umts hspa datatransmission performance improvement
TRANSCRIPT
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 1/43
HUAWEI TECHNOLOGIES CO., LTD.
www.huawei.com
Huawei Confidential
Security Level:2014/2/25
HSPA Data Transmission
Performance Improvement
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 2/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 2
HSDPA性能研究与华为解决方案汇报
HSDPA Basic Concepts and Process of Identifying
Data Transmission Problems
HSUPA Basic Concepts and Process of Identifying
Data Transmission Problems
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 3/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 3
Basic Concepts — Protocol Structure of HSDPA User Plane
The data transmission protocol layers consist of the physical layer, MAC layers (MAC-hs/ehs
layer and MAC-d layer), RLC layer, PDCP layer, TCP/IP layer, and application layer. Figure 2-1 shows
the involved NEs and the relationships among layers.
On the RAN side (excluding the UE), the physical layer, MAC layer, and RLC layer are involved.
The TCP/IP layer is adjacent to the PDCP layer and the RLC layer. Therefore, the TCP/IP layer may
also be affected by the RAN in some scenarios
FTP servers, streaming
servers, and websites.
PDCP Packet Data Convergence Protocol
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 4/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 4
Basic Concepts — Data Frame Structure of HSDPA User Plane Protocol
The rate at each layer is classified into the PDU rate and the SDU rate.
The PDU rate includes the overhead of the current layer, whereas the SDU rate does not include the
overhead of the current layer.
Therefore, the SDU rate of a layer equals to the PDU rate of the upper layer.
RLC RLC RLC RLC RLC RLC
RLC SDU
PayloadRLCHeader
MAC-d
MAC-c
MAC-d
MAC-ehs
Iub-FP Iub-FP Iub-FP
PayloadRLCHeader
PayloadRLC
Header
U-RNTI
PayloadRLCHeader
U-RNTIMAC-ehsHeader
(s)RBs for UE1 (s)RBs for UE2 PDU structure
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 5/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 5
Basic Concepts —CQI Report Principles
The UE measures the Ec/No of the Common Pilot Channel (CPICH) and adds aMeasurement Power Offset (MPO) as the Ec/No estimation value of the HS-PDSCH.
That is, the UE assumes that the NodeB transmits the HS-PDSCH according to CPICHPower + MPO.
Then, add the SF gain 10 * log16 to obtain the SNR of the HS-PDSCH.
SNR HS-PDSCH = HS-PDSCH Ec/No + SF gain 10 * log16
HS-PDSCH Ec/No (estimation value) = Ec/No cpich + MPO
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 6/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 6
The CQI is obtained according to the relationship between the SNR and CQI of the
simulated HS-PDSCH.
Generally, the difference between the CQI and SNR is a constant (4 or 4.5 dB).
CQI formula (for the HSDPA excluding 64QAM) :
The previous formula shows that CQI increase by 1 when SNR HS-PDSCH increase by 1.
With 64QAM: when the CQI is greater than 25, CQI increases by 1 when SNR increases by 2 dB.
In this case:
Basic Concepts —CQI Report Principles
CQI = SNRcpich + MPO + 4.5 if CQI <= 25
or CQI = 25 + (SNRcpich + MPO + 4.5 - 25)/2 if CQI > 25
CQI = CPICH Ec/No + MPO + 10 * log16 + 4.5
= SNRcpich + MPO + 4.5
= SNRhs-pdsch (based on assumed power) + 4.5 (depending on the UE implementation)
MPO = min(13, Pcell-Pcpich - MPO constant) dB The MPO constant is 2.5 by default.
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 7/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 7
1.终 端能力
HS-DSCH category
Maximum
number of HS-
DSCH codes
received
Minimum
inter-TTI
interval
Maximum number of
bits of an HS-DSCH
transport block
received within
an HS-DSCH TTI
Supported modulations
without MIMO
operation
or dual cell operation
Supported
modulations with
MIMO operation
and without dual
cell operation
Supported
modulations
with dual cell
operation
Category 1 5 3 7298
QPSK, 16QAM
Not applicable
(MIMO not
supported)
Not applicable (dual
cell operation not
supported)
Category 2 5 3 7298
Category 3 5 2 7298
Category 4 5 2 7298
Category 5 5 1 7298
Category 6 5 1 7298
Category 7 10 1 14411
Category 8 10 1 14411
Category 9 15 1 20251
Category 10 15 1 27952
Category 11 5 2 3630QPSK
Category 12 5 1 3630
Category 13 15 1 35280QPSK, 16QAM, 64QAM
Category 14 15 1 42192
Category 15 15 1 23370QPSK, 16QAM
Category 16 15 1 27952
Category 17 15 135280 QPSK, 16QAM, 64QAM –
23370 – QPSK, 16QAM
Category 18 15 142192 QPSK, 16QAM, 64QAM –
27952 – QPSK, 16QAM
Category 19 15 1 35280QPSK, 16QAM, 64QAM
Category 20 15 1 42192
Category 21 15 1 23370
- -
QPSK, 16QAM
Category 22 15 1 27952
Category 23 15 1 35280
QPSK, 16QAM,64QAMCategory 24 15 1 42192
Basic Concepts — UE Capabilities
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 8/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 8
The code assignment algorithm is involved in both RNC and NodeB.
RNC:Manual assignment : set the number of codes to be assigned.
Automatic assignment : set the maximum number of codes and minimum number of
codes to be assigned.
NodeB: Enable or disable the NodeB dynamic code function.
We will introduce different combinations of the two algorithms. The number of channels
used by the cell is set by default (four HS-SCCHs with HSUPA activated):
• Manually assign five codes on the RNC
•
Disable the dynamic codes on the NodeB :=> The HS-DSCH uses a maximum of 5 SF16 and a minimum of 5 SF16.
• Enable the dynamic codes on the NodeB:
=> The HS-DSCH uses a maximum of 14 SF16 and a minimum of 5 SF16
Basic Concepts— Code Assignment (1/3)
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 9/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 9
Basic Concepts — Code Assignment (2/3)
• Automatically assign a maximum of ten codes and a minimum of five codes on the RNC
• Disable the dynamic does on the NodeB:
=> The HS-DSCH uses a maximum of 10 SF16 and a minimum of 5 SF16.
• Enablethe dynamic does on the NodeB:
=> The HS-DSCH uses a maximum of 14 SF16 and a minimum of 5 SF16.
Over one SF16 codes are used by the Common Control Channels
=> maximum of 14 SF16 can be used for HSDPA
• The policy of RNC manual assignment + NodeB dynamic code enabled is
recommended for the existing network.
• RNC automatic assignment + NodeB dynamic code disabled is recommended if
networks do not support the NodeB dynamic codes.
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 10/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 10
To ensure that 15 codes can be used by the accessed DPA user, you need to modify the
configurations of the channels (Eg disable HSUPA, only 2 HS-SCCHs in one TTI..)
SRB over DCH : Each accessed HSDPA user consumes an associated DPCCH using one SF256SRB over HSDPA : F-DPCH is multiplexed by all users => save codes
Basic Concepts — Code Assignment (3/3)
Default configuration for codes
usage by common channels in the
• The CCH uses one SF32,
• Four HS-SCCHs, each one use one
SF128 => one SF32,
• The E-RGCH and E-HICH
multiplexes one SF128,
• The E-AGCH uses one SF256.
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 11/43
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 12/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 12
Basic Concepts — EL2 Principles (2/3)
64QAM, MIMO, 64QAM+MIMO, and DC+64QAM technologies are introduced in R8.
The theoretical peak rate of service data transmission can reach 21.6 Mbit/s, 28.8
Mbit/s, and 43.2 Mbit/s respectively.
• If the RLC PDU size is too large, no complete RLC PDU can be correctly transmitted
to the UE when the channel quality is poor; therefore, normal data transmission cannot
be performed and the coverage is reduced.
• If the RLC PDU size is too small and the RLC window size is increased to improve the
data transmission rate, many RLC PDUs are multiplexed in one MAC-hs PDU when the
user channel quality is good; therefore, much redundant information is brought in the
RLC PDU data heads, and the data transmission efficiency of the air interface is
reduced.
To fully match the data transmission capability of the air interface, improve the data
transmission efficiency, the L2 enhancement is introduced.
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 13/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 13
The L2 enhancement mainly affects RLC layer, MAC layer, and user plane FP data transmission of Iub
interface on the RNC.
• The variable PDU size of downlink AM is introduced at the RLC layer.
• The MAC-ehs is introduced at the MAC layer.
•The 64QAM, MIMO, and DC-HSDPA features must be supported by EL2 enhancement.
Basic Concepts — EL2 Principles (3/3)
The maximum RLC PDU is set to 302
bytes (2416 bits) in EL2 by default
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 14/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 14
Basic Concepts — Theoretical Rates of Layers (1/2)
Relationships among the throughput of layers :
• TCP/IP layer: Data packets at this layer are received from the application layer. The
TCP/IP layer matches data packets based on the MTU (Maximum Transfer Unit ) size.
Generally, the MTU size is 1,500 bytes, which is equal to 12,000 bits. The header
overhead of 40 bytes (320 bits) is added to each MTU.
•
PDCP layer: Data transmission is considered as transparent transmission at this layer,and therefore no overhead is added.
• RLC layer: This layer matches data packets of the PDCP layer based on the RLC SDU
size. The overhead of 16 bits is added to each SDU to form an RLC PDU. Then, the RLC
PDU is transmitted to the MAC-d layer.
• MAC-d layer:Data transmission is considered as transparent transmission at this layer,
and therefore no overhead is added.
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 15/43
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 16/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 16
Basic Concepts — Calculation of Theoretical Rates (1/4)
Example for UE CAT14 :
CAT14 supports EL2 and 64QAM (MAC-ehs entity)
The maximum TBS used in CAT14 is 42192 bits
1) The maximum rate of the MAC-hs layer is: 42192 bits /2 ms = 21.096 Mbit/s
2) At the RLC layer (EL2) :
- The maximum RLC PDU is set to 302 bytes (2416 bits) in EL2 by default
- If 42192 (minus fixed 8 bits overhead for EL2) => 42184 bits
- Nb of RLC PDU (Mac-ehs SDU) => int (42184/2416) = 17 PDU
- 17 PDUs of 2416 bits can be carried and 16-bit overheads are introduced (EL2).
42184 bits – (2416 bits * 17) – (16 * 17) = 840 bits.
- After subtracting the introduced 16-bit overhead, the PDU that can be carried is 824 bits.
The maximum PDU rate at the RLC layer is: (2416 bits * 17 + 824 bits)/2 ms = 20.95 Mbit/s
Maximum number of bits of an HS-DSCHtransport block received within
an HS-DSCH TTI
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 17/43
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 18/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 18
Downlink UE Throughput During RNC Real-Time Performance Tracing
Real-time performance tracing from RNC give the Downlink UE throughput at the MAC-d layer(that is, the SDU rate at the MAC-hs layer) or the PDU rate at the RLC layer.
Downlink Throughput Displayed in DU Meter (monitor TCP IP throughout)
Downlink throughput displayed in DU meter can be considered as the SDU rate at the RLC rate.
Basic Concepts — Calculated of Theoretical Rates (3/4)
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 19/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 19
IBLER / SBLER/ RBLER
• The 1st BLER is called initial BLER (IBLER).
• SBLER indicates the sum BLER, regardless of the initial TBS transmission or HARQ retransmission.
• Residual BLER (RBLER) indicates the BLERs for TBs still incorrectly transmitted after all the HARQ
retransmissions at the MAC-hs layer.
Example:
The maximum number of HARQ retransmission is 4, three TBSs must be transmitted:• The first TB is correctly transmitted during the initial transmission.
• The second TBS is retransmitted correctly after an initial transmission.
• The third TB is retransmitted incorrectly for four times.
• IBLER, the initial transmission of the first TB is successful and the initial transmissions of the second and
third TBs fail. Therefore, IBLER = 2/3 = 66.67%.
•
SBLER, the first TBS is transmitted only once with zero transmission failure. The second TBS istransmitted twice with one transmission failure. The third TBS is transmitted five times with transmission
failure five times. Therefore, SBLER = (0 + 1 + 5) /(1+2+5) = 75%.
• RBLER, the first TB is correctly transmitted during the initial transmission. The second TB is retransmitted
correctly after an initial transmission. The third TB fails to be retransmitted for four times. Therefore,
RBLER = 1/3 = 33.33%.
Basic Concepts — Calculation of Theoretical Rates (4/4)
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 20/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 20
Process of identifying stationary test-based HSDPA
data transmission problemsIdentify stationary
test-based HSDPA data
transmission problems
END
NCheck and
clear alarms
3.2.3
Check downlink
power resources
3.2.8
Is the problem
identified?
Is the problemidentified?
Y
N
N
Check cells’
HSDPA status
3.2.4
Check access
signaling
3.2.5
Check
licenses
3.2.6
Check the
DCCC setting
3.2.7
Check
downlink code
resources
3.2.9
Check radio
quality
3.2.11
Check packetloss on the Iub
interface
3.2.15
3.2.12
Collect data andreport problems
3.2.19
Is the problem identified?
Y
Contact CN
engineers to locate
the fault, which
must be supported
by the RAN
Y
N
Dial-up succeeds while
transmission fails
Identify who
transmission fails
3.2.1
Can data transmission be started
Is data transmission OK?
Y
Y
N
N
NY
RAN problems?
Replace the UE or
driver program
Y
N
Y
Is the problemidentified?
Check the bandwidth on the lub interface
3.2.13
Check RAN problems3.2.2
Check bandwidth
on the IU-PSinterface
3.2.14
Check packet
losses on the IU-
PS interface andTCP mechanism
3.2.16
Check andisolate UE faults
3.2.17
Check CPU
usage of laptops
3.2.18
Y
N
Check the number of
online users in a cell
3.2.10
Check if RLC downlink
window is full
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 21/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 21
Process of identifying Drive Test test-based HSDPA
data transmission problemsIdentify DT test-based HSDPA
data transmission problems
Does calldrop or handover failure (or
not timely) exist
Does abnormal
point exist (with high CQI but
low throughput)
Is the problem solved END
Check and clearalarms, especially,cell-level alarms
3.2.4
Check cells’ HSDPA status
3.2.5
Check the license,especially the
NodeB-level license
3.2.7
Check downlinkcode resources
in each cell
3.2.11
Check the bandwidthof the lub interface on
each NodeB
3.2.14
Collect data and report
problems
3.2.19
Y
Y
N
Perform static test of data transmission on a near point,
check whether anyproblem exists
Handle with the
problem
Check downlinkpower resources
in each cell
3.2.10
Handle with the
problem
Handle with the
problem
Perform DT in idle
period (with few
online users)
Optimize RF coverage
and improve the
average CQI
Y
Is the problem solved
N
N
Y
N
Y
N
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 22/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 22
Flowchart of analyzing HSDPA cell performance problems
Y
Does the
average single-user throughput reach
the required BER on the airinterface?
The item cannot be
evaluated at present.
Identify HSDPA
cell performanceproblems
Is the power
usage high?YIs the CQI
poor?Y
Optimize the
coverage
There is an
optimization scheme.
Are there
many UEs?
N
Perform
expansionY
Is the Iub
transmission usage
high?
Is the RLC
retransmission ratio
high
N
YPerform Iub
expansion
Y
Is the
transmission quality of
the Iub path poor?
Optimize the
transmissionY
N
Is the residual
BER on the air interfacehigh?
Is the code usage
high?
N
Check power
control parameters Y
N
UE problems
YPerform
expansion
N
Is upper-layer
data insufficient?
Other problemsN
Global or upper-
layer problemsY
END
N
N
Is the BER
on the air interface
high?
N
Y
Does the theoretical cell rate
meet the requirement?Y
N
END
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 23/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 23
Analysis on HSDPA cell performance problemsStep Evaluation Item Evaluation Result and Handling Suggestion
High Low
(1)Bit error rate (BER) on the airinterface in the cell Optimize the coverage. Go to step (2).
(2)
Power usage of the cell
Perform the following operations based on the CQI:
If the CQI is poor, optimize the coverage.
If the CQI is normal, add carriers.
Go to step (3).
(3)Usage of the Iub transmission
bandwidthExpand the Iub transmission bandwidth. Go to step (4).
(4)
RLC retransmission rate
Perform the following operations based on the IP pathtransmission quality on the Iub interface:
If the transmission quality is poor, optimizetransmission.
If the transmission quality is normal, check the residual bit errors on the air interface. For the cells with many bit errors on the air interface, check power control parameters.
Go to step (5).
(5)
Code resource usage Add code resources.
Check whether thetheoretical rate of the cellmeets the requirement. Ifthe theoretical rate meetsthe requirement, the upper-layer data sources areinsufficient.
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 24/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 24
HSDPA性能研究与华为解决方案汇报
HSDPA Basic Concepts and Process of Identifying
Data Transmission Problems
HSUPA Basic Concepts and Process of IdentifyingData Transmission Problems
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 25/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 25
Basic Concepts — Protocol Structure of HSUPA User Plane
Figure below shows the protocol structure of HSUPA data services. Different from the R99
user plane, a MAC entity MAC-e/es is added at the MAC-d layer on the HSUPA user plane.
PHY PHY
EDCH FP EDCH FP
Iub UE NodeB Uu
DCCH DTCH
TNL TNL
DTCH DCCH
MAC-e
SRNC
MAC-d
MAC-e
MAC-d
MAC-es /MAC-e
MAC-es
Iur
TNL
TNL
DRNC
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 26/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 26
Basic Concepts — Data Frame Structure of HSUPA User Plane Protocol
MAC-es header overhead:
• TSN : Transmission Sequence Number (6 bits)
MAC-e header overhead:
• DDI : Data Description Indicator (6 bits)• N: Number of Mac-d PDU in a logical channel (6 bits)
• SI system information : SI bits and padding bits in the end
MAC-d Flows
MAC-es PDUMAC-e header
DCCH DTCH DTCH
HARQprocesses
Multiplexing
DATA
MAC-d DATA
DATA
DDI N Padding
(Opt)
RLC PDU:
MAC-e PDU:
L1
RLC
DDI N
Mapping info signaled over RRC
PDU size, logical channel id, MAC-d flowid => DDI
DATA DATA
MAC-d PDU:
DDI
Header
MAC-es/e
NumberingMAC-es PDU: TSN DATA DATANumbering Numbering
1.Simplified architecture showing MAC inter-working in UE. The left part shows the functional split while the right part shows PDU construction
MAC-d PDU MAC-d PDU MAC-d PDU
MAC-es SDUMAC-es SDUTSN1 N1DDI1 MAC-es SDU
MAC-d PDUs coming from one Logical Channel
N1 MAC-es SDUs of size and LCh indicated by DDI1
MAC-es PDU1
DDI1 N1 DDI2 N2
DDI1 N1 DDI2 N2 DDIn Nn DDI0(Opt)
MAC-es PDU1 MAC-es PDU2 MAC-es PDUn
MAC-es PDU2 MAC-es PDU1 DDIn Nn MAC-es PDUn
MAC-e PDU
SI(Opt)
Padding(Opt)
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 27/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 27
Basic Concepts — Factors Affecting the Uplink Load
1.Simplified architecture showing MAC inter-working in UE. The left part shows the functional split while the right part shows PDU construction
Number of Users in the Cell
• Check the number of users in the cell in real time on the RNC LMT.
The uplink load is shared by all users. When the number of users increases, the available load of each
user is reduced, therefore affecting the throughput.
On the other hand, the HSUPA scheduling can only control the load of the HSUPA scheduling users.
If many non-HSUPA scheduling users exist in the cell, the available load of the HSUPA scheduling
user is affected, therefore affecting the actual rate of the HSUPA scheduling users.
• If the actual load is less than or equal to 75 , the HSUPA user throughput should be
higher than to MAX(GBR, 1 RLC PDU rate).
• If the actual load is greater than 75 and less than 95 , the HSUPA user throughput is
equal or les than MAX(GBR, 1 RLC PDU rate).
• If the actual load is greater than 95 , the HSUPA user throughput is hard to reach the
GBR but may meet 1 RLC PDU rate.
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 28/43
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 29/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 29
Basic Concepts — Factors Affecting the Uplink Load
External Interference :
External interference involves neighboring interference and foreign interference
If a burst of interference occurs, the load of the cell rises instantly causing the throughput fluctuation of
HSUPA users.
=> Check the RTWP of the cell when no user exists or identify the problem by using frequency
sweep.
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 30/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 30
Basic Concepts — Uplink CE Consumption Rules
Direction Rate
(kbit/s) SF
Number of CEs
Consumed
Corresponding
Credits Consumed
UL 8 64 1 2
UL 16 64 1 2
UL 32 32 1 2
UL 64 32 1 2
UL 128 16 2 4
UL 144 16 2 4
UL 256 8 4 8
UL 384 4 8 16
UL 608 4 8 16
UL 1450 2SF4 16 32
UL 2048 2SF2 32 64
UL 2890 2SF2 32 64
UL 5760 2SF2+2SF4 48 96
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 31/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 31
E-DCHCategory
Max. CapabilityCombination
E-DCH TTIMAX EDCHTBS (10ms)
MAX EDCHTBS (2ms)
Max. Data Rate (Mbit/s)
MAC Layer
10 ms TTI
MAC Layer
2 ms TTIAir Interface
Category 1 1 x SF4 10 ms only 7110 0.71 - 0.96
Category 2 2 x SF4 10 ms and 2 ms 14484 14000 1.44484 1.40 1.92
Category 3 2 x SF4 10 ms only 14484 1.44484 - 1.92
Category 4 2 x SF2 10 ms and 2 ms 20000 289000 2.0 2.89 3.84
Category 5 2 x SF2 10 ms only 20000 2.0 - 3.84
Category 6 2 x SF4 + 2 x SF2 10 ms and 2 ms 20000 57400 2.0 5.74 5.76
Category 7 2 x SF4 + 2 xS F2 10 ms and 2 ms 20000 1150000 2.0 11.50 11.52
Category 8 2 x SF4 + 2 xS F2 2 ms 20000 1150000 2.0 11.50 11.52
Category 9 2 x SF4 + 2 xS F2 2 ms 20000 2300000 2.0 23.00 23.04
To support HSUPA, 3GPP TS 25.306 defined nine UE categories. These UEs support different peak
rates at the MAC layer, ranging from 711 kbit/s to 23 Mbit/s.
Basic Concepts calculated of Theoretical Rates
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 32/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 32
Basic Concepts — calculated of Theoretical Rates
RLC PDU UL throughput :
= Total size of all RLC PDUs transmitted at the RLC layer within a measurement period/ Measurement period
• Total size of all RLC PDUs transmitted at the RLC layer :
- involves the PDUs transmission and retransmission.
- Data transmitted by the MAC-d layer including the header overhead of the
RLC PDU with total size of 16 bits.
• Measurement period : all the time whether data is transmitted or not.
Relationship between the RLC PDU UL throughput and the MAC-e PDU available rate:
RLC PDU UL throughput = MAC-e PDU available rate * (1 - MAC-e PDU header
overhead ratio)
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 33/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 33
RLC SDU Throughput UL
= Total size of all RLC SDUs transmitted at the RLC layer within a measurement period /Measurement period
• Total size of all RLC SDUs transmitted at the RLC layer :
= Total bits of RLC PDUs – Σ of retransmitted bits and RLC PDU header overhead (16bits in total).
The relationship between the RLC SDU Throughput UL and the RLC PDU Throughput UL :
RLC SDU throughput UL ≈“RLC PDU UL throughput” * “(1 - RLC PDU retransmission rate UL)” * “RLC PDU
header overhead ratio.”
Uplink Throughput involved in RNC Radio Performance Monitoring
MAC SDU rate (the input rate at the MAC layer) also called the RLC PDU rate (the output rate at the RLC layer) involves
the retransmitted data at the RLC layer.
HSUPA CAT3, MAC SDU size = 336 bits:
MAC-d SDU rate = int(14484/336) * 336/10 = 1.4448 Mbit/s
HSUPA CAT6, MAC SDU size = 336 bits:
MAC-d SDU rate = int(11484/336) * 336/2 = 5.712 Mbit/s
Basic Concepts — calculation of Theoretical Rates
MAC-d SDU rate = (TB-Size * Number of TBs) / TTI
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 34/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 34
Basic Concepts — HSUPA Theoretical Rates of Layers
CAT3:
1) Throughput at the physical layer: (3840000/4) * 2 (SF4) = 1.92 Mbit/s
2) Data rate at the MAC-e layer: Maximum TBS at the MAC-e layer /10 ms
= 14484 * 1000/10 ms = 1.448 Mbit/s
Number of MAC-d PDUs per MAC-e PDU = Maximum TBS at the MAC-e layer (bits) / MAC-d PDU size(bits)
Number of MAC-d PDUs (MAC-d PDU size = 336 bits) = int(14484/336) = 43,
RLC payload rate : CAT3 theoretical rate = (RLC payload size * Number of TBs) / TTI
RLC SDU size = MAC-d PDU - MAC header - RLC header = 336 -16 = 320 bits
CAT3 theoretical rate (MAC-d PDU size is 336 bits) = (320 * 43) /10 = 1.376 Mbit/s.
The MAC SDU rate monitored on the LMT = int(14484/336) * 336/10 = 1.448 Mbit/s.
The maximum throughput at the application layer ≈ the RLC payload rate / (1 + 10% of HARQ retransmission)
= 1.36 Mbit/s
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 35/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 35
CAT5:
Throughput at the physical layer : (3840000/2) * 2 (SF2) = 3.84 Mbit/sData rate at the MAC-e layer: Maximum TBS at the MAC-e layer * 1000/10 ms => 20000 * 1000/10 = 2 Mbit/s
Number of MAC-d PDUs per MAC-e PDU = Maximum TBS at the MAC-e layer (bits)/MAC-d PDU size (bits)
Number of MAC-d PDUs = int(20000/336) = 59
CAT5 theoretical rate (RLC payload rate) = (RLC payload size * Number of TBs)/TTI
RLC SDU size = MAC-d PDU - MAC header - RLC header = 16 bits
If the MAC-d PDU size is 336 bits, the CAT5 theoretical rate is 59 * 320/10 = 1.888 Mbit/s
The MAC SDU rate monitored on the LMT is: int(20000/336) * 336/10 = 1.9824 Mbit/s.
The maximum throughput at the application layer ≈ the RLC payload rate/(1 + 1% of HARQ retransmission) = 1.87 Mbit/s
CAT6 SRB OVER E-DCH :
Throughput at the physical layer: (3840000/2) * 2 (SF2) + (3840000/4) * 2 (SF4) = 5.76 Mbit/s
Data rate at the MAC-e layer: Maximum TBS at the MAC-e layer * 1000/2 ms => 11484 * 1000/2 = 5.742 Mbit/s
Number of MAC-d PDUs per MAC-e PDU = Maximum TBS at the MAC-e layer (bits)/MAC-d PDU size (bits)
Number of MAC-d PDUs = int(11484/336) = 34
CAT6 theoretical rate (RLC payload rate) = (RLC payload size * Number of TBs)/TTIRLC SDU size = MAC-d PDU - MAC header - RLC header = 16 bits
If the MAC-d PDU size is 336 bits, the CAT6 theoretical rate is 34 * 320/2 = 5.44 Mbit/s.
The MAC SDU rate monitored on the LMT is: int(11484/336) * 336/2 = 5.712 Mbit/s.
The maximum throughput at the application layer ≈ the RLC payload rate/(1 + 10% of HARQ retransmission) = 4.945 Mbit/s
Basic Concepts — HSUPA Theoretical Rates of Layers
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 36/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 36
CAT6 SRB OVER DCH :
Throughput at the physical layer: (3840000/2) * 2 (SF2) = 3.84 Mbit/s
Data rate at the MAC-e layer: Maximum TBS at the MAC-e layer * 1000/2 ms => 5772 * 1000/2 = 2.886 Mbit/s
Number of MAC-d PDUs per MAC-e PDU = Maximum TBS at the MAC-e layer (bits)/MAC-d PDU size (bits)
Number of MAC-d PDUs = int(5772/336) = 17
CAT6 theoretical rate (RLC payload rate) = (RLC payload size * Number of TBs)/TTI
RLC SDU size = MAC-d PDU - MAC header - RLC header = 16 bits
If the MAC-d PDU size is 336 bits, the CAT6 theoretical rate is 17 * 320/2 = 2.72 Mbit/s.
The MAC SDU rate monitored on the LMT is: int(5772/336) * 336/2 = 2.856 Mbit/s.
The maximum throughput at the application layer ≈the RLC payload rate/(1 + 10% of HARQ retransmission) = 2.473 Mbit/s
Basic Concepts — HSUPA Theoretical Rates of Layers
Process of Identifying Stationary Test-Based HSUPA
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 37/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 37
Process of Identifying Stationary Test-Based HSUPA
Problems Identify stationary test-based HSUPAdata transmission problems
Dial-up succeeds while datatransmission fails
Identify data
transmission failure
3.2.1
Is data transmission OK?
Check and
clear alarms
3.2.4
Check the minSF during the linksetup or re-configuration
3.4.1
Check UE
capabilities
3.4.2
Check cell
capabilities
3.4.3
Check the Assigned rate
of the CN
3.4.4
Check RANparameters
3.4.5
Check theDCCC
algorithm
3.4.7
Y
Is the problem solved?
Is the problem solved?
Is the problem solved?
Is the fault rectified
Y
N
N
Y
END
Y
N
Y
N
Y
Y
Y
N
N
Y
N
Is UE traffic volumerestricted
3.4.8
Is the UE transmissionpower restricted
3.4.7
N
N
Check UE location and outlooppower control
3.4.7
Check the RLClayer
3.4.8
Check uplink loadresources
3.4.9
Check uplink CEresources
3.4.10
Check Iub resources
3.4.11
Collect data and report3.2.19
Is the problemcaused by UE
Is the problemcaused by server or
CN
Replace the UE or driver
Contact CN engineersto locate the fault, which
must be supported by
the RAN
Is the
problem caused by
laptop
Modify the laptop setting
or replace the laptop
Y
Y
Y
N
N
N
Check the TCP layer
and higher layers
3.4.8
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 38/43
P f Id tif i P f M t
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 39/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 39
Process of Identifying Performance Measurement-
Based HSUPA Data Transmission Problems
END N
Y
N
N
N
Y
NY
N
Y
Others
Is MultiACkabnormal?
The power controlof the control
channel or UE isabnormal.
Otherproblems Differentialcauses
Y
Does the averagesingle-user throughput reach
the cell requirement?
Is the BLER on theair interface high?
The item cannot beevaluated at present.
Evaluation item Identify HSUPA cellperformance problems
Is the uplink load limited?
Are CE resources limited?
Is the residualBLER high?
Is the UE in theUnhappy state?
Y
N
Is the transmitpower of the UE
limited?
Is the RLCretransmission ratio
high?
Is the Iubtransmission quality
poor?
Out-loop power controlproblems
N
N
Y
The uplink RTWP ishigh or the UE is faraway from the cell
center.
YOptimize the
transmission quality
Power controlparameters are
incorrect.
Is the RTWP limited?
Y Are data sources
insufficient?
Y
N
Perform expansion,use dynamic CEs, and
optimize the GBR
Y
Are Iubresourceslimited?
N
Expand the Iubtransmissionbandwidth
Y
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 40/43
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 41/43
Thank youwww.huawei.com
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 42/43
HUAWEI TECHNOLOGIES CO., LTD. Huawei Confidential Page 42
Probe:
Scheduled rate = Total size of all TBs received by the MAC-hs layer within a measurement period/Total duration for scheduling
TBs within a measurement period
The total size of all TBs received at the MAC-hs layer within a measurement period: involves the TBs that are correctly andincorrectly received.
Total duration for scheduling TBs within a measurement period: includes only the time when TBs are received. For example,
within a measurement period of 100 subframes (200 ms), if only 50 subframes contain data, the duration for receiving TBs with
data within a measurement period is 100 ms.
Served rate = Total size of all TBs received by the MAC-hs layer within a measurement period/Measurement period
Total size of all TBs received at the MAC-hs layer within a measurement period involves the TBs that are correctly and
incorrectly received.
Measurement period indicates all the time when TBs are received and not received.
The relationship between the served rate and the scheduled rate is as follows:
Served Rate = Scheduled Rate * HS - SCCH success rate
HS-SCCH success rate = Total duration for receiving TBs within a measurement period /Measurement period. The HS-SCCH
success rate indicates the scheduling probability.
MAC layer rate = Total size of TBs correctly received at the MAC-hs layer within a measurement period/Measurement period
Total size of TBs correctly received at the MAC-hs layer within a measurement period only involves the TBs that are correctly
received.
Measurement period indicates all the time when TBs are received and not received.
The relationship between the MAC layer rate and the served rate is as follows:
MAC layer rate = Served rate * (1 - SBLER)
SBLER = Total size of TBs incorrectly received within a measurement period/Total size of all TBs received within a
measurement period. The SBLER indicates the BLER of TBs.
To reflect user experience more approximately, the rate at the MAC layer is usually used.
Basic Concepts—calculated of Theoretical Rates
8/12/2019 UMTS HSPA DataTransmission Performance Improvement
http://slidepdf.com/reader/full/umts-hspa-datatransmission-performance-improvement 43/43
Basic Concepts——calculated of Theoretical Rates
Probe
MAC-e PDU non-DTX rate = Total size of all TBs during non-DTXs/(Number of non-DTXs * TTI)
Measurement period: The measurement period for a single log packet is 20 TTIs.
The MAC-e PDU non-DTX rate is the actual MAC-e rate excluding the TB transmission during DTXs but including TBretransmission;
Total size of all TBs during non-DTXs involves the TBs that are initially transmitted and retransmitted.
"Number of non-DTXs * TTI" indicates only the duration in which TBs are transmitted. For example, within a measurement period
of 100 subframes (200 ms), if only 50 subframes contain data, the "number of non-DTXs * TTI" is 100 ms.
MAC-e PDU served rate = Total size of all TBs during non-DTXs/(NUM_SAMPLES * TTI)
The MAC-e PDU served rate is the MAC-e service rate including the TB transmission during DTXs and TB retransmission.
Total size of all TBs during non-DTXs involves the TBs that are initially transmitted and retransmitted.
NUM_SAMPLES * TTI indicates duration in which TBs are transmitted and not transmitted. For example, within a measurement
period with 100 subframes (200 ms), if only 50 subframes contain data. However, the "NUM_SAMPLES * TTI" is still 200 ms.
The relationship between the MAC-e PDU served rate and the MAC-e PUD non-DTX rate is as follows:
Served rate = MAC-e PDU non-DTX rate * Non-DTX probability
Non-DTX probability = Number of non-DTXs/NUM_SAMPLES * 100%
MAC-e PDU available rate = Total size of all TBs during non-DTXs and when COMB_HIGH is ACK and
ACK_NS/(NUM_SAMPLES * TTI)
Total size of all TBs during non-DTXs and when COMB_HIGH is ACK and ACK_NS" involves only the TBs that are correctlytransmitted.
NUM_SAMPLES * TTI indicates all the time whether data is transmitted or not.
The relationship between the MAC-e PDU available rate and the MAC-e PUD served rate is as follows:
MAC-e PDU available rate ≈"MAC-e PDU served rate" * (1 - SBLER)
SBLER = (Number of non-DTXs – Number of ACKs or ACK_NSs)/Number of non-DTXs * 100%