Radio Resource Management for Wireless Networks - TUT · Radio Resource Management for Wireless...
Transcript of Radio Resource Management for Wireless Networks - TUT · Radio Resource Management for Wireless...
1
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Radio Radio ResourceResource Management for Management for WirelessWireless NetworksNetworks
Janne Kurjenniemi
Magister Solutions
Prof. Tapani RistaniemiTampere University of Technology
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ContentsContents
Power controlHandoverAdmission controlLoad control
2
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ChapterChapter 1: 1: PowerPower controlcontrol
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
WhyWhy transmittertransmitter powerpower controlcontrol ??Reduce terminal power consumptionMitigate near-far effects
efficient handling of mobilityefficient reduction of the dynamic range of received power
Reduce interference / improve quality”BS tx power = interference to the system””MS tx power = interference to other cells””BS rx power = interference to intra-cell users”
PC is a radio resource management toollow vs. high complexity (how much in assumed to be knowabout the system)slow vs. fast (where the decision is done)
3
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
OpenOpen looploop PCPCTwo purposes for open loop PC:
initial power level in the beginning of a callquality target updating
Open loop PC can be non-linearthis is to allow fast response for negative corrections (e.g. upto 10 dB in 1 ms) but slow response for positive corrections(e.g. 1 dB in 1 ms)Why non-linear PC: When attenuation is suddenly decreased, it is important to quicly decrease the TX power in order notto cause additional interference to other users.
Since the open loop method cannot estimate reverselink fading (due to different band and relatedindependent fading), open loop PC becomes inaccurate. In 3GPP, accuracy requirement for OL PC is 10 dBThis is why closed loop PC in needed, too.
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ClosedClosed looploop PCPC
signalerror waveformsampling
estimationpower in error instant sampling at thepower received
instant sampling at thepower unit mobile
−−−−−
n
n
n
n
tn
ESN
nRnP
The base station will be estimating the received power in order to find out whatkind of correctionis needed.
The impact of noise on the overall process is removed by averaging B samplesThe result is compared with desired one and a sample of error signal is created.Error signal is transmitted on the downlink after propagation delay of D samples.
Dnt
nt
n EPP −+ +=1Log-linear PC model
4
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
WCDMA WCDMA PowerPower ControlControlFast Power Control
Fast Power Control with 1.5kHz is supported in both UL and DL inWCDMAOnly slow power control with about 2Hz is employed in GSMIn IS-95, fast power control with 800Hz is supported only in UL
Gain of Fast Power ControlRequired Eb/No values with and without power control
Required relative transmission powers with and without fast power control
Slow Power Control(dB)
Fast 1.5kHz power control(dB)
Gain from fast power control(dB)
ITU Pedestrian A 3 km/h 11.3 5.5 5.8ITU Vehicular A 3 km/h 8.5 6.7 1.8ITU Vehicular A 50 km/h 6.8 7.3 -0.5
Slow Power Control(dB)
Fast 1.5kHz power control(dB)
Gain from fast power control(dB)
ITU Pedestrian A 3 km/h 11.3 7.7 3.6ITU Vehicular A 3 km/h 8.5 7.5 1ITU Vehicular A 50 km/h 6.8 7.6 -0.8
Fast PC with 1dB
step
Fast PC with 1dB
step
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
GainGain of of fastfast PC in WCDMAPC in WCDMAGenerally, slow PC is meant to compensate the effects of path loss and shadowing, whereas fast PC is used to compensate also for fast fading.
From the tables just presented we see that fast PC gives a cleargain in WCDMA.
It can be seen that the gain from fast PC is larger:for low UE speeds than for high UE speedsin required Eb/No than in transmission powersfor those cases where only a little multipath diversity is available, as in the ITU Pedestrian A channel.
The negative gain at 50km/h indicate that slow PC is better thanFast PC. The reasons are
inaccuracies in the SIR estimationpower control signaling errors, and the delay in the power control loop.
5
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
RxRx and and TxTx powerspowers vs. vs. interferenceinterference levelslevels
Slow and fast PC are also important in reducing the level of interference
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
WCDMA PC in soft WCDMA PC in soft handoverhandoverTwo major issues that are different from the single-link case:
power drifting in Node B powers in the downlink reliable detection of the UL power control commands in the UE.
6
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
SolutionsSolutions for DL for DL powerpower driftingdriftingThe simplest method is to set relatively strict limits for the downlink power control dynamics.
Naturally, the smaller the allowed power control dynamics, the smaller the maximum power drifting.On the other hand, large power control dynamics typically improve power control performance.
Another solution: RNC can receive information from the Node Bs concerning the transmission power levels of the soft handover connections. These levels are averaged over a number of power control commands, e.g. over 500 ms or 750 power control commands.Based on those measurements, RNC can send a reference value for the DL transmission powers to the Node Bs. The idea is that a small correction is periodically performed towards the reference power.
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
SolutionSolution for for reliablereliable UL PC UL PC commandcommand detectiondetectionTo make UL PC command detection more reliable, a simple solution is to improve PC signaling quality in downlink as follows:
The power of DPCCH channel is more than that of DPDCH
7
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
OuterOuter looploop PC in WCDMAPC in WCDMAOuter loop power control (OL PC) is needed to keep the quality of communication at the required level by setting the target forthe fast power control:
OL PC aims at providing the required quality: now worse, no bettertoo high quality would waste capacity
- OL PC is needed in both uplink and downlink becausethere is fast PC in both links, too.
- The uplink outer loop is located in RNC and the downlink outer loop is located in UE
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
General General outerouter looploop PC PC algorithmalgorithm
Frequency of the outer loop power control typically 10-100Hz.
8
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
GainGain of OL PCof OL PC
6.91203-path equal powers
6.4503-path equal powers
6.4203-path equal powers
6.033-path equal powers
7.1120ITU Pedestrian A
6.850ITU Pedestrian A
6.820ITU Pedestrian A
5.93ITU Pedestrian A
5.3-Non-fading
Multi-path UE speed (km/h) Average SIR target (dB)
The higher the variation in the received power, the higher SIR target needs to be to provide the same quality (here, 1 % BLER).
For example, if SIR target would be fixed to 7.1 dB, unnecessary highpower would be used in other channels.
Since we have different SIR targets in different channels, we need to dynamically adjust the SIR target of the fast closed PC by OL PC !
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
EstimationEstimation of of receivedreceived qualityqualityEstimation of received quality is needed by the OL PCTwo main approaches:
The results of CRC (cyclic redundancy check)simple reliable detector for frame errors
Soft frame reliability information. Such information could be e.g.
estimated bit error rate before channel decoder, called raw BER or physical channel BERsoft information from Viterbi decoder with convolutional codessoft information from Turbo decoder, for example BER or BLER after an intermediate decoding iterationreceived SIR
9
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
HighHigh qualityquality servicesservices
High quality service with very low BLER(< 0.001) are required to be supported by 3G networks.With low BLER errors are very rare events.With CRC algorithm, the adjustments of SIR target are very slow and the convergence of SIR target to the optimal value takes a long time. Therefore, for high quality services the soft frame reliability information has advantages.
Soft information can be obtained from every frame even if there are no errors.
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Limited PC Limited PC dynamicsdynamics
UE Tx power
block errors
UL Eb/No target
Eb/No target gets unnecessarily high because max. Tx powers are already in use=> when UE gets closer to the Node B its Eb/No target (and hence Tx power) is unnesessarily high for some time until the OL PC is able to correct the Eb/No target again to the optimal value.
attenuationbetween UE and Node B
There is no need to increaseEb/No target anymore !
10
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ChapterChapter 2: 2: HandoverHandover controlcontrol
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
HandoverHandover typestypes in UMTSin UMTSSeveral types of handovers are possible in UMTS
FDD soft/softer HO (intra-system intra-frequency HO)FDD inter-frequency (intra-system inter-frequency HO)WCDMA-GSM HO (inter-system HO)GSM-WCDMA HO (inter-system HO)FDD/TDD HO (inter-system HO)TDD/FDD HOTDD/TDD HOHSDPA/HSDPA HOR99/HSDPA HO
f1
GSMGSM
f2
f1
GSM
f1 f1 f1f1WCDMA
WCDMA
Softer Soft
Inter-frequency
Inter-systemInter-system
11
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
IntraIntra--frequencyfrequency HOsHOsSoft handover uses typically pilot (CPICH) Ec/Io as the handover measurement quantity, which is signaled to RNCTerminology:
Active set - the cells that a UE connects in soft handover connectionNeighbor set - the list of cells that the UE continuously measures, but whose pilot Ec/Io are not strong enough to be added to the active set
Handover Control supports the following intra-frequency measurement report events
Event 1a: a P-CPICH enters the reporting rangeEvent 1b: a P-CPICH leaves the reporting rangeEvent 1c: a non-active P-CPICH become better than an active P-CPICHEvent 1d: change of the best cellEvent 1e: a P-CPICH becomes better than an absolute thresholdEvent 1f: a P-CPICH becomes worse than an absolute threshold
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
General HO General HO schemeschemeIf “Pilot_Ec/Io > Best_Pilot_Ec/Io –Window_add” for a period of ∆T & active set not full Event1A (Radio Link Addition)If “Pilot_Ec/Io < Best_Pilot_Ec/Io – Window_drop” for a period of ∆T
Event1B ( Radio Link Removal)If the active set is full and “Best_candidate_Pilot_Ec/Io > Worst_Old_Pilot_Ec/Io + Hysteris_event1C” for a period of ∆T Event1C (Combined Radio Link Addition and Removal)
Window_drop
P-CPICH #1
P-CPICH #2
Ec/No
Window_add
Absolute threshold
T
T T
P-CPICH #3
Event 1D
Hysteresis_event1C
Connected to cell1 Connected to cell1&2 cell2&3 Connected to cell2
Event 1E
UE connect tocell(s) with
Event 1A
Event 1C
Event 1B
12
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
HandoverHandover measurementsmeasurementsThe intra-frequency handover measurementprocedure is as follows:
UE synchronizes to all cells that are within thedetection range using P-SCH,S-SCH and CPICH,and identifies the cell.
UE decodes System Frame Number(SFN) fromBCH of the neighbor cell, since Node Bs can be asynchronous.
If the reporting trigger (e.g. Window_Add or Window_Drop) is fulfilled, UE reports handover measurements to RNC.
RNC sends active set update command to the UE
Notice: the more peaks the UE can distinguish with its MF, the longer the cell identification takes.
Generally, the cell identification time depends on:-Number of multi-paths-Number of cells within detection range-Number of already found cells-Size of the neighbourlist
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Soft HO Soft HO GainsGainsSoft handover gains can be achieved by:
Macro diversity gain: a diversity gain over slow fading and sudden drops in signal strengthMicro diversity gain: a diversity over fast fadingDL load sharing: a UE in soft handover receives power from multiple Node Bs which means that the maximum transmit power to a UE in X-way soft handover is multiplied by factor X (i.e. improved coverage)
All these gains can be mapped to improved coverage.
13
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ExampleExample: : LinkLink GainsGains withwith 22--branch branch NodeNode BsBs
The gains are relative to the single link case where the UE is connected to the best Node BThe highest gain is obtained when the path loss is the same to both Node BsThe max soft handover gain of DL is more than that of UL, because UL has already gained from antenna diversityFor DL, in case the path loss difference is more than 4-5dB, the UE does not experience a gain from the signal transmitted from the Node B with largest path loss
the power transmitted from that Node B to the UE will only contribute to total interference in the network.
Typical values: Window_add = 1-3 dB, Window_drop = 2-5 dB
-3
-2
-1
0
1
2
3
-10 -6 -3 0Soft
HO g
ain
inUL
Txpo
wer
-3
-2
-1
0
1
2
3
-10 -6 -3 0
Soft
HO g
ain
in D
L Tx
pow
er
Relative path loss to BS1 vs. BS2 Relative path loss to BS1 vs. BS2
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Soft HO Soft HO overheadoverheadSoft handover overhead (SHO) is defined as
where N is the active set size and Pn is the average probabilityof a UE being in n-way soft handover.Factors affecting to SHO
ParametersWindow_addWindow_dropActive set size
Other factorsNetwork topology: site distance, sectors per siteAntenna radiation patternsPath loss & shadow fading characteristicsThe average number of Node Bs that a UE can synchronize to.
∑=
−=N
nnnP
11β
14
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ExampleExample: SHO vs. : SHO vs. Window_addWindow_add
1,1
1,2
1,3
1,4
1,5
1,6
1,7
1,8
1,9
1 1,5 2 2,5 3 3,5 4 4,5 5
Window_add(dB)
Sof
t han
dove
r ov
erhe
ad +
1 cell radii:2000mcell radii:666m
Assumptions:- 3 sector sites- 65 degree antenna- Okumura-Hata path-loss model- 8 dB standard deviation for shadow fading- CPICH power:
•10% for 666m•20% for 1000m
- SCH power is –3dB to P-CPICH- Active set size=3- Window_drop=window_add+2dB
- SHO increases approximately linearly when Window_add is increased- SHO is typically larger for small cells with the same soft handover parameter setting- Window_add=1-3dB is suitable if SHO’s design goal is 20-40% for 3 sector sites- For 6 sector sites, SHO is increased by approximately 30%. Thismeans that Window_add/drop should be lowered.
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ActiveActive set set updateupdate raterate (ASUR)(ASUR)Definition:
It is counted as the time between consecutive active set update commands and includes all the events: addition, removal and replacement
ASUR is relevant for RNC dimensioning since ASU causes signaling load to the RNC.ASUR depends on
average mobility of the userscell sizenetwork topologyhandover parameters
SHO and ASUR are related: smaller SHO can be obtained at the expence of a higher active set update rate.
15
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
WCDMA to GSM ISHOWCDMA to GSM ISHO
Motivation for inter-system handovers (ISHOs)
coverage compensationload balancing reasons
Inter-system measurements are not active all the time, i.e., only when there is a needMeasurement trigger is vendor specific. It can be based on quality (BLER) or required transmission power
RNC commands UE to start inter-system measurements with compressed mode(measurement trigger is RAN vendor specific)
UE measures signal power of GSM frequenciesIn the neighbor list (typically,1second of measurements needed to average fading out)
RNC commands the UE to decode BSIC of the best GSM candidate(on average takes 1 second)
RNC sends handover command to the UE
ISHO procedure
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
CompressedCompressed modemodeCompressed mode is needed for inter-system and inter-frequency measurements if UE has only a single receiverCompressed mode means that transmission is halted for a periodof time during which measurements are performed.During compressed mode,
fast power control can’t be applied part of the interleaving gain is lost
Thus, higher Eb/No is needed, which leads to some capacity degradationHowever, if compressed mode is active only when needed, the capacity reduction is negligible.For example, if 10% of the users are simultaneously in compressed mode and every 3rd frame is compressed, the capacity lost is around 2 %.
16
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
EffectsEffects of of compressedcompressed modemodeCompressed mode affects the UL coverage area of the real time services since the bit rate can’t be lowered during compressed mode.
Thus, the coverage reason inter-system handover procedure has to be initiated early enough at the cell edge.
Compressed mode affects coverage in 2 waysSince the same amount of data is transmitted in a shorter time, Txpower should be increased: if there exists a 7-slot gap in a 15-slot frame, power increase is 10*log(15/(15-7)) = 2.7dBThe Eb/No performance degrades during the compressed mode (~2dB)
Normal frame Normal frameGapCompressed frame
10ms 10ms 10ms
Effect of compressed mode on coverage
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ISHOsISHOs withinwithin WCDMAWCDMADifferent ways to use several WCDMA frequencies
Compressed mode measurements are used also in the inter-frequency handover in the same way as in the inter-system handovers
f1 f1 f1 f1f2 f2
f1 f1 f1f2 f2 f2 f2 f2 f2 f2 f2
High capacity sites with 2 frequencies f1 and f2
Micro layer with frequency f2
f1
17
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ISHO ISHO proceduresprocedures
RNC commands UE to start inter-frequency measurements with compressed mode(measurement trigger is RAN vendor specific)
UE finds P-SCH peaks
UE identifies the cell with S-SCH and CPICH and reports measurements to RNC
RNC sends handover command to the UE
3GPP requirement: cell identification in 5 seconds with pilot Ec/Io>-20 dB
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Handover measurements
Typical handover measurement reporting from UE to RNC
Typical handover reasonsHandover type
WCDMA intra-frequency
Measurements all the time with matched filter
event-triggered reporting
-Normal mobility
WCDMA->GSM inter-system
Measurements started only when needed,compressed mode used
Periodic during compressed mode
-Coverag -Load -Service
WCDMA inter-frequency
Measurements started only when needed,compressed mode used
Periodic during compressed mode
-Coverage -Load
SummarySummary of WCDMA of WCDMA handovershandovers
18
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
WCDMA HSDPA WCDMA HSDPA handovershandoversNo soft handovers in HS-DSCH (high speed downlinkshared channel)UTRAN determines the serving HS-DSCH cell for an HSDPA capable UEServing HS-DSCH cell change is triggered by UE measurement reports and determined by UTRANRNC dictates the time moment when serving cell is changedA new UE measurement event is included in Release 5 to inform RNC of the best serving HS-DSCH cell
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
MeasurementMeasurement eventevent for for bestbest servingserving HSHS--DSCH DSCH cellcell
UE reports the best servingHS-DSCH cell to the servingRNC
Based on pilot Ec/Io or RSCP measurements
Hysteresis margin can beused to avoid fast change of serving HS-DSCH cellCell Individual Offset value(CIO) can be used to favourcertain cells (to increasecell’s coverage area)
19
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
WCDMA HSDPA WCDMA HSDPA handoverhandover typestypesThere are three handover types
Intra-Node B HS-DSCH to HS-DSCH handoverInter-Node B HS-DSCH to HS-DSCH handoverHS-DSCH to DCH handover
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Intra-Node B HS-DSCH to HS-DSCH HONode-B and UE are informedabout incoming handover bySRNCTime moment for the handoveris specified by the SRNCAll transmission from the sourcecell stops at that specified timeand the packet scheduler in the target cell is then allowed to control transmission to UEMAC-hs preservation:
Buffered data in source cell is moved to target cell inside the Node B ⇒ no data loss!No higher layer retransmissions(e.g. from RLC protocol)
If MAC-hs preservation is notsupported ⇒ same as inter-NodeB handover case
20
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Inter-Node B HS-DSCH to HS-DSCH HONode Bs are potentially underdifferent RNCNode-Bs, target cell RNC and UE areinformed about incoming handover bycurrent SRNCAt the time of cell change, the MAC-hs for the user in the source cell is reset (user data in buffers is deleted)At the same time the MAC-hs flowcontrol unit in target cell starts to request PDUs from the new SRNCHigher layer retransmissions (RLC) are needed to recover the data thatwas destroyed in buffer resetIf RLC is used in unacknowledgedmode and if user application does nothave retransmission mechanisms, some data is lost forever when a handover occurs!
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
HS-DSCH to DCH HandoverNeeded when user movesfrom HSDPA capable cell to a cell that does not supportHSDPANode Bs, target cell RNC and the user are informed aboutincoming handover eventBuffers in Node B are resetas in previous case and similar retransmissions arerequired because of lostbuffer dataTransmission continues in DCH
21
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
SummarySummary –– WCDMA HSDPA WCDMA HSDPA HOsHOs
No (if RLC acknowledgedmode is used)
No (if RLC acknowledgedmode is used)
No (if MAC-hspreservation is used)
Packet losses
No bufferforwardingSRNC RLC retransmissions
No bufferforwardingSRNC RLC retransmissions
bufferforwarding
Packetretransmissions
HS-DSCH to DCH
Inter-Node BHS-DSCH to HS-DSCH
Intra-Node BHS-DSCH to HS-DSCH
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
WCDMA HSUPA handoversWCDMA HSUPA handoversHandover control in the RNC decides, which
cells are in the active setcell is the serving HSUPA cell
Definition of active set is very similar to wideband code division multiple access (WCDMA) Release 99 handover control
Maximum active set size equals 4 for HSUPA, while Release 99 UEs must support up to 6 cells in the active set.
The serving cell algorithm decides which cell is in control of the HSUPA userThe serving cell for HSUPA can be different from the serving cell in HSDPA, but typically
the serving cell for HSDPA and for HSUPA are the samethe serving cell change takes place at the same time
22
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ChapterChapter 3: 3: AdmissionAdmission and and loadload controlcontrol
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Air Interface Load MeasurementAir Interface Load MeasurementIf radio resource management is based on the interference levels in the air interface, the air interface load needs to be measured.Load estimation can be based on
wideband received powerThroughput
Uplink load estimation based on wideband received power Itotal=Iown+Ioth+PN
Itotal: received wideband powerIown: power of intra-cell usersIoth: power of inter-cell usersPN: background and receiver noise
Noise rise= Itotal/ PN = 1/(1-ηUL)ηUL=1- PN/ Itotal=(Noise rise-1)/Noise rise = UL load factor
UL load factor can be used as a UL load indicator.
23
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
Uplink load estimation based on throughputUL load factor can be calculated as the sum of the load factors of the UEs that are connected to the Node BN is the UE number in own cellW is the chip rateR_j is the bit rate of jth UE(Eb/No)_j is Eb/No of the jth UEv_j is the voice activity factor of jth UEi is the other-to-own cell interference ratio
∑= +
+=N
jjvjRjoNbE
Wi
UL 1)/(
1
1)1(η
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
ComparisonComparison of UL of UL loadload estimationestimation methodsmethods
Throughput-based threshold
Wideband power-basedthreshold
Wideband power level Itotal(Coverage)
Own cell load factor η(Capacity)
i=1.0i=0.5 i=0
Soft capacity depending on other cell loading I with wideband power based
RRM
+ Wideband power-based load estimation keeps the coverage with the planned limitsand delivered capacity depends on the loading in the adjacent cell (soft capacity). Thus, it prevents cell breathing.- Measured interference can include interference from adjacent frequencies. Thus, it may overestimate the load of own carrier due to any external interference
+Throughput-based estimation method keeps the throughput of the cell at planned level- Coverage will be reduced if the loading in adjacent cells is high+/- Soft capacity can only be obtained in RNC as RNC has the adjacent cells’ loading information
24
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
DownlinkDownlink loadload: : powerpower--basedbased estimationestimationDL load factor
Note that here the total Node B transmission power PTX_TOTAL
does not give accurate information concerning how close to the DL air interface pole capacity the system is operating.In a small cell, the same PTX_TOTAL corresponds to a higher air interface loading than in a large cell.
MAXTX
TOTALTX
PP
DL_
_=η
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
DownlinkDownlink loadload: : throughputthroughput--basedbased estimationestimationDL load factor as a sum of DL allocated bit rates:
N is the number of DL connections, including the common channelsR_j is the bit rate of the jth UERmax is the maximum allowed throughput of the cell
Alternatively, weigth UE bit rates with Eb/No’s
W is the chip rate(Eb/No)j is the Eb/No of the jth UE, needs to be assumed based on the typical values of that environmentα is the average orthogonality of the cell, can be estimated by Node B based on the multipath propagation in the uplinki is the average DL other-to-own cell ratio of the cell, can be obtained in RNC based on the adjacent cell loading
MAX
N
jj
R
RDL
∑== 1η
+−⋅=
⋅= ∑ iW
joNbEjN
jjRDL α
υη 1
/
1
25
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
AdmissionAdmission controlcontrolAdmission Control Principles
AC ensures the coverage/quality of the existing connections before accepting a new RAB and maintains network stabilityAC functionality is located in the RNCAC estimates whether a request to establish a RAB is admitted in RAN or notAC needs estimation of a load increase that a new connection would causeNew call admitted only if both UL and DL AC admits it.
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
WidebandWideband powerpower--basedbased AC AC strategystrategyIn uplink, a new UE is admitted if
The only problem is how to estimate the load increaseTwo methods for that:
the derivative methodthe integral mathod
thresholdoldtotal III <∆+_
same as max. UL noiseraise (set by RNP)
Interferencelevel
∆I Max planned noise rise set by RNP
load∆L
totalI
η
26
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
LoadLoad increaseincrease estimationestimationLoad increase estimation takes into account the derivative of load curve,
which can be calculated as follows:
ηddItotal
2)1(
1
11raise Noise
ηη
η
η
−=
⇒−
=
⇒−
==
Ntotal
Ntotal
N
total
Pd
dI
PI
PI
∆I
totalI
ηload∆L
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
DerivativeDerivative methodmethodReplacing derivative with differentials, we have
where ∆L is a load factor of the new UE. It is estimated as
Finally, this results in the following:
ηddI
LI total≈
∆∆
LILPLd
dII totalNtotal ∆−
=∆−
=∆≈∆ηηη 1)1( 2
vRNEWL
ob /1
1
+=∆ assumed values for
the new connection
∆I
totalI
ηload∆L
27
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
IntegrationIntegration methodmethodDerivative of the interference is integrated from the old loadfactor value to the new one:
LL
I
PL
L
PL
P
dP
dII
total
N
NN
L N
L
total
∆∆−−
=
−∆−−∆=
−−
∆−−=
−=
=∆
∫
∫∆+
∆+
η
ηη
ηη
ηη
η
η
η
η
1
11
11
)1( 2
∆I
totalI
ηload∆L
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
DownlinkDownlink ACACIn downlink the strategy is similar: a new UE is admitted ifthe new total DL Tx power does not exceed the predefinedtarget:
Notice that the increase of total Tx power includes both the power of the new UE requesting capacity and the additional power rise of the existing UEs in the system due to the additional interference contributed by the new UE.This total Tx power increase can be estimated based on a priori knowledge of the required Eb/No, the requested bit rate and the pilot report from the UE.
The pilot report can provide information on the path loss towards the new UE as well as the interference level experienced by the UE.
thresholdtotaloldtotal PPP <∆+_Set by RNP
28
TAMPERE UNIVERSITY OF TECHNOLOGY, Institute of Communications Engineering
LoadLoad ((congestioncongestion) ) controlcontrolPurpose: update the load status based on the measurements and estimations provided by AC and PSReturn the system quickly and controllably to/under the target load in case of an overload situation.LC functionalities locate in RNC and BTSAction taken
DL fast load control: deny DL power up commands received from the UEUL fast load control: reduce the UL Eb/No targetReduce the throughput of packet data trafficHO to another WCDMA carrierHO to GSMDecrease bit rates of RT usersDrop calls in controlled fashion