& & +ower #ontrol 1ain changes in Ericsson )1 system &"0/B &"0
5 -." -.2 -.$
Enineerin )i(e%ine# %nteractions with other features re4uency planning aspects &ecommendations
, ." .2 .$
P&r&'e"er# 1ain controlling parameters +arameters for special adjustments 5alue ranges and default values
7
Reeren$e#
8
.ppen(i .
1 I nt r oduc t i on 6ith the 7ynamic BT +ower #ontrol feature the output power of a Base Transceiver tation 8BT9 can be controlled during a connection. The control strategy is to maintain a desired received signal strength and 4uality in the mobile station 819. This :ser 7escription describes the BT +ower #ontrol and A1& +ower #ontrol algorithm for circuit switched connections only.
2 Gl os s ar y 2 . 1 Concept s e)re'en" Repor"
e)re'en" Re#)%"
1essage consisting of measurements done by the 1 which is sent from the 1 to the BT.
1essage consisting of the 1easurement &eport and measurements done by the BT which is sent from the BT to the B#.
2 . 2 Abbr evi at i onsandAcr onyms ms .R
Adaptive 1ulti &ate
BCCH Broadcast #ontrol #hannel C/I
#arrier to %nterference &atio
CN.
#ellular ;etwor! Administration
DT
7iscontinuous Transmission
*PRS )eneral +ac!et &adio ervice RP
7
Reeren$e#
8
.ppen(i .
1 I nt r oduc t i on 6ith the 7ynamic BT +ower #ontrol feature the output power of a Base Transceiver tation 8BT9 can be controlled during a connection. The control strategy is to maintain a desired received signal strength and 4uality in the mobile station 819. This :ser 7escription describes the BT +ower #ontrol and A1& +ower #ontrol algorithm for circuit switched connections only.
2 Gl os s ar y 2 . 1 Concept s e)re'en" Repor"
e)re'en" Re#)%"
1essage consisting of measurements done by the 1 which is sent from the 1 to the BT.
1essage consisting of the 1easurement &eport and measurements done by the BT which is sent from the BT to the B#.
2 . 2 Abbr evi at i onsandAcr onyms ms .R
Adaptive 1ulti &ate
BCCH Broadcast #ontrol #hannel C/I
#arrier to %nterference &atio
CN.
#ellular ;etwor! Administration
DT
7iscontinuous Transmission
*PRS )eneral +ac!et &adio ervice RP
SDCCH tand Alone 7edicated #ontrol #hannel
3 Capabi l i t i es 3 . 1I nt er f er ence The aim with BT +ower #ontrol is to increase the number of 1s with sufficiently good #/%. BT +ower #ontrol will improve #/% if traffic is maintained or maintain #/% when traffic is increased or tighter fre4uency re=use is realised. The gain is obtained by a reduction of the over all interference level 8%9 in the networ!. 6hen BT +ower #ontrol is used in all BTs in the networ! the total amount of radiated power is reduced compared to when it is not used. This implies that the downlin! co= and adjacent channel interference in the networ! is reduced. ince 1s with low signal strength or bad 4uality use full BT output power reduced interference level imply increased #/% for these connections. >n the other hand the #/% is decreased for connections with high signal strength and good 4uality since they are subjected to a reduced BT output power. &eduction of #/% will not affect the speech 4uality of these connections since they have a margin to the lowest tolerable #/%. re4uency *opping together with BT +ower #ontrol and 7T? improve the possibilities to achieve very tight fre4uency reuse see further :ser 7escription 7iscontinuous Transmission and Transmission and :ser 7escription re4uency *opping. *opping .
3 . 2 Bat t er ybackuppowerconsump mpt i on %f the power supply for the base station is cut off a battery bac!up is used. 6hen BT +ower #ontrol is used the battery consumption is reduced and the ma,imum possible speech time will increase.
3 . 3 Re ce i v ers at ur a t i on The high signal energy from BTs transmitted transmitted to 1s that are close might saturate the 1 receiver. The sensitivity of the receiver will then decrease and the speech 4uality become poor. %f the output power of the concerned BTs BTs is lowered the ris! for this !ind of radio fre4uency bloc!ing is reduced. The receiver might still be bloc!ed if an 1 is very close to the base station but the probability for this is significantly reduced.
3 . 4 Qua l i t ya nds i gna ls t r e ngt hi mpa ct
Both 4uality and signal strength is considered by the algorithm. (uality is the estimated bit error rate which is represented by rxqual . ignal strength is represented by rxlev . Bad 4uality as well as low signal strength will increase the output power of the BT.
4 T ec hni c al des c r i pt i on 4 . 1 Gener al %mportant notice@ The algorithms in 1 +ower #ontrol and BT +ower #ontrol are the same. %n igure " the BT output power and the signal strength in the 1 versus path loss between a BT and an 1 is shown. A BT can only transmit at distinct power levels this is illustrated in the figure.
Figure 1
Base station output power and MS signal strength versus path loss. Quality is not taken into account.
6hen a connection has low path loss 8left part of igure " 9 the BT transmits at its lowest possible power level. Although the 1 receives a signal that e,ceeds the desired value the BT can not reduce the transmitted power any further. #onversely when a connection e,periences high path loss 8right part of igure "9 the BT transmits at the ma,imum allowed power level for the cell. The power cannot be increased even if the received signal strength in the 1 is low. ;ote that this is dependent on the path loss compensation used 8see ection '.2.' 9. 6hen 4uality is ta!en into account the output power is regulated up or down depending on the received 4uality 8see igure 2 9. The base station power then varies with the 4uality measured by the 1. 6hen an 1 has low rxqual 8high
4uality9 the base station sends on low power and when an 1 has high rxqual on high power. The higher the rxqual the higher the power and vice versa.
Figure 2
Exaple o! B"S output power versus rxqual. Signal strength is not taken into account.
4 . 2 Al gor i t hm 421
*ener&%
7ynamic BT +ower #ontrol is performed for Traffic channels 8T#*s9 as well as for 7##*s. +ower control of the 7##*s is enabled with the switch SDCCHRE*. All time slots on the B##* fre4uency are transmitted on full power i.e. there is no +ower #ontrol of these time slots. 7uring a call the 1 measures the downlin! signal strength and 4uality. These measurements are sent to the BT in the 1easurement &eport and further on to the B# in the 1easurement &esult message where they are used for calculation of a new BT output power. The measurements from the 1easurement &esult that are used in the 7ynamic BT +ower #ontrol algorithm are shown in Table ". Table 1
Measurements used by BTS Power Control
D&"& (e#$rip"ion
So)r$e
signal strength
downlin!
full set
8"9
1
signal strength
downlin!
subset
8"9
1
4uality
downlin!
full set
8"9
1
4uality
downlin!
subset
8"9
1
power level used by BT
BT
7T? used by BT or not
BT
8"9
The 1 performs signal strength and signal 4uality measurements on the downlin!. 1easurements are made on the full set of frames 8full set9 as well as on the subset of frames where there is always traffic 8subset9. 6hich of the sets will be used depends on whether 7T? downlin! has been used or not during the measurement period 8see also :ser 7escription 7iscontinuous Transmission 9. The minimum time period between two consecutive power orders is controlled by the parameter RE*INTD. RE*INTD is set in units of A##* periods 8'0 ms9 between " and "0. The BT is able to change its output power on a time slot basis. The resolution in output power is in steps of 2 dB and the ma,imum configurative change is $0 dB. or a single connection the ma,imum change per A##* period is also $0 dB. 7own regulation can be limited to 2 dB per A##* period by means of the parameter STEPID. The default value of this parameter is >. The 7ynamic BT +ower #ontrol algorithm consists of three stages@ ". Preparation of input data The output power level used in the latest measurement period is converted from a relative scale. A decision is ta!en about which set of measurements 8full set or subset 8"99 to use. ignal strength and 4uality are compensated for fre4uency hopping and power control. 2. Filtering of measurements 1easurements are filtered in e,ponential non=linear filters in order to eliminate variations of temporary nature. $. Calculation of power order Two power orders are calculated according to the algorithm using two different parameter settings. The one with the ma,imum power order 8minimum attenuation9 is chosen. A number of constraints 8according to hardware limitations and parameter settings9 are applied to the chosen power order. 422
Prep&r&"ion o inp)" (&"&
The output power level used by the BT 8T&:9 at A##* period k is given by #$used 8see e4. "-9 as a number of 2 dB steps downwards from the configured output power.
B"S %"&'( output power %k( %dB( BSPRT = 2 C #$used
8"9
%n the 1easurement &esult message the BT sends information about whether 7T? 8see :ser 7escription 7iscontinuous Transmission 9 has been used during the measurement period or not. This information is used by the B# to decide which set of downlin! measurements full set or subset to use on T#*s. The subset of measurements should be used if 7T? was used during the measurement period by the BT. >n 7##*s the full set of measurements are always used. To be able to use the desired 4uality 8 DESD9 and the measured rxqual in the calculations both must be converted to )*+ e,pressed in dB according to Table 2. The mapping between rxqual and )*+ is non=linear due to that faster regulation is needed for low and high rxqual values. Table 2
Table with relations due to non-linear r!ual to C"# mapping
DESD Ddt4u
0
"0
20
$0
'0
-0
0
30
rxqual
0
"
2
$
'
-
3
)*+ DdB
2$
"F
"3
"-
"$
""
'
DESD defines a desired value for rxqual that the regulation will aim for in the regulation process and is given in dt4u 8deci=transformed 4uality units9. 7ifference between dt4u and rxqual is a factor of ten. %f DESD is not e4ual to the values given in Table 2 linear interpolation is used to realiGe )*+ . E,ample of DESD interpolation@ %f DESD $- then )*+ "-H8"$="-9C0- "' dB DESD e,pressed in #/% is called Q,ES,$-dB which is the value used in the calculations. The B##* fre4uency is not subjected to power control. 6hen fre4uency hopping 8:ser 7escription re4uency *opping 9 is applied and the B##* fre4uency is included in the hopping set the BT output power will vary from burst to burst depending on which fre4uency the burst is sent on. A compensation is necessary to obtain a correct estimation of the measured signal strength see e4. 2.
SS") SSM = 8BSPR-BSTPR H2C#$used 9 / /!
829
where SS") is the signal strength on the down regulated T#* carriers SSM the measured signal strength reported by the 1 BSPR is the BT output power on the B##* fre4uency in the <&+ 8see :ser 7escription
SS-)0M# SS
H 2C #$used
")
8$9
where SS-)0M# is the signal strength compensated for both down regulation and fre4uency hopping. %f the B# does not receive the 1easurement &esult from a BT the power regulation is inhibited for that connection. At the same time the RE*INTD counting is suspended. 6hen a 1easurement &esult is received again power regulation and RE*INTD counting are resumed. The signal strength filter will not be updated when signal strength results 8measured in the 1easurement &eport9 are missing. This means that the output from the signal strength filter is held until the ne,t value is received. 1issing 4uality values in the 1easurement &eport are set to the worst possible value. This means that missing 4uality values are interpreted as rxqual 3. %f information about the BT power level used is missing in the 1easurement &eport the missing values are set to the latest calculated power order. 423
6i%"erin o 'e)re'en"#
The filtering for both signal strength and 4uality is done with e,ponential non=linear filters. SSF+$"E&E, in e4. ' is the filtered signal strength compensated for down regulation i.e. the signal strength that would have been received by the 1 if no power control was used. SSF+$"E&E, is defined as@
SSF+$"E&E, %k( 3 SS-)0M#%k( 4 a 3 SS F+$"E&E, %k51(
8'9
where and a 8 %15a(9 represent the filter coefficients SS-)0M# is the signal strength compensated for both down regulation and fre4uency hopping and k is a se4uence number. #oefficient a is given by the length of the e,ponential filter 8see Appendi, A9. Each filter length 8$9 corresponds to a certain value of a and $ is determined in the following way@
if
SS-)0M#%k( 6 %
then
$ SSEND
else
$ SSEND C UPDNR.TIO * 177
8-9
where $ is rounded upwards to A##* periods. 6hen the length e,ceeds $0 A##* periods the length is set to $0. To enable calculating and sending the power order immediately after assignment or handover the filter is initiated with SSF+$"E&E, %k51( SSDESD. This leads to that the regulation starts immediately after the first valid 1easurement report. (uality filtering is performed in the same way as for signal strength i.e. with e,ponential non=linear filters. The filtering is done according to e4. .
QF+$"E&E, %k( 3 Q-)0M#%k( 4 a 3 Q F+$"E&E, %k51(
89
where QF+$"E&E, is the filtered 4uality compensated for down regulation i.e. the estimated #/% 8in dB9 that would have been received by the 1 if no power control was used. Q-)0M# is the compensated 4uality part according to e4. 3.
Q-)0M# &8Q'9$-dB H 2C#$used
839
where &8Q'9$-dB is the measured rxqual transformed to )*+ 8in dB9 according to ection '.2.2. The coefficient a in e4. above is given by the length of the e,ponential filter 8see Appendi, A9 in the same way as for the signal strength case only that this time $ is determined in the following way@
if
Q-)0M#%k( 6 (%
then
$ END
else
$ END C UPDNR.TIO * 177
89
where $ is rounded upwards to A##* periods. To enable calculating and sending the power order immediately after assignment or handover the 4uality filter is initiated with Q F+$"E&E, %k51( Q,ES,$-dB. 424
C&%$)%&"ion o poer or(er
The calculation of the power order is made in three steps@
". The two basic power orders are calculated. 2. #ertain constraints are applied. $. The output data is finally converted to power order units before it is transmitted to the BT as a power order. The actual information sent to the BT is the power level #$used according to ection '.2.. The basic power orders for regulation 8 pu1 and pu29 are given by the following e,pression@
pui
3 8SSDESD 5 SSF+$"E&E, 9 H
i
i
C 8(7E7
8F9
i 1: 2 where the parameters
and
i
are defined as follows@
i
COPD * 177
8pathloss compensation9
8"09
COPD * 177
84uality compensation9
8""9
7.;
8pathloss compensation9
8"29
7.<
84uality compensation9
8"$9
1
1
2
2
The parameters i and i control the compensation of path loss and 4uality. The parameters 1 and 1 can be set by means of COPD and COPD while parameters 2 and 2 are fi,ed. These values have been optimised to get the regulation towards the noise floor fast without jeopardising the 4uality. The setting of 2 and 2 is however not critical since these parameters merely serve as a limitation for regulation close to the noise floor 8see ection '.'9. The two power orders are calculated simultaneously 8e4. F9 and the one with the highest value 8minimum down regulation9 is used. This resulting power order is called the unconstrained power order pu.
pu ax%pu1 :pu2 ( 425
8"'9
Poer or(er $on#"r&in"#
7ynamic power range limitation is applied if the unconstrained power order is outside the dynamic range@
The highest allowed power order is Gero 809. This corresponds to full power according to BSPRT .
The lowest allowed power order is given by the ma,imum of a. =$0 b. BSPRT = 81iminum BT output power 8*6 limit99 c. BSTPR - BSPRIN
;ote that even if the actual output power BSPRT in the BT is set to the minimum value lower power levels can actually be achieved when BT +ower #ontrol is active. or an &B2000 )1F00 1*G with minimum output power possible to configure e4ual to $- dBm 8 BSPRT@ $- to '3 dBm odd values only9 the lowest achievable output power is '3 = $0 "3 dBm when BT power control is active. 42,
Coner#ion o o)"p)" (&"&
The new power order has to be converted from the internal dBm scale to #$used representation before it can be transmitted to the BT. %n reality this means that the constrained power order is 4uantisiGed in steps of 2 dB according to@ #$used %nt8=pu/2 9 D0.."- where #$used is the power level. #$used 0 represents full power and #$used "represents $0 dB down regulation. The power is always truncated to a higher value 8lower down regulation9. 427
Re)%&"ion pro$e()re
6hen a T#* connection is set up ma,imum configurative output power is always used for e,ample in the following situations@
assignment of a T#*. assignment failure or handover failure. intra=cell handover and subcell change. inter=cell handover.
7own regulation always starts after the first valid 1easurement report 8see ection '.2.$9. The response time for up regulation is controlled by the parameters END and SSEND. END determines the response time on high interference and SSEND on signal strength drops. The values of END and SSEND corresponds to a F0 J rise time of the e,ponential filters. The response time for down regulation is determined by the e,pressions END CUPDNR.TIO /"00 and SSEND CUPDNR.TIO /"00 where UPDNR.TIO is the ratio between up= and down regulation speed. This results in a 4uic! up regulation and a smooth down regulation. UPDNR.TIO is a B# e,change property.
6hen a power order is sent it ta!es RE*INTD A##* periods before the ne,t power order can be sent. %f this power order differs from the previous one it is sent. %f it does not differ from the previous one a new order is calculated every A##* period until a different power order is obtained. Then that order is sent and RE*INTD A##* periods must elapse before a new order can be sent again. 428
)%"i#%o" $oni)r&"ion
%f the T#* channel is a part of a channel combination it can be either a main bi= directional or a uni=directional channel. %f the channel is a main channel in a multislot configuration the difference between the computed power order and the previous power order must e,ceed a hysteresis of two dB before a new power order is sent. BT power regulation on bi=directional channels is done independently of the other channels. or uni=directional channels BT +ower #ontrol is activated without starting normal power regulation. ;o 1easurement reports will be received for uni=directional channels. %nstead the BT power value of the main channel is distributed to the uni= directionals in the multislot configuration. %n a multislot configuration only the main channel is affected by the handover power boost see ection '.$. ee further :ser 7escription #hannel Administration and :ser 7escription *igh peed #ircuit witched 7ata 8*#79 .
4 . 3 Handoverpowerboost 6ith *andover power boost the handover command is sent by the B#/BT to the 1 on ma,imum configurative power. *andover command includes information about which uplin! power the 1 shall use in serving cell. The 1 then ac!nowledges the handover command using ma,imum configurative power. %n case of a *> failure the *> failure message is also sent on ma,imum configurative power. 6hen handover power boost is triggered normal regulation is inhibited until the 1 has received the handover command. The BT ignores all BT or 1 power orders sent by the B# in the serving cell until the 1 has ac!nowledged the handover command. The speech/channel coding and interleaving in )1 is very robust. A small number of bursts/frames can be lost without speech degradation 8the number depends on the error distribution9. +ower #ontrol should therefore also be used for connections close to the cell border. ince the signaling for the handover procedure 8e.g. *andover #ommand9 is more critical and error=sensitive it should be sent on ma,imum power in order to ma,imise the handover performance. *>+B is useful when the 4uic!ly drops for e,ample when the 1 moves around a street corner. %n this case due to the system delay and the limited up=regulation speed the signaling would be sent on a too low power without *>+B. Thus in order
to ma,imise the probability of a successful handover *andover +ower Boost should be used. ince the ma,imum configurative power is only used for a short time before the handover activating *>+B has a minor impact on the overall interference level in the networ!. ;ote that *>+B only improves the *> performance if power control is activated. *andover power boost is activated by setting the state variable *+BTATE.
4 . 4 Powerr egul at i onexampl e The most important thing for good comprehension of the BT +ower #ontrol algorithm is to understand how the two algorithms wor! in parallel and how different settings of the available parameters will influence the regulation. The e4uations given in ection '.2.' can be used to find out how much the output power will be down regulated for a certain signal strength and 4uality. But to get an overview picture of the algorithm as a whole the dependence between signal strength 4uality and down regulation must be understood. A suitable way of studying these three 4uantities is in a three dimensional plot describing the static behaviour of the algorithm.
Figure ;
#rincipal !igure !or downregulation
As it can be seen in igure $ the surface is raised for rxlev K "' and rxqual L -. The down lin! for 1s in this area is down regulated. The level of the down regulation is shown on the G=a,is. ;ote that rxqual and rxlev in igure $ corresponds to the measured values collected from the 1easurement &eport before any compensation has been done. The static behaviour is calculated by assuming an initial down regulation of Gero and that the path loss to the 1 is constant. Then for a certain value of initial &,
4 . 5 GPRS/ EGPRS )+&/E)+& BT +ower #ontrol is not supported in B &"0. ull output power is used on all )+&/E)+& channels.
4 . 6 AMRFRPowerCont r ol 4,1
*ener&%
Adaptive 1ulti &ate 8A1&9 is a speech and channel codec feature for full rate channels that ma!es it possible to acheive improved speech 4uality for mobile connection as well as better capacity see :ser 7escription Adaptive 1ulti &ate. The A1& +ower #ontrol is used to minimiGe the interference in the radio networ! for A1& & connections by reducing the output power of the A1& & connections. 4,2
.R Poer Con"ro% .%ori"!'
The A1& +ower #ontrol is based on the the 7ynamic BT +ower #ontrol and 7ynamic 1 +ower #ontrol respectively see :ser 7escription 7ynamic 1 +ower #ontrol.
The A1& & speech coding is more robust and can perform well on low #/% levels. This results in a possibility to down regulate the output power of A1& & connections more than for non=A1& connections. This means that A1& & +ower control parameter set can be set more aggressive than for non=A1& parameter setting. To be able to set the parameter more aggressive for A1& & connections two new parameters are implemented SSDESD.6R and DESD.6R in the 7ynamic BT +ower #ontrol. This means that the two power orders for A1& & connections are calculated according to@ pui i C 8SSDESD.6R = SS F+$"E&E,9 H QF+$"E&E,9
C 8(7E7
i
8"-9
i 1:2 The Q,ES,$9F&-dB is DESD.6R e,pressed in )*+ 8in dB9 according to ection '.2.2. Then the remaining calculations in ection '.2.' are the same.
4 . 7 Mai nchangesi nEr i cssonGSM sys t em R10/ BSSR10 A1& +ower #ontrol is introduced.
5 Engi neer i nggui del i nes 5 . 1I nt er act i onswi t hot herf eat ur es The gain of BT +ower #ontrol increases in high capacity systems utiliGing a tight fre4uency reuse. The primary application is a system that uses a combination of 7ynamic BT +ower #ontrol 7ynamic 1 +ower #ontrol re4uency *opping and 7T?. The mutual interaction between these features provides a very powerful method to increase system performance and thereby system capacity 8see further in :ser 7escription 7iscontinuous Transmission :ser 7escription re4uency *opping and :ser 7escription 7ynamic 1 +ower #ontrol 9. +referably power regulation should be configured to be performed before an intra= cell handover occurs. Also power regulation should be configured to always occur before a bad 4uality urgency handover is attempted. The desired regulation performance can be achieved through a well balanced combination of the following@
the BT +ower #ontrol parameters SSDESD and DESD that set the limits for how close to the noise floor 8how low rxlev 9 and how high in interference 8how high rxqual 9 BT down regulation can be performed.
the A1& & +ower #ontrol parameters SSDESD.6R and DESD.6R that set the limits for how close to the noise floor 8how low rxlev 9 and how high in interference 8how high rxlev 9 A1& & down regulation can be performed. the 4uality compensation factor COPD and the path loss compensation factor COPD that determine the angles of inclination of plane $ in igure $. the intra=cell handover area defined by O66SETD and O66SETD.6R 8:ser 7escription %ntra #ell *andover 9. the threshold triggering bad 4uality urgency handovers ID and ID.6R 8:ser 7escription
E,ample@ DESD $0 O66SETD - and ID --. 6ith this setting full power will always be used before an intra=cell or urgency handover occurs.
5 . 2 Fr equencypl anni ngaspect s %n order to utiliGe BT +ower #ontrol in an optimum way it is preferable to use a dedicated B##* band. This means that a B##* carrier is never used as a T#* carrier and vice versa. The level of interference will in this way be decreased for all T#* carriers. The B##* carriers are unaffected but will depending on the fre4uency plan e,perience less adjacent channel interference from the down regulated T#* carriers. The B##* carriers can either be allocated in a contiguous B##* band or in a staggered B##* band. %n a contiguous band carrier no. "="- can for e,ample be used as B##* carriers whereas in a staggered band for e,ample every second fre4uency can be used as B##* carriers 8"$-..$"9. There are pros and cons with both these strategies. or BT +ower #ontrol it is probably beneficial to use the contigous B##* band since when using staggered B##* the down regulated T#* carriers in between B##* carriers will suffer from adjacent channel interference from the on full power always transmitting B##* carriers. %n a networ! with tight reuse and if the B##* carriers are allocated in a contigous band it is beneficial to use a more aggressive setting than the recommended e.g. by increasing (#>1+7< to -.
5 . 3 Rec ommendat i ons 531
*ener&%
6hen attempting to decrease the downlin! co=channel and adjacent channel interference in the system the BT +ower #ontrol feature should be considered.
*owever since downlin! power regulation is never performed on B##* carriers the impact of downlin! regulation will be greater in systems having three or more Transceivers 8T&?s9 per cell. 6hen introducing BT +ower #ontrol into a system it is recommended to begin with moderate settings for the controlling parameters. The majority of the gain obtained from using power control originates from the first decibels of regulation. Therefore a good strategy is to down regulate many connections with a few dB. To get the best effect it is important to reduce the BT output power for as many connections as possible also those connections to 1s in the cell border regions being closest to neighbouring users. or such 1s however the interference levels are often considerable and great care has to be ta!en not to degrade such calls. 532
T)nin o "!e &%ori"!'
The shown down regulation in igure $ and in the graphs in this section is a target regulation that the algorithm aims for. %t is important to understand that the down regulation is determined by the $o'+in&"ion of the parameters SSDESD and DESD or SSDESU.6R and DESU.6R for A1& & connections not one of the parameters alone. ince the environment changes 4uic!ly and the filtering of signal strength and 4uality introduces delays the target down regulation is never reached directly. The recommended strategy 8see igure $9 is a good parameter setting that is not particularly aggressive according to any regulation strategy. By changing the parameters the regulation can be made more aggressive towards 4uality or signal strength or combinations depending on the needs of the customer. ;ote it is not recommended to limit the down regulation with the parameter BSPRIN. %f used the parameter will seriously limit the regulation towards interference and also introduce a delay in the regulation algorithm. %nstead it is recommended to use a more restrictive parameter setting e.g. according to igure . To get a regulation that is more aggressive towards 4uality 8i.e. allows higher interference before it regulates up to full power9 DESD can be set to a higher value e.g. DESD '0. This will lead to if no other parameters are changed an increase of the raised surface in igure $ that grows mainly to the right 8towards worse 4uality9 but also a little bit to the left 8towards lower signal strength9. And if the inclination of plane $ is left unchanged the result is also an upwards shift of this plane. As an e,ample igure ' shows more aggressiveness towards 4uality signal strength and down regulation compared to igure $. till the only parameter that has been changed is DESD.
Figure < 9ggressive paraeter setting towards quality. "his setting is rather aggressive: also towards signal strength and down regulation. 0nly paraeter Q,ES,$ has een changed copared to recoended setting %see Figure ;(. or the parameter setting in igure ' the 4uality part of the power control will always fully compensate for bad 4uality. ull power should be reached 4uic!ly in case of high rxqual 8rxqual - or 39. This is in order to minimise the ris! of having poor speech 4uality due to too much down regulation and also prevent unnecessary intra=cell handovers and urgency handovers. *ence a shorter 4uality filter might be needed 8see ection -.$.'9. As an e,ample of more aggressive regulation towards signal strength study igure -. The only parameter changed compared to the recommended setting is SSDESD which is set to =F3. or this setting the downlin! for 1s with rxlev "0 and rxqual 0 is down regulated ' dB. ;ote that this might sound a bit more aggressive than it is since at this low signal strength noise will impose occasional bit errors to the connection. This will ma!e the regulation to MbounceM on the noise floor. 5ery few connections will then manage to be as much as ' dB down regulated. %nstead most connections will alter between 0 and 2 dB down regulation.
Figure =
9ggressive regulation towards low rxlev. MSs with low signal strength also get down regulated in case o! good quality.
As an e,ample of a more careful regulation strategy see igure . This shows how DESD can be decreased compared to the recommended setting to get a very moderate setting. 1a,imum "0 dB down regulation is then allowed.
Figure >
Moderate paraeter setting. 0nly paraeter Q,ES,$ has een changed copared to recoended setting %see !igure ;(
To compensate for this low setting of DESD one alternative could be to allow more down regulation for those 1s that have good 4uality. igure 3 show how this can be done. The parameter COPD is increased and as a result the inclination of
plane $ is changed. The algorithm then allows more down regulation for 1s with good 4uality but is still careful when it comes to regulation towards bad 4uality.
Figure ?
Moderate paraeter setting: ore aggressive towards down regulation.
Another way of changing the inclination of plane $ would be to change the path loss compensation parameter COPD . %n igure COPD has been set to "0 while all other parameters are the same as in igure . This results in that the 1s with high signal strength regardless of 4uality gets more down regulated.
Figure @ Moderate paraeter setting with path loss copensation !actor $)0M#,$ set to 17. "his results in a very aggressive ehaviour towards down regulation.
6ith the setting in igure plane $ has become very large and dominating. This setting has regulation towards signal strength and is more aggressive towards down regulation. The ma,imum down regulation is here " dB compared to "' dB for the old recommended setting. %mportant notice@ The default values given in Table $ are also ;>T recommended to useN 533
E&'p%e# o p&r&'e"er #e""in#
Below are some e,amples of static behaviour with different parameter settings shown. The first figure illustrates the recommended setting and the rest of the e,amples are sorted in order of increasing MaggressivenessM. These e,amples can all be considered as recommendations for different MaggressivenessM levels.
Figure A
"he recoended setting.
Figure 17
Figure 11
Figure 12
Figure 1;
Figure 1<
Figure 1=
Figure 1> 534
6i%"er ")nin
)enerally for up regulation the BT +ower #ontrol 4uality filter END can be set to a value between 2 or -. This is fairly uncritical since instability in the control loop has not shown to be a problem with this control strategy. Therefore it is better to have a short power control 4uality filter since the response to bad 4uality then becomes 4uic!. %t is not useful to set END ". This would only lead to e,tremely nervous behaviour resulting in less average down regulation. Tests have shown that the difference in fast up regulation between END 2 and END $ is insignificant. %n order to avoid unstable behaviour the down regulation must be slow. Tests have shown that a filter with lengths between and F is good. >f course longer filters can also be used. This would result in an even more cautious behaviour. The filter length on the down regulation is determined by parameters END and UPDNR.TIO. UPDNR.TIO sets how much longer the down regulation filter is compared to the up regulation filter in percent. %t is recommended to use high UPDNR.TIO instead of using STEPID. As an e,ample of how the system reacts to bad 4uality see igure "3. E,ample@ END is 2 and UPDNR.TIO is 00. This gives 2 A##* periods filter length for up regulation and 2C00J 2C "2 A##* periods filter length for down regulation.
Figure 1? Step response to ad quality. #araeter setting Q$E/,$ ; and '#,/&9"+0 ;77 was used. /ote the logarithic ehaviour o! the down regulation. The BT +ower #ontrol signal strength filter is less critical. The regulation is done in the same way as for 4uality filtering. The length of the up regulation filter is set by the parameter SSEND and for the down regulation by SSEND and UPDNR.TIO. or up regulation SSEND $ is recommended. The parameter UPDNR.TIO should be tuned for the 4uality filter. %f it is tuned for 4uality filtering it is also valid for signal strength filtering. Thus for down regulation a filter length of to F is recommended but longer filter lengths can be used if necessary. ee also igure ".
Figure 1@ Step response to low signal strength. #araeter setting SS$E/,$ ; and '#,/&9"+0 ;77 was used. 9ggressive paraeter setting gave 1> dB down regulation e!ore the low signal strength occurred. /ote the logarithic ehaviour o! the down regulation.
RE*INTD should be set to RE*INTD " in order to ma!e the up regulation 4uic! in bad 4uality situations.
6 Pa r a me t e r s 6 . 1 Ma i nc ont r ol l i ngpa r a me t e r s SSDESD defines the target value for the desired signal strength measured by the receiver in the 1 at the outer rim of the regulation area. The parameter is set per subcell. DESD defines the target value for the desired 4uality level measured by the receiver in the 1. %t is measured in r,4ual units and transformed into dB units before is used in the algorithm. The parameter is set per subcell. SSDESD.6R defines the target value for the desired signal strength for A1& & connection measured by the receiver in the 1 at the outer rim of the regulation area. The parameter is set per subcell. DESD.6R defines the target value for the desired 4uality level for A1& & connection measured by the receiver in the 1. %t is measured in r,4ual units and transformed into dB units before is used in the algorithm. The parameter is set per subcell. COPD is the parameter that determines how much of the path loss that shall be compensated for in the algorithm that regulates towards 4uality. The parameter is set per subcell. COPD is the parameter that determines the weight of the 4uality compensation. This parameter ranges between 0 and "00 and is set per subcell.
6 . 2 Pa r a me t e r sf ors pe ci a la dj us t me nt s RE*INTD defines the regulation interval. The parameter is set per subcell. SSEND defines the length of the signal strength filter. The parameter is set per subcell. END defines the length of the 4uality filter. The parameter is set per subcell. SDCCHRE* is a switch for the regulation of 7##* channels. The switch is set per subcell. BSPRIN defines the minimum allowed output power for the BT on the non= B##* fre4uencies. The parameter is set per subcell.
BSTPR defines the ma,imum allowed power level for BTs in the current subcell. The parameter is also used in
6 . 3 Va l uer a nge sa ndde f a ul tv a l ue s Table $
De&)%" &%)e
Re$o''en(e( &%)e
9&%)e r&ne
Uni"
=30
=F0
=""0 to ='3
dBm
DESD
20
$0
0 to 30
dt4u
SSDESD.6R 8"9
=30
=F08$9
=""0 to ='3
dBm
DESD.6R
20
'08$9
0 to 30
dt4u
COPD
30
-
0 to "00
J
COPD
$0
--
0 to "00
J
RE*INTD
-
"
" to "0
A##* periods
SSEND
-
$
$ to "-
A##* periods
END
$
" to 20
A##* periods
SDCCHRE*
>
>;
>; >
BSPRIN
=20
=20
=20 to H-0
dBm
0 to 0
dBm
P&r&'e"er n&'e
SSDESD
8"9
BSTPR 829
UPDNR.TIO
200
$00
"00 to 300
STEPID
>
>
>; >
J
8"9 SSDESD and SSDESD.6R ta!es the corresponding positive value in 11< commands and #;A. 829 The value of this parameter is highly dependent on the cell planning. ;o default value is provided. 8$9 These recommended values are based on assumptions/simulations and have not been live tested.
7 Re f e r en c es ". 2. $. '. -. . 3. .
:ser 7escription 7iscontinuous Transmission :ser 7escription re4uency *opping :ser 7escription
8 Appendi xA 6i%"er $oei$ien"# Table %
Coefficients for the eponential filters used&
6i%"er %en"! '
6i%"er $oei$ien" a
"
0."000
2
0.$"2
$
0.''2
'
0.-2$
-
0.$"0
0."$
3
0.3"F3
0.3'FF
F
0.33'$
"0
0.3F'$
""
0."""
"2
0.2-'
"$
0.$33
"'
0.'$
"-
0.-33
"
0.0
"3
0.3$$
"
0.3FF
"F
0.-F
20
0.F"$
2"
0.F2
22
0.F00
2$
0.F0'3
2'
0.F0-
2-
0.F"20
2
0.F"-2
23
0.F"$