Nok
Nok
ia
ia
Sie
Sie
men
men
s Ne
s Ne
two
two
rks
rks
–
–
3G
3G
Upl
Upl
ink
ink
Optimization
Optimization
1
1 © © NNookkiia a SSiieemmeenns s NNeettwwoorrkks s 22001111 Customer ConfidentialCustomer Confidential
NSN response to Annex 6, Chapter 5 in
NSN response to Annex 6, Chapter 5 in T-Mobile Netherlands Single RAN RfQT-Mobile Netherlands Single RAN RfQ
September 2011 September 2011
Introduction
Introduction
Target Target Target Target ThThisis prpresesenentatatitionon is is inintetendndeded to to prprovovididee rerespspononsese to to AnAnnenexx 6 6 ChChapapteterr 5 5 in in T-T-MoMobibilele Nethe
Netherlandrlands Sings Single RAN RfQle RAN RfQ whewherere the Supthe Supplieplier is requester is requested to provd to provide an ovide an overvieerview ofw of measures taken to reduce the radio and
measures taken to reduce the radio and basebandbaseband/RNC resource allocation in /RNC resource allocation in a higha high smartphone
smartphone penetration penetration environmenenvironment with extret with extremely high sigmely high signaling loadnaling load.. Confidentiality
Confidentiality Confidentiality Confidentiality
2
2 © © NNookkiia a SSiieemmeenns s NNeettwwoorrkks s 22001111 Customer ConfidentialCustomer Confidential
All
All infinformormatiationon relrelateatedd to to the the NokNokia ia SieSiememens ns NetNetwoworks rks 3G 3G UplUplinkink OptOptimiimizatzationion feafeaturtures,es, fu
funcnctitiononalalitity’y’ss anand rod roadadmamapsps prpresesenentetedd in tin thihiss dodocucumementnt araree ststririctctlyly NoNokikia Sia Siememenens Nes Netwtwororksks Custo
Customermer conficonfidentdential.ial. No in
No infoformrmaatitioonn shshaallll bbee ddisisclcloosesedd to ato annyy ththirirdd papartrtyy wwitithhoouutt pepermrmisissisioonn frfromom NoNokikia Siea Siememennss Networks.
Uplink vs Downlink Traffic in Live 3G Networks (1)
Europe 2 South America Europe 1 0.25 0.20 0.15HSUPA / HSDPA daily volume ratio at selected operators
MEA
3 © Nokia Siemens Networks 2011 Customer Confidential
Source: NSN Analysis, April 2011 0.10
0.05
0
01.03.2010 01.07.2010 01.10.2010 01.01.2011 01.04.2011
• Uplink traffic volume is
15-20% of downlink
• Uplink volume is growing
faster than downlink (due to HSUPA)
Uplink vs Downlink Traffic in Example Live 3G Network
HSDPA vs HSUPA ratio 10x now and getting smaller due to higher
HSUPA penetration
HSDPA vs HSUPA + WCDMA UL ratio 6x and
stabile
Uplink vs Downlink Capacity in Theory
1.06 1.11 1.31 1.44 1.52 1.74 0.79 1.0 1.2 1.4 1.6 1.8 2.0 / H z / c e l lEvolution of HSPA efficiency
Downlink Uplink • Downlink 1.31 bps/Hz/cell • Uplink 0.33 bps/Hz/cell (0.53 with IC) • => Theoretical ratio 4x
5 © Nokia Siemens Networks 2011 Customer Confidential
0.55 0.33 0.33 0.33 0.53 0.65 0.65 0.0 0.2 0.4 0.6 . H S P A R 6 H S P A R 6 + U E e q u a l i z e r H S P A R 7 6 4 Q A M H S P A R 8 D C - H S D P A H S P A R 9 D C - H S D P A + M I M O H S P A R 1 0 Q C - H S D P A + M I M O L T E R 8 b p
Smartphones Increase Signalling Load
•
Smartphones create frequenttransmission of small packets
• More multi-RABs
due to smartphones
6 © Nokia Siemens Networks 2011 Customer Confidential
changes (DCH allocations) and RACH signalling which increases uplink interference
Starting Point for Uplink Optimization
•
In theory, the networks should be downlink limited because the traffic is 5-6x in downlinkwhile the capacity is 4x.
•
The higher uplink noise rise is mainly caused by the control overhead– RACH preambles and messages, like RRC requests, uplink capacity request and user plane data,
especially related to smartphone traffic
– DPCCH overhead. For example, with AMR5.9 kbps 64% of interference comes from DPCCH.
– DPCCH overhead from PS 0/0 kb s users
7 © Nokia Siemens Networks 2011 Customer Confidential
– HS-DPCCH overhead for HSUPA
•
It is possible to improve the situation because we are not hitting any fundamental theoreticallimit. The limit is ”only” system protocol design and configuration. There are already promising indications since RU20 ontop releases have stabilized the uplink in many networks.
•
NSN uses interference based uplink RRM while some RAN vendors use throughput basedsolution (number of users). The interference based solution has the benefit that cell breathing can be controlled. But interference based solution requires also careful control of the uplink interference sources to provide optimal performance.
NSN Solutions for Uplink Interference Control – Summary
HS-RACH Interference
cancellation Dynamic HSUPA
power offset RU20
RU30
RU40 Cell_PCH
Load aware outer
loop power control RU30
3GPP Release 99-6
HSUPA DPCCH
interpolation RU30
Fast dormancy RU20
High noise
optimized RRM1 RU20
Cell level control of
uplink parameters RU30
3GPP Release 7
Continuous packet
connectivity RU20
3GPP Release 8
8 © Nokia Siemens Networks 2011 Customer Confidential
High noise
optimized RRM2 RU20 Dynamic initial bit
rate allocation RU20
DPCCH overhead
calculation RU20 Downgrade of DCH
in SHO congestion RU20 RRC IPhone
workaround RU20
Fast BTS load control
RACH access class
barring RU40
Dynamic parameter settings
Dynamic CQI frequency
Dynamic HSUPA Power Offset
DynPwrOffsetTable2 (High/low power offset indication table for 2ms TTI)
RSCP [dBm] EcNo [dB]
<-108 <- 14 L L L L L L
•
Two sets of DPCCH offset values defined. Lower DPCCH power is used when the condition(H=High load) is fulfilled
9 © Nokia Siemens Networks 2011 Customer Confidential
-105…-108 -13…-14 L L L L L L -101…-104 -11…-12 L L L L L L -98..100 - 10 H H L L L L -98..-95 - 9 H H H L L L >-95 >-9 H H H L L L 0-1 2-3 4-6 7-12 13-20 >20 # of HSPA serving cell users
High Noise Optimized RRM1
•
Five features for optimizing the power based uplink RRM•
Correction of the filtering parameter MaxIncrInterferenceUL• Filter out the short term spikes of the measured RTWP for avoiding the unnecessary admission
control blockings during the period of the spike
•
Corrections in power increase and decrease estimations of the estimated R99 power10 © Nokia Siemens Networks 2011 Customer Confidential
•
orrec on o e power ncrease an ecrease es ma ons n e ce s•
Reference power of the management parameter DeltaPrxMaxUpInitial SIR Target Optimization
Parameters 1-RX 2-RX 4-RX nonSHO SHO beta_d / beta_c R_b SIRDPCCHInitialDCHHS256 4.5 2.5 -0.5 -1.5 1.3 2.0 SIRDPCCHInitialDCHHS128 4.5 2.5 -0.5 -1.5 1.3 2.0 1.2 16 kbps SIRDPCCHInitialDCHHS64 4.5 2.5 -0.5 -1.5 1.3 2.0 SIRDPCCHInitialDCHHS32 6.0 4.0 1.0 0.0 1.3 2.0 1.4 64 kbps SIRDPCCHInitialDCHHS16 7.5 5.5 2.5 1.5 1.0 1.6 1.7 128 kbps SIRDPCCHInitialDCHHS8 8.0 6.0 3.0 2.0 1.0 1.6 SIRDPCCHInitialDCHHS4 9.0 7.0 4.0 3.0 0.8 1.3 2.5 384 kbps SIRDPCCHInitialDCH64 4.5 2.5 -0.5 -1.5 1.2 AMR 12.2
Initial DPCCH SIR w HSDPA AmplitudeRatioACK
SIRDPCCHInitialDCHOffset -2 dB SIRDPCCHInitialDCHRxDiv2 -3 dB SIRDPCCHInitialDCHRxDiv4 -4 dB SIRDPCCHInitialDCHMax 6 dB
SFDPCCH 256
Graphs are assuming activity factor as given below: 16kbps – 63%, 64kbps – 16%, 384kbps – 3%
11 © Nokia Siemens Networks 2011 Customer Confidential UL noise rise at initial SIR with DPCCH and HS-DPCCH overhead
0 1 2 3 4 5 6 7 8 9 10 1 3 5 7 9 1 1 1 3 1 5 1 7 1 9 2 1 2 3 2 5 Number of links (DPCCH + HS-DPCCH + DPDCH) U L n o i s e r i s e ( d B ) 16 kbps 64 kbps 384 kbps
UL noise rise at initial SIR with DPCCH and HS-DPCCH overhead
0 1 2 3 4 5 6 7 8 9 10 1 3 5 7 9 1 1 1 3 1 5 1 7 1 9 2 1 2 3 2 5 Number of links (DPCCH + HS-DPCCH + DPDCH) U L n o i s e r i s e ( d B ) 16 kbps 64 kbps 384 kbps
High Noise Optimized RRM2
•
12 features for optimizing the power based uplink RRM• Emergency call failure for the power blocking
• PrxNoise autotuning only in the cell without any CELL_DCH traffic
• PrxNoise is autotuned only if all cells of the same frequency in the BTS are on the low traffic level
• Adjusting of the increased reference noise floor value in the loaded cell
• Detection of the common measurement reports filtered by BTS
12 © Nokia Siemens Networks 2011 Customer Confidential
• Candidate prioritization and bit rate selection in PBS
• R99 Overload Control procedure
• Downgrading the PS NRT DCH for the soft handover branch addition congestion handling
• PRFILE parameter control for triggering the channel type switching from the SIR error
• Limited value of UL DPCCH power offset for the first RL setup in the RTWP spiking cell
• Power based Admission Control for the HSUPA call setups
• Correction in updating the the MIN and MAX PRXTOTAL counters of the Received Rel99 wideband
Dynamic Initial Bit Rate Allocation
•
Allows more PS NRT users admitted at initial and minimum bit rates in and keep the existingPS NRT users longer in the CELL_DCH state.
•
High bit rate PS DCH users are selected first for downgrade, the QoS priority is applied onlywhen the PS calls of the cell are not using higher than the initial DCH bit rates
•
PBS candidates will be prioritized in all congestion cases as follows:• PS NRT DCHs users having higher bit rate than initial bit rate users, in the QoS priority order.
•
13 © Nokia Siemens Networks 2011 Customer Confidential
, .
• Finally the minimum bit rate users, in the QoS priority order.
•
Initial/minimum DCH bit rate selection of the PS call triggered the PBS:• New functionality applies to the UL interference, DL power and UL load congestion.
• If BRM detects congestion and the PBS triggers, then:
• If high bit rate (higher than initial) PBS candidates are available, then the incoming user gets the initial
bit rate
• If only low bit rate (lower or initial) PBS candidates are available, then the incoming user gets the
Downgrading NRT DCH in Soft Handover Congestion
•
Present implementation does not allow the downgrading of the DCH bit rates of the PSbearers if a congestion occurs in the soft handover branch addition. If the target cell is better than the ones in the active set, the failed soft handover may cause significant spiking of the RTWP, unless the PS DCH is removed.
•
In the new implementation, the PS DCH is downgraded to the minimum bit rate and thenattempted the branch addition once more.
•
If the congestion occurs still, the UE is switched to CELL_FACH state without applying the14 © Nokia Siemens Networks 2011 Customer Confidential
management parameter EnableRRCRelease. If the UE has also the AMR, the PS bearers are downgraded to DCH 0/0 as it is done already in the original implementation.
RRC IPhone Workaround
UE1 Cell1 RNC
RACH: RRC: RRC Connection Request UE starts decoding FACH
FACH: RRC: RRC Connection Setup, state indicator: Cell-DCH (Spreading code 1)
RACH: RRC: RRC Connection Request cause: protocol error UE decodes some rubbish from FACH
RNC thinks first RRC connection re uest has RL setup procedure, spreading code1
UE 2 Cell1
Solution: RNC ignores the repeated RRC connection request with protocol
error cause and wait for the RRC connection setup complete for the first
RRC connection setup.
15 © Nokia Siemens Networks 2011 Customer Confidential
failed, and releases resources, and setup resources for second RRC connection request RL deletion procedure, spreading
code 1
RL setup procedure, spreading code2
FACH: RRC: RRC Connection Setup state indicator: Cell-DCH (Spreading code 2)
RNC allocates spreading code1 to UE2 UE decodes first RRC Connection
setup message, and starts using spreading code 1 in Cell_DCH state
UE1 and UE2 decoding the same DL spreading code and TPC bits. UE1 can cause uncontrolled interference.
DPCCH Overhead Calculation
•
DPCCH overhead included in load factor estimation has too conservative value based oninitial UL SIR target. This modification multiplies the initial SIR target value with the activity factor of the signaling link DCH.
Interference Cancellation
•
Estimate the physical channel data after channel(Turbo) decoding. The physical channel data is
generated by encoding the decoded data. Large gain from channel decoding
•
Uplink throughput gain 23-62%•
SW upgrade to Flexi Rel.2 baseband17 © Nokia Siemens Networks 2011 Customer Confidential
RX
RX RAKERAKE
IC
IC RAKERAKE DECODERDECODER
DECODER
DECODER ENCODERENCODER
no PIC PIC
PIC
(uncod.) PIC
PIC
(uncod.) no PIC PIC
PIC (uncod.) PIC PIC (uncod.) 1 user 5,84 7,61 2 user 4,78 5,92 5,92 24% 24% 5,53 7,69 7,69 39% 39% 3 user 3,96 5,87 5,39 48% 36% 4,29 6,93 6,19 62% 44%
thrput Mbps trp gain thrput Mbps trp gain
Cell Level Control of Uplink Parameters
•
Some of the existing UL interference impacting parameters that are controlled in RNC level,change to cell level
•
HSDPAinitialBitrateUL•
HSDPAminallowedBitrateUL•
TrafVolPendingTimeDL18 © Nokia Siemens Networks 2011 Customer Confidential
•
TrafVolPendingTimeUL•
Prx NoiseMaxTuneAbsoluteLoad Aware of Outer Loop Power Control
•
Target: prevent too high increase of SIR target during high load•
Reason: SIR target increases if UE hits its max power•
Freeze AMR SIR targets and decrease NRT PS SIR targets until noise rise gets lower•
Potentially also decrease AMR SIR and/or increase BLER target with higher noise riseRACH Access Class Barring
•
The access classes [0,…9] which are barred are actually rotated by specified intervals.•
If during first time interval, the access classes [1,2,3] were barred, in the next time interval [4,5,6] would be barred covering access classes 0,…,9, i.e. rotation by mod 10. Rotation time needs to be long enough.-0.1 0
Fast BTS Load Control
•
Reduce SIR target if noise rise exceeds PrxTarget + offset (2 dB)(
)
(
)
10 _ , _ arg , , arg , arg arg , Offset P rise Noise SIR SIR Offset P rise Noise If et t rx RNC et t BTS et t et t rx − − − = + >21 © Nokia Siemens Networks 2011 Customer Confidential 0 2 4 6 8 10 12 14 16 18 20 -1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 Noise rise [dB] S I R t a r g e t c o r r e c t i o n [ d B ] PrxTarget 8 dB SIR target adjustment
Dynamic Parameter Setting
•
High interference cases can be solved by using suitable timer and other parameters duringmass events. Some of those parameters are not good for the non-congested cells. Therefore, the parameters should be automatically tuned according to the instantaneous load.
•
Example parameters• WaitTimeRRC
•
Dynamic CQI Frequency
•
Channel Quality Information (CQI) frequency is 4 ms currently•
CQI frequency could be lowered during high uplink load to minimize the interference in thesame way as DPCCH offset values are optimized
•
Current CQI interference contribution with 4 ms period = 0.41 x DPCCH. If we would lowerCQI frequency to 10 ms or 20 ms, the interference would reduce to 0.08..0.16 x DPCCH.
•
The total uplink interference from HSDPA users without any uplink activity would be reduced- =
23 © Nokia Siemens Networks 2011 Customer Confidential
. . A/N CQI A/N CQI DPCCH HS-DPCCH DPCCH
A/N CQI CQI power offset –2 dB in
single link and +4 dB in SHO compared to DPCCH
Continuous Packet Connectivity
•
Uplink DPCCH and E-DPCCH gatingreduces interference especially for low data rate users
•
Gating is part of Continuous packetconnectivity (CPC). It is part of 3GPP Release 7 400 500 600 700 800 900 1000 1100 1200 1300 l l T h r o u g h p u t ( k b p s ) PedA_not gated PedA_9/15 400 500 600 700 800 900 1000 1100 1200 1300 l l T h r o u g h p u t ( k b p s ) PedA_not gated PedA_9/15 PedA_not gated PedA_9/15
24 © Nokia Siemens Networks 2011 Customer Confidential
(E-)DPCCH E-DDCH Web page
download User readingweb page
0 100 200
0 5 10 15 20 25 30 35 40 45 50 55 Number of no-data UEs in CELL_DCH
C _ PedA_12/15 PedA_9/15 ideal PedA_12/15 ideal 0 100 200 0 5 10 15 20 25 30 35 40 45 50 55 Number of no-data UEs in CELL_DCH
C _ PedA_12/15 PedA_9/15 ideal PedA_12/15 ideal _ PedA_12/15 PedA_9/15 ideal PedA_12/15 ideal
HSUPA DPCCH Interpolation
•
Release 7 solution allows to minimize DPCCH overhead for low data rate HSUPA users•
Fixed DPCCH power in Release 6 leads typically to too high DPCCH overhead at low kbpsOptimised gain factors
E-TFC Pred
Relative power of E-DPDCHs over DPCCH 2 ms TTI
16 18
25 © Nokia Siemens Networks 2011 Customer Confidential 32 6 64 7.1 128 8.1 256 8.9 384 9.9 512 8.1 768 8.1 1024 6 1450 6 1920 6 2900 7.1 3800 8 4 6 8 10 12 14 0 2 5 6 5 1 2 7 6 8 1 0 2 4 1 2 8 0 1 5 3 6 1 7 9 2 2 0 4 8 2 3 0 4 2 5 6 0 2 8 1 6 3 0 7 2 3 3 2 8 3 5 8 4 3 8 4 0 4 0 9 6 Data rate [kbps] P ( E - D P D C H s ) [ d B ] Computed P(E-DPDCHs) Optimal P(E-DPDCHs)
Fast Dormancy
Other vendors
2 s IDLE
IDLE
DCH/HSPA inactivity timer
CELL_FACH inactivity timer
30 signaling messages
Heavy signaling load Low battery life time
Idle = <5 mA Cell_PCH = <5 mA Cell_FACH = >100 mA Active = >200 mA IDLE Fast Dormancy to save battery
26 © Nokia Siemens Networks 2011 Customer Confidential
Nokia Siemens Networks <0.3 s 3 signaling messages PCH PCH DCH/HSPA inactive CELL_FACH inactive 12 signaling messages Idle = <5 mA Cell_PCH = <5 mA Cell_FACH = >100 mA Active = >200 mA
Network avoids signaling storm Battery lasts longer
HS-RACH
•
HS-RACH allows carrying medium size data packets without allocation of dedicated resourceskB per DCH or HS-DSCH Allocation 315 199 180 200 220 240 260 280 300 320
•
The avarage data volume perallocation is typically 10-60 kB for the smartphones. The median value is even smaller: 60% of
allocations are below 1 kB ⇒ large
part of smartphone traffic could be
27 © Nokia Siemens Networks 2011 Customer Confidential 47 25 4650 6868 3238 2 34 14 44 21 2 155 1116 117 12 8 100 65 2034 6 14 131123205 1 2 4 3 6 3 78 17 94 1 8 15 53 0 20 40 60 80 100 120 140 A p p l e i P a d 3 G ( A 1 3 3 7 ) A p p l e i P h o n e 3 G ( A 1 2 4 1 ) A p p l e i P h o n e 3 G A 1 2 4 1 A p p l e i P h o n e 3 G S - A 1 3 0 3 A p p l e i P h o n e 3 G S ( A 1 3 0 3 ) A p p l e i P h o n e 3 G S A 1 3 0 3 A p p l e i P h o n e 4 ( A 1 3 3 2 ) A S U S T e k N ³ v i f o n e A 5 0 H T C D R E A 1 1 0 H T C D e s i r e H T C H e r o H T C L e g e n d H T C W i l d f i r e H T C M A P L 1 1 0 H T C P B 9 2 2 0 0 H T C P B 9 9 1 0 0 H T C P B 9 9 2 1 0 H T C P D 9 8 1 0 0 H u a w e i E 1 6 0 E H u a w e i E 1 6 9 H u a w e i E 1 7 6 2 L G G T 5 4 0 M o t o r o l a M B 5 0 1 , M E 5 0 1 N o k i a E 5 - 0 0 N o k i a E 6 3 - 1 N o k i a E 7 1 - 1 N o k i a E 7 2 - 1 N o k i a N 9 7 - 4 N o k i a N o k i a E 6 3 - 1 N o k i a N o k i a E 7 1 - 1 N o k i a N o k i a E 7 2 - 1 N o k i a N o k i a N 8 - 0 0 Q i s d a S t r e a k R I M 9 0 0 0 R I M 9 3 0 0 R I M 9 7 0 0 R I M 9 7 0 0 ( G e n e r i c ) R I M 9 7 8 0 R I M B l a c k b e r r y 9 8 0 0 S a m s u n g G T - I 5 8 0 0 S a m s u n g G T - I 8 7 0 0 S a m s u n g G T - I 9 0 0 0 S a m s u n g S a m s u n g G T - I 9 0 0 0 S i e r r a W i r e l e s s M C 8 7 7 5 V S o n y E r i c s s o n E 1 0 i S o n y E r i c s s o n X 1 0 i S o n y E r i c s s o n X 1 0 i carried by HS-RACH
•
HS-RACH reduces controloverhead considerably
•
No setup signalling•
Immediate stop of controlchannel transmission
compared to DCH with >1 second timer
Summary
•
3G networks have turned to be uplink limited due to interference limited nature of CDMAuplink. The main problems come from the control channel and signalling overhead which is driven by increased smartphone traffic and HSUPA high data rates
•
NSN has introduced a large number of features in RU20 and RU30 to improve the uplinkperformance. The features have already shown to be highly useful in the practical networks.
28 © Nokia Siemens Networks 2011 Customer Confidential