Professional Documents
Culture Documents
LTE INTERNAL
Product Version
Total 69 pages
eRAN16.1
Change History
Date Issue CR/Defect ID Section Change Description Author
Distribution List
Copy No. Holder's Name & Role Issue Date
Keywords:
heavy traffic, Benchmark test, downlink throughput
Abstract:
This document describes the LTE downlink capacity solution and the activation method. It
aims to instruct users to improve the network performance and robustness on the live
network, therefore enabling Huawei to exceed the industry benchmark in terms of the
THP vs SINR curve in benchmark and traffic statistics performance competition for the
same city or telecom operator.
Contents
1 Introduction.......................................................................................................................1
1.1 Purpose...................................................................................................................................................... 1
1.2 Scope......................................................................................................................................................... 1
1.3 Organization............................................................................................................................................... 1
1.3.1 Relationship with Other Documents......................................................................................................... 2
1.3.2 Life Cycle................................................................................................................................................. 2
1.4 Open Issues............................................................................................................................................... 2
1.5 Assumptions............................................................................................................................................... 2
2 Solution Overview.............................................................................................................3
2.1 Scenario Analysis....................................................................................................................................... 3
2.1.1 Commercial Network Application Scenarios............................................................................................ 3
2.1.2 Benchmark Scenarios.............................................................................................................................. 5
2.1.3 Heavy Traffic Scenarios........................................................................................................................... 5
2.2 Benefits....................................................................................................................................................... 5
2.3 General Solution......................................................................................................................................... 6
2.3.2 Commercial Network Application Solution............................................................................................... 8
2.3.3 Benchmark Solution................................................................................................................................. 8
2.3.4 Heavy Traffic Solution.............................................................................................................................. 9
1 Introduction
1.1 Purpose
This document serves as an application guide of the LTE downlink capacity increase
solution. It describes downlink capacity increase feature portfolios and end-to-end (E2E)
deliverable applications of the solution, and provides guidelines for future solution-level
application and maintenance and for secondary development of service solutions for
existing LTE systems.
1.2 Scope
This document mainly provides guidelines for implementing the LTE downlink capacity
increase solution. Application guides of related features are optional in this document.
In addition to designs of this solution for the current version, this document includes all
designs for earlier versions.
It also serves as a reference for preparing user guides for other solutions, covering
capacity, radio networking, services, and quality of experience (QoE).
1.3 Organization
Chapter Description
Chapter Description
1.5 Assumptions
None
2 Solution Overview
Services with the highest penetration rate include instant message (IM), web browsing,
and social networking service (SNS), followed by streaming media and Email.
● 99% of packet intervals are smaller than 2 seconds and packet transmission has
the abruptness feature on the whole.
● For TCP services, the relationship between the downlink service rate and
downlink TCP status report transmission rate is reflected in the following formula:
40byte
RUL RDL
2 MTUbyte
The MTU value may be 1500 bytes or 576 bytes. The numerator is 60 bytes
instead of 40 bytes for IPv6. The formula shows that a smaller MTU value
indicates a larger value of the downlink TCP status report transmission rate.
Typical calculation examples are as follows:
Table 2.1.1.1.1.1.1.4.1 Ratio of uplink MTU to downlink MTU
(X, MTU) 40 Bytes, 40 Bytes, 576 60 Bytes, 1500 60 Bytes, 576
1500 Bytes Bytes Bytes Bytes
2.2 Benefits
Values and general goals of the LTE downlink throughput solution are as follows:
● Maintaining the competitiveness of each vendor during Benchmark tests.
− Intra-city test
− Intra-site test
− Inter-city test
● Maintaining the competitiveness during the swapping of competitors' networks
● Maintaining the competitiveness in the P3 test
● Improving service experience of users on live networks continuously.
− Scenarios when data cards are in the majority
− Scenarios when intelligent terminals are in the majority
− Big event scenarios
● Complying with protocols and maintain the leading roadmap competitiveness
2.3 General Solution
Factors affecting downlink throughput are as follows:
Adjustment mechanism
DL AMC
CSI obtaining
Downlink
PDCCH Reliability&Capacity
DL throughput Control
channel
PUCCH Reliability&Capacity
DL feedback
channel
PUSCH Reliability&Capacity
Frequency selective
DL UE
scheduling&Non-Frequency Priority selection for multi-UE scheduling
selection
selective scheduling
Networking
Channel quality
Power configuration
distribution on the
entire network
Interference control
RRU performance
TCP mechanism
Terminal maturity
The following functions are added to eRAN8.X (including eRAN8.0 and eRAN8.1):
1 Enhanced downlink IBLER adaptation: Downlink MCSs are selected based on
the CQIs reported by UEs and target IBLER. If the target value of IBLER is low,
a lower downlink MCS is likely to be selected. As a result, transmission
efficiency decreases. If the target value of IBLER is too high, a higher downlink
MCS is likely to be selected. As a result, downlink packet retransmission
increases and system resources are wasted. The best target IBLER value is
10% by default. In some scenarios, however, increasing the IBLER value can
improve downlink spectral efficiency of UEs, such as cell edge UEs or UEs at
the cell center or close to the eNodeB performing small-packet services. In
eRAN8.0, the following algorithm is added based on existing target IBLER value
adaptation: The degree of channel quality fluctuation is determined based on
CQIs reported by UEs. Then, the target IBLER value 5%, 10%, or 30% is
selected for data transmission based on the degree. When the degree of
channel quality fluctuation is greater than a certain threshold, the target IBLER
value increases. When it is less than a certain threshold, the target IBLER value
decreases. Otherwise, the existing target IBLER value adaptation process starts.
This function improves downlink spectral efficiency of UEs.
2. Optimized downlink frequency selective decision: Downlink frequency selective
scheduling uses channel quality differences on UE frequency bands to obtain
frequency selective scheduling gains of channels. To calculate scheduling
priorities and select resources, downlink frequency selective scheduling requires
accurate subband CQIs. When UEs are moving at a high speed, the subband
CQIs cannot reflect channel changes in time. Therefore, frequency selective
scheduling applies to scenarios where UEs are moving at a low speed. In
versions earlier than eRAN8.1, the period of downlink frequency selection
decision is 10s by default. In eRAN8.1, the period can be selected to support
fast downlink frequency selective decision, thereby increasing the proportion of
downlink frequency selective scheduling in all scheduling and downlink spectral
efficiency of UEs.
3. Enhanced PDCCH power control: In eRAN8.1, PDCCH resource selection is
enhanced and CCE-level PDCCH power control is introduced. For the same
SNR, the transmit power is increased based on the existing aggregation level for
data transmission. Compared with earlier versions, PDCCH resource selection
can be implemented in a finer way, thereby saving PDCCH resources, reducing
PDCCH congestion probabilities, reducing PDCCH overheads, and increasing
user-perceived downlink rate.
4. PUCCH outer loop power control: The power control for PUCCH Format2x is
implemented based on the SINR of PUCCH Format1a/1b. According to live
network data, interference on Format2x is stronger than that on Format1a/1b. As
a result, the SINR on Format2x is still low when the SINR on Format1a/1b is
converged, resulting in CQI and RANK false check and affecting downlink
transmission efficiency. Therefore, Format2x channel quality needs to be
considered during PUCCH power control. A simple method is adding Format2x-
based BER for outer loop power control based on Format1a/1b SINR power
control so that the target SINR value for Format1a is adjusted adaptively to
adapt to Format2x channel quality changes. This scheme improves downlink
spectral efficiency and user-perceived rate.
5. SRB RLC parameter adaptation: In heavy traffic scenarios, the MAC scheduling
interval becomes larger. If the SRB polling timer is set to a small value, the timer
may expire for multiple times when the peer end does not return a status report
(the scheduling interval is large). As a result, invalid retransmission occurs for
multiple times. If the scheduling interval is excessively large, the retransmission
times may reach 32 and cause call drops. In eRAN8.1, adaptive adjustment is
implemented for the SRB1 and SRB2 polling timers on the eNodeB side to
reduce invalid SRB retransmission. In this manner, usable downlink PRBs and
user-perceived downlink rate can be increased.
Enhanced PDCCH symbol adaptation coordinates well with downlink frequency selection.
When the common PDCCH symbol adaptation function is enabled, downlink frequency
selection is enabled only when the PDCCH symbol quantity is 3. After the function of
PDCCH symbol quantity adaptation is enhanced, downlink frequency selection is enabled
when the symbol quantity is adjusted to 1, thereby improving downlink spectral efficiency
and increasing user-perceived downlink rate.
Parameter Configurations
MOD CELLPDCCHALGO: LocalCellId=0, PdcchSymNumSwitch =ECfiAdaptionON
1. The PDCCH symbol quantity decreases. The average PDCCH symbol quantity is
calculated as follows:
(L.ChMeas.PDCCH.SymNum.1 x 1 + L.ChMeas.PDCCH.SymNum.2 x 2 +
L.ChMeas.PDCCH.SymNum.3 x
3)/sum(L.ChMeas.PDCCH.SymNum.1:L.ChMeas.PDCCH.SymNum.3)
2. You can observe traffic measurement counters to determine whether user-perceived
downlink rate increases: (L.Thrp.bits.DL –
L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI (ms).
3.1.2 Downlink Frequency Selection
Introduction
Frequency selective scheduling uses channel quality differences on UE frequency bands
to obtain frequency selective scheduling gains of channels. An eNodeB uses subband
CQIs reported by UEs to calculate the scheduling priority of each subband and perform
scheduling based on the subband scheduling priority of each UE on each subband so
that each UE is scheduled on the subband with the best channel quality. As shown in the
following figure, a is assigned to UE1, b is assigned to UE2, and c is assigned to UE1.
This is because UE1 and UE2 have the highest priorities in the scheduling queues
corresponding to the allocated RBGs.
When the uplink load is high, the eNodeB increases the period of reporting the subband
CQIs and adaptively adjusts the CQI reporting mode of UEs. The prolonged CQI
reporting period and the changed CQI reporting mode decrease the gain of frequency
selective scheduling.
When the downlink frequency selective scheduling mode is used, an eNodeB needs to
periodically determine the channel fluctuation and moving speed of each UE. When
downlink frequency selection conditions are met, downlink frequency selective scheduling
can be performed. When downlink frequency selection conditions are not met, UEs exit
downlink frequency selective scheduling. The CellDlschAlgo.FreqSelJudgePeriod
parameter is added to control the period of making downlink frequency selective
decisions. If the period is set to a smaller value, the UEs meeting frequency selection
conditions can be scheduled in the downlink sooner and downlink spectral efficiency
enhancement is greater. However, aperiodic CQI scheduling required for downlink
frequency selection increases, increasing uplink interference. When configuring fast
frequency selective decision, you are advised to turn on Doppler measurement reporting
period switches at the same time.
Less than 3 No
OFF - - No
When downlink frequency selection is enabled, you are advised to turn on the switch for
controlling simultaneous transmission of aperiodic CQIs and ACK/NACK among
compatibility switches. The purpose is to be compatible with the exceptions that UEs
report aperiodic CQIs in closed-loop MIMO mode.
Parameter Configurations
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=FreqSelSwitch-1;
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=FreqSelJudgeIgnorDopplerSwitch-1;
MOD CELLDLSCHALGO:LOCALCELLID=0,FREQSELJUDGEPERIOD=x;
MOD CELLULSCHALGO:LOCALCELLID=0,DOPALGOSWITCH=FastReportDopSwitch-1;
MOD CELLPDCCHALGO: LocalCellId=0, PdcchSymNumSwitch =EcfiAdaptionON;
Parameter Configurations
MOD CELLALGOSWITCH: LocalCellId=0, UlSchSwitch=PuschDtxSwitch-1;
1. Trace items of PUSCH DTX are available on the U2000 for you to observe the
percentage of PUSCH DTX, which is less than 5% in most cases.
Round down: RB resources can be fully used but the number of scheduling times will
increase.
Round up: All data can be scheduled at a time but a few RB resources will be wasted.
Adaptation: When resources for allocation exceed one RBG, use the roundup method.
When resources for allocation are less than one RBG, allocate resources based on the
actual RB quantity.
When 5 RBs are required by a UE, 3 RBs are used for scheduling in the current TTI and
remaining data is scheduled in subsequent TTIs if the round down method is used. If the
round up method is used, 6 RBs are used for scheduling but 1 RB is Padding. Six RBs
are used for scheduling when adaptation is used.
When 2 RBs are required by a UE, 2 RBs are used for scheduling in the current TTI if the
round down method is used. If the round up method is used, 3 RBs are used for
scheduling but 1 RB is Padding. Two RBs are used for scheduling when adaptation is
used.
RBG adaptation:
Parameter Configurations
MOD CELLPDCCHALGO: LocalCellId=0, AggLvlSelStrageForDualCW = STRATEGYBASEDONCOVERAGE;
Effect Evaluation Methods
The scenario where rank 2 is used in downlink scheduling and the PDCCH BLER is high
(the PDSCH is not a bottleneck) exists when the pilot of an intra-eNodeB neighboring cell
strongly interferes the PDCCH of the cell. In this scenario, the optimization point reduces
the PDCCH BLER.
PDCCH BLER:
(L.ChMeas.PDCCH.DL.DTXNum.AggLvl1 + L.ChMeas.PDCCH.DL.DTXNum.AggLvl2
+ L.ChMeas.PDCCH.DL.DTXNum.AggLvl4 +
L.ChMeas.PDCCH.DL.DTXNum.AggLvl8)/(L.ChMeas.PDCCH.AggLvl1Num +
L.ChMeas.PDCCH.AggLvl2Num + L.ChMeas.PDCCH.AggLvl4Num +
L.ChMeas.PDCCH.AggLvl8Num)
Parameters are introduced to enable the downlink MCS0 order to be fixed within a
specified period before and after the handover.
Aperiodic CQI fields are added to handover commands. In addition, the aperiodic CQI
reporting is triggered through downlink services after the handover to obtain downlink
channel quality information in advance, thereby improving downlink throughput in
handover areas.
Parameter Configurations
MOD CQIADAPTIVECFG: LocalCellId=0,HoAperiodicCqiCfgSwitch=ON;
MOD CELLDLSCHALGO: LocalCellId=0,HoStaticMcsTimer=10;
The following figures show the downlink SINR-THP before and after optimization,
respectively. The number of points indicating high SINR and low throughput decreases
greatly, that is, the number of points indicating low downlink throughput caused by low
MCS index in handover areas decreases greatly.
Figure 3.1.6.1.1.1.1.1 Downlink SINR-THP optimization scatter chart
To solve the preceding problem that application scenarios are limited, for capability-
limited UEs, the eNodeB now lowers the MCS to increase the number of RBs for the last
UE to be scheduled (the single UE scenario is expanded to the multiple UE scenario). For
small packet service scheduling, the eNodeB now lowers the MCS to increase the
number of RBs for the last service to be scheduled (the single service scenario is
expanded to the multiple service scenario).
After the preceding changes, the competitiveness of downlink throughput and delay is
improved. In addition, downlink performance in the commercial scenario can be
improved.
Parameter Configurations
MOD CELLDLSCHALGO:LOCALCELLID=0,RBPRIMCSSELECTSTRATEGY=MULTIUSER_STRATEGY,
RBPRIMCSSELECTRATIOTHD=10;
Effect Evaluation Methods
● The method of determining whether the feature has taken effect is as follows:
The proportion of MCS index 0 and RB usage that is calculated based on
L.ChMeas.PDSCH.MCS.0 and L.ChMeas.PRB.DL.Used.Avg can be used to
determine whether the feature has taken effect. If the proportion of MCS index 0
and RB usage increase sharply, the feature has taken effect.
● Feature gain evaluation:
Observe throughput improvement. The throughput is calculated by using the
following formula: Throughput = (L.Thrp.bits.DL –
L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
To solve the preceding problem, the maximum TBS index reduction for the last UE to be
scheduled has been changed to a configurable parameter. In addition, when interference
randomization takes effect, the eNodeB limits the range of RBs whose the index can be
reduced for the last to-be-scheduled UE to further control the interference to neighboring
cells.
Observe the downlink IBLER and the average downlink MCS index. If their values are
lower than those before the function is enabled, this function has taken effect.
Observe the throughput rate improvement. The throughput rate is calculated by using the
following formula: Throughput rate = (L.Thrp.bits.DL –
L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
3.1.9 Rank Filter
Introduction
Currently, when open-loop and closed-loop multiplexing are used for Huawei downlink
MIMO mode, the eNodeB depends on the rank reported by a UE for scheduling, which is
not reliable due to the following reasons:
● Some reported ranks that are not reliable can be filtered out. Currently, most
false rank reporting identified during simulation and tests is the early rank 2
handover and single-codeword scheduling can reduce interference between
networks compared with dual-codeword scheduling. Therefore, the eNodeB
uses rank 1 instead of rank 2 for scheduling.
● There is a probability of missing check or false check on baseband RI reliability.
The probability of false check on baseband RI reliability is smaller than 1%,
however, the loss is caused once false check occurs. For example, the
throughput decreases by 50% if rank 2 is encoded to rank 1. Figure
3.1.9.1.1.1.1.2 shows that there is a sharp decrease in the peak throughput.
Parameter Configurations
MOD CELLDLSCHALGO: LocalCellId=0, DlRankOptimizeSwitch=ON;
Parameter Configurations
MOD CELLALGOSWITCH: LocalCellId=0, DlSchSwitch= EpfEnhancedSwitch -1;
Effect Evaluation Methods
Observe throughput improvement. The throughput is calculated by using the following
formula: Throughput = (L.Thrp.bits.DL –
L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
In the typical MBB service scenario, the expected gain ranges from 5% to 10%.
For multi-UE non-full-buffer service scenarios, reported CQIs do not match channels well
during scheduling due to network interference fluctuation. In this scenario, AMC can be
optimized noticeably. Enhancement of downlink target IBLER value adaptation uses a
relative spectral efficiency difference to implement target IBLER value adaptation
decision. When the difference is greater than a threshold, a larger target IBLER value is
used. Otherwise, a smaller target IBLER value is used.
Parameter Configurations
MOD CELLALGOSWITCH:LOCALCELLID=0,CQIADJALGOSWITCH=DlEnVarIblerTargetSwitch-1;
Introduction
When the bandwidth is 10 MHz/15 MHz/20 MHz and downlink channel quality is high,
PDCCH CCE can be set to 1 to reduce PDCCH overheads and increase downlink
capacity.
Parameter Configurations
MOD CELLPDCCHALGO:LOCALCELLID=0,PdcchMaxCodeRate=14;
Parameter Configurations
MOD CELLPDCCHALGO:LOCALCELLID=0,PdcchMaxCodeRate=95;
Parameter Configurations
MOD CELLPDCCHALGO:LOCALCELLID=0,PdcchPowerEnhancedSwitch=ON;
Parameters: to be supplemented (the parameter workflow for this feature has not been
reviewed).
The DL power is not fully used, for example, the rated RRU power is 60 W, but only 40 W
is actually used.
Three types of RRU modules are as follows: RRU3268 (2.6 GHz), RRU3642 (800 MHz),
and RU3635 (800 MHz). The last two types are used for China Telecom.
If the baseband processing unit (BBP) is UBBPd3/4/5/6, the TX/RX mode is not set to
4T4R, and the cells are configured to work in 5/10/15/20 MHz, the AWPC algorithm can
be activated to further improve throughput of CEUs.
Off:
You can observe the value of the following formula to determine whether the DL
perceivable throughput of the CEU increases:
L.Thrp.bits.DL.CEU/L.Thrp.Time.DL.CEU
You can observe the value of the following formula to determine whether the DL
perceivable throughput of the UE increases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Parameter Configurations
On:
Off:
PUCCH FMT2 code channel allocation optimization: Widen the interval of the first half
code channel of each RB for reporting CQI. For example, six CQI code channels can be
allocated on an RB. The first three code channels are numbered 0, 4 and 8, and the last
three code channels are numbered 2, 6 and 10. If not optimized, the code channels
allocation sequence is 0, 2, 4, 6, 8, and 10.
Parameter Configurations
On:
Off:
You can observe the value of the following formula to determine whether the DL
perceivable throughput of the UE increases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Parameter Configurations
On:
Off:
MOD CELLULSCHALGO:LOCALCELLID=0,AperiodicCsiUlTxMode=CsiOnly;
(L.Thrp.bits.UL – L.Thrp.bits.UE.UL.SmallPkt)/L.Thrp.Time.UE.UL.RmvSmallPkt
You can observe the value of the following formula to determine whether the DL
throughput of the UE decreases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
3.1.18 256QAM
Introduction
3GPP Release 12 introduces the 256 quadrature amplitude modulation (256QAM)
scheme. 256QAM is a supplement of the downlink quadrature phase shift keying (QPSK),
16QAM and 64QAM. It is used to increase bit rate of the UE when radio conditions are
good. Downlink 256QAM improves CCUs' downlink spectral efficiency and DL
throughput. Gains are affected by the quality of the radio channel, the EVM of the RF TX
channel of the eNodeB and that of the UEs. Each symbol in 256QAM carries information
about eight bits. Compared with 64QAM, 256QAM supports larger TBS, and the
theoretical peak spectral efficiency increases by 33%.
Parameter Configurations
On:
You can observe the counter L.Thrp.bits.DL to determine whether the DL throughput of
the UE increases.
You can observe the value of the following formula to determine whether the DL
throughput of the UE decreases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Parameter configurations
Run the following command to enable this function:
The following formula can be used to determine whether the user-perceived uplink
throughput increases:
(L.Thrp.bits.UL – L.Thrp.bits.UE.UL.SmallPkt)/L.Thrp.Time.UE.UL.RmvSmallPkt
You can check whether the optimization takes effect by viewing the sum of the counters
L.ChMeas.CQI.DL.SingleCW.Aperiodic, L.ChMeas.CQI.DL.DualCW.Code0.Periodic,
and L.ChMeas.CQI.DL.DualCW.Code1.Aperiodic. If the sum does not change
obviously, the value of the NOSCHSTOPACQITHD parameter is large.
3.1.20 Intra-eNodeB CSPC
Introduction
This algorithm comprehensively considers the resource allocation status of cells
configured on a BBP. By using the CQIs reported by intra-BBP CSPC UEs receiving
intra-BBP interference, this algorithm calibrates the reported CQIs based on the
interference change at the measurement and scheduling moments, therefore improving
the spectral efficiency of such UEs.
The eNodeB estimates the resource allocation status of cells configured on a BBP and
obtains the number of resource block groups (RBGs) of each cell.
Based on the RBG overlapping between intra-BBP interference neighboring cells and the
serving cell, the eNodeB calibrates the modulation and coding scheme (MCS) for joint
AMC (JAMC) UEs before performing downlink scheduling. This helps the eNodeB obtain
a more accurate MCS for downlink scheduling.
This optimization applies to the scenario where the downlink PRB usage is greater than
10%. The optimization gains increase as the traffic load increases. When the average
PRB usage reaches 50%, the gains reach the peak value. When the PRB usage
continues to increase, the gains decrease gradually. The optimization does not present
gains when the PRB usage reaches the maximum.
Parameter Configurations
Enable this function:
The following formula is used to determine whether the downlink CEU throughput
increases:
L.Thrp.bits.DL.CEU/L.Thrp.Time.DL.CEU
The following formula is used to determine whether the downlink user throughput
increases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
3.1.21 CQI Initial Filtering Optimization
Introduction
After the initial access or incoming handover, the eNodeB performs initial filtering
optimization on full-band CQIs reported by UEs at the medium or long distance from the
cell center, which reduces the time of increasing the CQI filtering value from the initial
value to the target value. This method increases the MCS, improves the spectral
efficiency, and increases the downlink user throughput.
Parameter Configurations
Run the following command to enable this function:
If there are only a small number of UEs in a cell (fewer than 100 UEs in a 10 MHz cell)
when the CQI initial filtering optimization is enabled, the user throughput increases by up
to 5%.
The following formula is used to determine whether the downlink user throughput
increases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
To disable MCS lowering in RBG initial transmission, run the following command:
Performance Prediction
● If the RBG allocation policy is ROUND UP or ADAPTIVE in the test scenario and
the proportion of scheduled small packets (tail packets) is high (for example,
greater than 20%), this function has the following impacts on KPIs:
− The IBLER decreases.
− The user-perceived downlink throughput increases by 1% to 5%.
− The average downlink cell throughput increases by 1% to 5%.
− The average downlink MCS indexes of UEs decrease, depending on the
proportion of UEs on which this function takes effect.
− Other KPIs related to the control plane are basically not affected.
● If the RBG allocation policy is ROUND UP or ADAPTIVE in the test scenario and
the proportion of scheduled small packets (tail packets) is low (for example,
lower than 20%), this function does not improve KPIs significantly and the KPIs
related to the user plane and control plane are basically not affected.
● If the RBG allocation policy is ROUND DOWN in the test scenario, this function
does not take effect and KPIs are not affected.
● In the benchmark scenario, the full buffer service will occupy all available RBs
and it is unnecessary to lower the MCS index. In this scenario, this function does
not improve KPIs significantly.
Performance Prediction
● Since the DRX offset of UEs is set in a centralized manner (including changes
between 2T and 4T), CQI resources are allocated in more centralized time
domains. Currently, this change does not affect network KPIs.
● The decrease of PUSCH RBs (two RBs decreased in maximum) will lead to the
decrease of uplink throughput. The proportion of the maximum number of RBs
that can be decreased at or near the cell center is calculated according to the
formula: 2 RBs/Number of RBs at the bandwidth.
● According to the test result in a lab, when the ElasticCarrierSwitch parameter
is set to ON in the networking scenario with three 4T cells, turning on
EmimoElasticPortSwitch can increase downlink user throughput by 5%.
● The measurement result of the serving cell is inaccurate. As demonstrated in the
test result, the RSRP of the serving cell fluctuates significantly, affecting the
accuracy of reported measurement reports (MRs). In addition, the handover
success rate may decrease (by less than 2%).
On the live network, breathing pilot offers the following benefits:
● If the DL PRB usage of the network is less than 1% and the average number of
RRC_CONNECTED UEs in an entire day is less than 20, user-perceived
downlink throughput increases by 15%.
● If the DL PRB usage of the network ranges from 1% to 5% and the average
number of RRC_CONNECTED UEs in an entire day is less than 20, user-
perceived downlink throughput increases by 5% to 10%.
● If the DL PRB usage of the network is greater than 5% and the average number
of RRC_CONNECTED UEs in an entire day is less than 20, this function does
not offer benefits.
In the drive test (DT) scenario, breathing pilot offers the following benefits:
● If the DL PRB usage of the DT area or the neighboring cell is less than 1% and
the average number of RRC_CONNECTED UEs in an entire day is less than 20,
the throughput of cell center users (CCUs) does not increase, the throughput of
users near the cell center increases by about 30%, and the throughput of cell
edge users (CEUs) increases by about 60%.
● If the DL PRB usage of the DT area or the neighboring cell ranges from 1% to
5% and the average number of RRC_CONNECTED UEs in an entire day is less
than 20, the throughput of CCUs does not increase, the throughput of users near
the cell center increases by about 5%, and the throughput of CEUs increases by
about 10%.
● If the DL PRB usage of the DT area or the neighboring cell is greater than 5%
and the average number of RRC_CONNECTED UEs in an entire day is less
than 20, breathing pilot does not offer benefits to CCUs, users near the cell
center, or CEUs.
Effect Evaluation Methods
You can observe the counters in the following formula to determine whether the user-
perceived downlink throughput increases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Some PDCCH power is not used owing to a small number of scheduled users in each TTI
or CCE fragmentation. After RB scheduling is complete in a TTI, the eNodeB redistributes
the remaining power to desired users so as to improve their PDCCH performance.
● Increasing CEUs' PDCCH power and throughput, and decreasing CEUs' DTX
proportion and service drop rate
● Increasing the PDCCH power and the number of downlink scheduling times for
users performing large-packet services, thereby improving spectral efficiency
and user throughput in a cell
● Increasing the number of schedulable RBs or scheduling times for DT users
Parameter Configurations
To enable remaining PDCCH power redistribution, run the following commands:
Performance Prediction
If the PdcchSparePowerAllocStrage parameter is set to WEAKCOVERAGEUSER, the
PDCCH performance of CEUs can be enhanced and the DTX proportion of CEUs can be
decreased. The benefits of this function become more and more obvious as the number
of users with low MCS indexes grows. Low MCS indexes indicate indexes 0 to 5. If the
proportion of users with low MCS indexes is less than 10%, it is predicted that user-
perceived throughput increases by up to 1%. If the proportion of users with low MCS
indexes is from 10% to 20%, it is predicted that user-perceived throughput increases by
up to 2%. If the proportion of users with low MCS indexes is greater than 20%, it is
predicted that user-perceived throughput increases by 1% to 2%.
Effect Evaluation Methods
You can observe the counters in the following formula to determine whether user-
perceived downlink throughput increases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Check whether the proportion of DTX users to which PDCCH resources are allocated
with aggregation level 8 decreases according to the following formula:
L.ChMeas.PDCCH.DL.DTXNum.AggLvl8/L.ChMeas.PDCCH.AggLvl8Num
Parameter Configurations
To enable small bandwidth RLC optimization, run the following command:
MOD CELLALGOSWITCH:LOCALCELLID=X,SmallBandOptSwitch=SmallBandRlcOptSwitch-1.
MOD CELLALGOSWITCH:LOCALCELLID=X,SmallBandOptSwitch=SmallBandRlcOptSwitch-0.
Performance Prediction
In the heavy-load scenarios when the bandwidth is 5 MHz, 3 MHz, or 1.4 MHz:
Parameter Configurations
To enable adaptive compensation for the initial value of CQI adjustment:
Performance Prediction
In MBB scenarios where the downlink IBLER is far lower than the target value and the
downlink PRB usage is greater than 20%, the user-perceived downlink rate increases by
1% to 5%. When the downlink PRB usage is lower than or equal to 20%, the user-
perceived downlink rate increases by 0% to 2%. The downlink IBLER and residual block
error rate (RBLER) are expected to increase.
You can observe the counters in the following formula to determine whether downlink
IBLER increases:
DL IBLER = (L.Traffic.DL.SCH.QPSK.ErrTB.Ibler +
L.Traffic.DL.SCH.16QAM.ErrTB.Ibler + L.Traffic.DL.SCH.64QAM.ErrTB.Ibler)/
(L.Traffic.DL.SCH.QPSK.TB + L.Traffic.DL.SCH.16QAM.TB +
L.Traffic.DL.SCH.64QAM.TB)
Parameter Configurations
To enable enhancement of adaptive downlink HARQ retransmission:
Performance Prediction
The user-perceived downlink rate increases by 1% to 5% in heavy-traffic scenarios where
the downlink PRB usage is greater than 50% and the downlink IBLER is between 9% and
30%, the proportion of UEs at the cell center is lower than or equal to 25%. The downlink
RBLER is expected to possibly increase.
You can observe the counters in the following formula to determine whether the number
of downlink RB retransmissions reduces:
Parameter Configurations
To enable flexible target IBLER adaptation and set the TBS index threshold for both low
target IBLER and high target IBLER:
MOD CELLALGOSWITCH:LOCALCELLID=0,CQIADJALGOSWITCH=DlEnVarIblerTargetSwitch-1;
MOD CELLDLSCHALGO:LOCALCELLID=0,LowIblerTargetTbsIdxThld=21,HighIblerTargetTbsIdxThld=5;
MOD CELLDLSCHALGO:LOCALCELLID=0,LowIblerTargetTbsIdxThld=255;
MOD CELLALGOSWITCH:LOCALCELLID=0,CQIADJALGOSWITCH=DlEnVarIblerTargetSwitch-0;
Performance Prediction
In heavy-traffic scenarios, the PRB usage is from 10% to 50%, and the CPU usage is less
than 55%. The downlink user throughput is expected to increase by 0% to 3%, and the
downlink IBLER is expected to increase.
In DTs, the downlink user throughput at the cell center is expected to increase by 1% to
3%, and the downlink IBLER decreases. The downlink user throughput at the cell edge is
expected to increase by 0% to 5%, and the downlink IBLER increases.
You can observe the counters in the following formula to determine whether downlink
IBLER increases:
DL IBLER = (L.Traffic.DL.SCH.QPSK.ErrTB.Ibler +
L.Traffic.DL.SCH.16QAM.ErrTB.Ibler + L.Traffic.DL.SCH.64QAM.ErrTB.Ibler)/
(L.Traffic.DL.SCH.QPSK.TB + L.Traffic.DL.SCH.16QAM.TB +
L.Traffic.DL.SCH.64QAM.TB)
Parameter Configurations
To set a recommended PDCCH step:
Performance Prediction
● Scenarios of positive gains: DTX occurs in PDCCH.
● Scenarios of negative gains: the eNodeB cannot identify the DTX caused by
feedback errors in the case of severe uplink interference. If the uplink check
error rate is higher than 1%, this function generates negative gains. If the uplink
interference is greater than -105 dB after preliminary uplink interference
observation, you are not advised to change the default parameter values.
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Parameter Configurations
This function is enabled inside RRUs and invisible to customers. There are no MML
commands or parameters.
Performance Prediction
In the scenarios of low-power RRUs (with a power headroom other than 1 dB, 3 dB or 6
dB), the SINR at the cell center increases by 0.5 dB to 1 dB.
When RRUs have a power headroom of 2.5 dB, the downlink SINR of a UE at the cell
center increases by 1 dB. When full-buffer services are performed at the cell center, the
throughput increases by over 1% in 64QAM modulation mode and by 3% to 5% in
256QAM modulation mode.
When the power headroom for RRUs is set to 1 dB, 3 dB, or 6 dB (the same as the
original solution), this function generates no gains to the SINR at the cell center and the
EVM.
Parameter Configurations
To enable multi-band optimal carrier selection:
Run the MOD CELLALGOSWITCH command, with the MbfcsSwitch option selected in
the Load balancing algorithm switch parameter.
To disable multi-band optimal carrier selection:
Run the MOD CELLALGOSWITCH command, with the MbfcsSwitch option deselected
in the Load balancing algorithm switch parameter.
Expected Gains
This function generates gains in the following scenarios:
Performance Prediction
On multi-band networks with three to five carriers and an average PRB usage of 30% to
70%, the average user-perceived downlink rate increases by 5% to 20%.
On light-traffic networks (with the PRB usage smaller than 30%), this function generates
no gains.
The increased number of UE handovers may decrease the network handover success
rate and increase the handover signaling overheads. The increased number of handovers
for a UE may increase the service drop rate.
Inter- and intra-frequency measurements increase the uplink overheads and may slightly
decrease the uplink throughput.
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Parameter Configurations
To enable variable step of CQI adjustment optimization for CRS transmission disabling:
MOD CELLALGOSWITCH:LOCALCELLID=0,CQIADJALGOSWITCH=PilotOffCqiAdjOptSwitch-1;
To disable variable step of CQI adjustment optimization for CRS transmission disabling:
CELLALGOSWITCH:LOCALCELLID=0,CQIADJALGOSWITCH=PilotOffCqiAdjOptSwitch-0;
Performance Prediction
In DTs, the downlink user throughput at the middle and edge of a cell is expected to
increase by 0% to 5%.
● The eNodeB allocates as much data as possible to the SCC for pre-scheduling.
− If the SCC is idle, scheduling capabilities of the SCC are fully utilized.
− If the SCC is congested and UEs cannot obtain resources, pre-scheduling fails. In
this situation, the PCC schedules buffered data without proportion restrictions,
fully utilizing scheduling capabilities of the PCC.
In pre-scheduling, the data allocation proportion between the PCC and SCC can
be adjusted for level-l CA scheduling.
● The last packet can be scheduled on the SCC. If pre-scheduling on the SCC
fails, the PCC schedules the last packet.
Start the level-1 CA scheduling
process.
Yes
Is the data volume
buffered at the RLC layer
less than the threshold?
No
End
Parameter Configurations
To enable this function:
Expected Gains
This function brings obvious gains when all of the following conditions are met:
● PUCCHAlgoSwitch.Dl2CCAckResShareSw is set to ON
● Both the PRB usage and the penetration rate of CA UEs are greater than 10%.
● The traffic proportion of activated CA UEs is greater than 10% and less than
40%.
Traffic proportion of activated CA users = L.Thrp.bits.DL.CAUser/L.Thrp.bits.DL
Performance Prediction
When PUCCHAlgoSwitch.Dl2CCAckResShareSw is set to ON and both the PRB
usage and CA user proportion are greater than 10%, the throughput for CA users and
that for all users increases by 5%- 15% and 1%-5%, respectively.
Gains of precise scheduling for CA UEs are related to the penetration rate of CA UEs and
traffic proportion of CA users, as illustrated in the following figures.
5%
3%
1%
10%
5%
5%
3%
1%
15%
10%
5%
The following figures show simulation results of precise scheduling for CA users.
3.1.34 NACK/DTX Determination Optimization During PUCCH
Demodulation in DL 2CC Scenarios
Introduction
In DL 2CC scenarios using PUCCH format 3, PUCCH demodulation enables the eNodeB
to accurately identify NACK fed back by the SCC when NACK and DTX messages fed
back at Layer 1 are confused, obtaining HARQ combination gains.
Parameter Configurations
Run the following commands to enable this function:
When this function is enabled, HARQ results are reported to Layer 2 in PUCCH format 3.
The SCC detects no HARQ from the UE, and therefore NACK is filled. The value of
ulAck1 is displayed as 0.
When this function is disabled, the value of ulAck1 is displayed as 15.
Performance Prediction
The downlink throughput increases by 2% according to DT results.
3.1.35 CSI Reporting Optimization in CA and Closed-Loop
MIMO Scenarios
Introduction
For activated CA UEs in closed-loop MIMO mode, enhanced aperiodic CQI reporting
enables CA UEs to periodically feed back PMIs, improving downlink benchmark
performance of CA UEs.
Parameter Configurations
Run the following commands to enable this function:
Expected Gains
This function brings obvious gains in benchmark DT scenarios with CA and closed-loop
MIMO enabled. In addition, the test UE must be a CA UE.
KPI Prediction
The downlink throughput increases by more than 5% according to DT results.
● The eNodeB monitors UEs in the active state of DRX mode. If it detects that a
UE has no data to receive or send, it sends a DRX Command MAC Control
Element message to the UE, instructing the UE to enter the sleep state. In this
way, the eNodeB increases the CRS transmission disabling rate. The function of
increasing the CRS transmission disabling rate is controlled by the
CellDlschAlgo.PilotOffRateIncreaseSw parameter. If this parameter is set to
NOT_BASED_ON_MEASUREMENT, this function takes effect in the local cell.
● If this parameter is set to BASED_ON_MEASUREMENT, this function takes
effect in the neighboring cells if they are notified that there are large-packet
users in the local cell. eNodeBs exchange information through eX2 interfaces.
Parameter Configurations
To enable this function:
Gain-Obtaining Scenarios
This function brings obvious gains in the following scenarios:
● There is significant intra-frequency interference in contiguous coverage areas,
for example, in areas where the typical inter-site distance is 500 m.
(recommended scenario for this function)
● The average number of UEs is less than 20 and the average downlink PRB
usage is less than 10% during off-peak hours in a cell.
● For LTE FDD, the cell bandwidth is 10 MHz, 15 MHz, or 20 MHz. For LTE FDD,
the cell bandwidth is 5 MHz, 10 MHz, 15 MHz, or 20 MHz.
Performance Prediction
When this function is enabled with the breathing pilot function, the average downlink
throughput will further increase by 1%-5%.
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Performance Prediction
The user-perceived throughput is expected to increase by 1% to 5% when all of the
following conditions are met:
● The uplink and downlink CCE usage is higher than 30%.
● The RB usage is 30% to 90%.
● The average number of PDCCH symbols is greater than 2.
● The uplink and downlink CCE failure rate is greater than 5%.
In other scenarios, this function has no gains or even negative gains.
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
● When the number of available CCEs is greater than a threshold, CCEs are
considered sufficient, and the CFI does not increase.
● When the number of available CCEs is less than the threshold and CCE
allocation fails, the CFI increases. If this function takes effect, the average CFI
decreases, the number of available symbols for PDSCHs increases, and user-
perceived downlink throughput increases in heavy-load scenarios. Parameter
Configurations
Run the following command to set the CCE threshold for CFI adaption to the
recommended value (20 MHz bandwidth is used as an example):
Run the following command to set the threshold to the default value:
MOD CELLPDCCHALGO: LocalCellId=0, CfiAdjCceThld =INF;
Performance Prediction
When CfiAdjCceThld is set to a value other than INF, the average CFI is smaller than
that when CfiAdjCceThld is set to INF, the downlink throughput increases (by 0 to 5%
with the recommended setting), but the number of CCE allocation failures increases and
the uplink throughput slightly decreases (by 2% with the recommended setting).
In moderate- and heavy-load scenarios (downlink RB usage > 50%; average CFI > 2), the
throughput gains are obvious. The smaller the value of CfiAdjCceThld, the smaller the
average CFI, and the higher the downlink throughput.
In moderate- and low-load scenarios, the throughput gains are unstable. Changing the
value of CfiAdjCceThld makes little difference.
(L.ChMeas.PDCCH.SymNum.1 + 2 x L.ChMeas.PDCCH.SymNum.2 + 3 x
L.ChMeas.PDCCH.SymNum.3 + 4 x
L.ChMeas.PDCCH.SymNum.4)/(L.ChMeas.PDCCH.SymNum.1 + ... +
L.ChMeas.PDCCH.SymNum.4)
You can observe counters in the following formula to determine whether the average
downlink throughput increases:
3.1.39.1 Introduction
Smart frequency selective scheduling adds multi-user admission and large-packet
identification to the original frequency selective scheduling, and dynamically adjusts the
triggering period and validity period of aperiodic CQI reporting. It applies to heavy-load
hotspot scenarios where multiple low-speed UEs are running services and improves
downlink user experience.
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=FreqSelSwitch-1;
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHEXTSWITCH=SMART_FSS_SWITCH-1;
Deactivation
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHEXTSWITCH=SMART_FSS_SWITCH-0;
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=FreqSelSwitch-0;
The buffered traffic volumes of multiple UEs are constantly large, for example, in a multi-
UE packet injection test, drive test, or downlink peak demonstration test.
Parameter Configurations
Run the following command to enable this function:
MOD CELLALGOSWITCH:LOCALCELLID=0,CQIADJALGOSWITCH=CqiFltSpectralEffOptSwitch-1;
MOD CELLALGOSWITCH:LOCALCELLID=0,CQIADJALGOSWITCH=CqiFltSpectralEffOptSwitch-0;
Performance Prediction
The expected gains increase by 1% to 3% when UEs move at a low speed and the
following conditions are met:
3.1.41.1 Introduction
NPREO is an acronym of nomPDSCH-RS-EPRE-Offset, which is a field defined in the
CQI-ReportConfig information element (IE) since 3GPP Release 8. The NPREO constant
is used to compensate for PDSCH SINR estimation for non-TM9/TM10 UEs. From
eRAN11.1 onwards, the NPREO constant can be configured using the
CELLDLPCPDSCHPA.NomPdschRsEpreOffset parameter.
NPREO constant optimization includes cell-level NPREO constant adjustment and UE-
level NPREO constant adjustment. After the cell-level NPREO constant adjustment
function is enabled, the eNodeB automatically searches for the optimal cell-level NPREO
configuration. Try different cell-level NPREO constant values and collect corresponding
data such as the cell traffic, user throughput, and downlink PRB usage. Use the collected
data to fit the polynomial curve of user throughput and cell traffic with different NPREO
constant values, and then estimate respective UE throughput. If the estimated UE
throughput with NPREO set to A is greater than that with NPREO set to B and the gain is
greater than CellDlPcPdschPA.CellNpreoAdjThreshold, A is considered to be better
than B. Once the optimal cell-level NPREO value is determined, the value replaces the
original GUI value of CellDlPcPdschPA.NomPdschRsEpreOffset.
NPREO constant optimization enables the eNodeB to select an optimal NPREO constant
for a UE so that the CQI and rank reported by the UE better match the actual channel
fluctuation. In this way, the eNodeB can select a more accurate rank and MCS for
scheduling, improving the downlink spectral efficiency.
Deactivation
After the downlink NPREO constant optimization function is enabled, different NPREO
constant values are attempted during cell-level NPREO constant adjustment. User
experience is expected to be affected for about one week, and the downlink UE
throughput may decrease by a maximum of 5%. NPREO constant optimization directly
changes the CQI and rank reporting of UEs. Therefore, this function changes items such
as the rank proportion, average CQI, average MCS index, CQI distribution, and MCS
distribution. In addition, the IBLER and RBLER may change slightly. During the
adjustment, the optimal NPREO is reconfigured for UEs. Therefore, after this function is
enabled, the number of RRC connection reconfiguration signaling increases.
3.1.42.1 Introduction
Downlink AMC enables the eNodeB to select an appropriate downlink MCS based on the
channel quality. The selected MCS determines the downlink spectral efficiency. The
traditional AMC algorithm sets a certain target IBLER value and adjusts the reported CQI
based on the ACK/NACK feedback. In this way, the final IBLER converges to the target
value. Parameters such as the target IBLER and adjustment step are fixed and cannot be
adaptively adjusted to the optimal values as the environment changes. Therefore, data
learning-based downlink smart AMC is introduced to adaptively select the downlink MCS
with the optimal downlink spectral efficiency by learning historical downlink scheduling
data.
With data learning-based downlink smart AMC a large amount of cell user data is
collected, including the downlink MCS, ACK/NACK feedback, air interface channel
quality, and channel fluctuation information. The collected data is trained using a neural
network and the resultant trained model can guide the selection of the optimal downlink
MCS in real time, improving the transmission spectral efficiency of UEs. This increases
the downlink user-perceived data rate.
Deactivation:
MOD CellIntelAmcConfig: LocalCellId=0, IntelligentAmcSwitch =
DL_INTEL_AMC_SELECTION_SW-0;
3.1.43.1 Introduction
Two-dimensional frequency selective scheduling is an optimization based on smart
frequency selective scheduling. During downlink RB resource allocation, the eNodeB
reorders RBGs based on the scheduling priorities of subbands and schedules RBGs in
descending order of scheduling priorities. In addition, the eNodeB supports
retransmission avoidance for frequency selective scheduling, fully utilizing frequency
selective gains of channels.
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHEXTSWITCH=SMART_FSS_SWITCH-1;
MOD CELLALGOEXTSWITCH:LOCALCELLID=0,DLSCHENHSWITCH=2D_FSS_SWITCH-1;
Deactivation:
MOD CELLALGOEXTSWITCH:LOCALCELLID=0,DLSCHENHSWITCH=2D_FSS_SWITCH-0;
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHEXTSWITCH=SMART_FSS_SWITCH-0;
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=FreqSelSwitch-0;
3.1.43.3 Performance Prediction
In scenarios where the downlink PRB usage is greater than 30%, the average number of
downlink activated UEs is greater than 2, and the cell is a low-speed cell, two-
dimensional frequency selective scheduling is expected to increase User Downlink
Average Throughput by 0–5%.
3.1.44.1 Introduction
In 4T4P scenarios, some Type B symbols do not transmit pilot signals. Therefore, the
power of these symbols is not fully utilized. The power of these symbols can be increased
to maximize power utilization and improve network performance in 4T4P scenarios.
After mutual power aid between VAM ports is used to equalize the power of Type B1 and
Type B2 symbols on ports 0/1 and ports 2/3, the PB parameter is set to the maximum
value to increase the remaining power of all UEs.
Value scenario: If the baseline PA/PB is set to 0/0, the pilot power can be increased by 3
dB and the data power remains unchanged after mutual power aid is performed between
VAM ports and PA/PB is set to –3/0. If the baseline PA/PB value is set to –3/1, data
power can be increased after mutual power aid is performed between VAM ports and the
PA/PB is set to –3/0.
//In scenarios where the remaining power is used to increase the data power with the
baseline PA/PB being –3/1
MOD CELLDLPCPDSCH: LOCALCELLID=0, RemainingPwrUseStrategy=PORT_PWR_BALANCING;
//In scenarios where the remaining power is used to improve coverage, wth the baseline
PA/PB being 0/0, and the value of ReferenceSignalPwr being 152.
MOD CELLDLPCPDSCH: LOCALCELLID=0, RemainingPwrUseStrategy=PORT_PWR_BALANCING;
Deactivation:
//In scenarios where the remaining power is used to increase the data power with the
baseline PA/PB being –3/1
MOD PDSCHCFG:LOCALCELLID=0, Pb=1;
//In scenarios where the remaining power is used to improve coverage, wth the baseline
PA/PB being 0/0, and the value of ReferenceSignalPwr being 152.
MOD PDSCHCFG :LOCALCELLID=0, ReferenceSignalPwr=152;
In scenarios where the remaining power is used to improve coverage, the remaining
power of 4T cells can improve downlink coverage by 3 dB and more cell edge users can
access the network. This may increase the cell traffic volume, decrease the average
downlink UE throughput, causing a slight fluctuation in the RRC connection setup
success rate, RRC connection reestablishment success rate, E-RAB setup success rate,
RB usage, and service drop rate.
The initial value of closed-loop adjustment is set to –10. If 20 consecutive packets do not
receive DTX, the value is increased by 5 until it is adjusted to 0. If 20 consecutive packets
do not receive DTX, the initial value is no longer adjusted subsequently. The purpose of
this design is to ensure the access success rate in some extreme scenarios, especially
when only uplink services are available. The minimum aggregation level needs to start
from 2, resisting interference in extreme scenarios. However, system capacity will be
affected if the CCE aggregation level is excessively low and the number of scheduling
times is small.
Parameter Configurations
MOD CELLPDCCHALGO: LocalCellId=0, PdcchCapacityImproveSwitch =ON;
Expected gains: Modifying the initial value of closed-loop PDCCH adjustment applies to
the scenario where the DRB has taken effect, lowering the PDCCH DRB aggregation
level and increasing the number of scheduled UEs.
Greater gains can be obtained when PDCCH CCE usage is 60% to 80%.
Therefore, the step adjustment scheme is introduced. The step is adjusted according to
the target IBLER value offset in the low scheduling probability scenario. When the offset
is large, the large step size is used. When the offset is small, the small step size is used.
This scheme can improve IBLER convergence for services with sparse packets to
improve downlink performance.
Parameter Configurations
MOD CELLALGOSWITCH:LocalCellId=0,CqiAdjAlgoSwitch=StepVarySwitch-1;
Effect Evaluation Methods
Observe throughput improvement and IBLER convergence.
For downlink packet bundling, multiple data packets are packed to consume less control
channel resources and use more data channel resources, thereby increasing capacity
and improving user experience. In eRAN6.0, according to the PDBs of different QCIs,
downlink packet bundling based on delay is implemented. The purpose is to ensure that
all service delays are similar so that service experience is similar. In addition, multiple
small packets are bundled to form a large packet. By doing this, the data channel
capacity can be increased when the control channel capacity does not change.
Therefore, packet bundling applies to the scenario where the scheduling capacity in each
TTI is limited or the PDCCH is limited whereas PDSCH resources are not fully utilized.
The scenarios include:
VoIP scenario: When the number of VoIP services connected to a cell is increasing, the
user satisfaction rate is decreasing starting from a user number until dissatisfaction
appears (the satisfaction rate is lower than the threshold). If packet bundling is available,
the delay of some packets may increase and the mean opinion score (MOS) decreases.
However, the number of satisfied users increases greatly.
Scenario where VoIP and common data services are mixed: Improve the throughput of
common data services.
Decrease the delay of common data services. The delay of some data packets with large
delay noticeably decreases. The delay of some data packets with small delay increases
but the satisfaction remains within the PDB, which means the packet delay distribution is
more balanced. In addition, the resource usage and cell throughput may increase.
Parameter Configurations
MOD CELLALGOSWITCH: LocalCellId=0, DlSchSwitch=NonGbrBundlingSwitch-1;
Effect Evaluation Methods
Target scenarios:
The methods are used when PDCCHs are insufficient while PDSCHs are sufficient:
1. Whether the PDCCH is insufficient
Number of PDCCH CCEs used by common signaling messages:
L.ChMeas.CCE.CommUsed
Number of PDCCH CCEs used by uplink scheduling: L.ChMeas.CCE.ULUsed
Number of PDCCH CCEs used by downlink scheduling:
L.ChMeas.CCE.DLUsed
Total duration of downlink data transmission in a cell (s): L.Thrp.Time.Cell.DL
The PDCCH is insufficient when PDCCH CCE usage is greater than 80%.
PDCCH CCE usage = (L.ChMeas.CCE.CommUsed + L.ChMeas.CCE.ULUsed
+ L.ChMeas.CCE.DLUsed)/(L.Thrp.Time.Cell.DL x 1000 x Number of CCEs
corresponding to the maximum CFI).
In the preceding description, 80% is a reference value. If the UE is nearer to the
eNodeB (SINR = About 25 dB), the value needs to be larger. If the UE is farther
from the eNodeB (SINR = About 0 dB), the value can be smaller.
Table 3.3.3.1.1.1.1.1.1 Number of CCEs corresponding to the largest CFI for each
bandwidth
Bandwidth 3M 5M 10 M 15 M 20 M
CFI = CCE 12 20 41 62 84
corresponding to 3
Result data
4. Increased downlink throughput
Total traffic volume of downlink PDCP SDUs in a cell (by QCI): L.Thrp.bits.DL
Duration of downlink PDCP SDUs in a cell (by QCI): L.Thrp.Time.DL
Average downlink throughput: (by QCI, approximate value): Average Dl Thp =
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
5. Increased downlink PRB usage
RB usage = L.ChMeas.PRB.DL.Used.Avg/Number of RBs in the system
bandwidth x 1000
6. Decreased variance of the average downlink packet delay
Total processing delay of downlink data packets of DRB services in a cell (by
QCI): L.Traffic.DL.PktDelay.Time.QCI.9 (such as QCI9)
Number of successfully transmitted downlink PDCP SDUs of DRB services in a
cell (by QCI): L.Traffic.DL.PktDelay.Num.QCI.9 (such as QCI9)
Average downlink packet delay (by QCI): Average Packet Delay =
L.Traffic.DL.PktDelay.Time.QCI.9/L.Traffic.DL.PktDelay.Num.QCI.9
In this scenario, you are advised to use the PDCCH and PDSCH resource balancing
function. A scheduler randomly selects a UE performing large packet services and
reserves PDCCH resources for the UE. When the last UE is scheduled in the downlink,
PDSCH resources are allocated to the selected UE, thereby improving downlink RB
usage and throughput.
Parameter Configurations
MOD
CELLDLSCHALGO:LOCALCELLID=0,RBPRIMCSSELECTSTRATEGY=SINGLEUSER_STRATEGY,RBPRIMC
SSELECTRATIOTHD=10;
MOD
CELLDLSCHALGO:LOCALCELLID=0,DATATHDINPDCCHPDSCHBAL=18000,UENUMTHDINPDCCHPDSCHB
AL=300;
Length of the polling retransmission timer = Radio transmission delay of the polling PDU
+ Status report scheduling delay at the peer end + Status report radio transmission delay.
Length of the timer for prohibiting status reporting from the receiver = Status report radio
transmission delay + Retransmitted data scheduling delay at the peer end + Radio
transmission delay of retransmitted data.
In the preceding formulas, the radio transmission time can be considered as the fixed
time. In this case, the lengths of the two timers are closely related to the status report
scheduling delay or retransmitted data delay.
On the network, scheduling delay is related to the number of UEs. If there are a large
number of UEs (in the heavy traffic scenario), the interval between scheduling chances of
each UE becomes larger. As a result, the status report or retransmitted data delay also
becomes larger. If the length of the polling retransmission timer is smaller than the
scheduling interval at the peer end, the timer times out and unnecessary retransmission
is triggered when the status report at the peer end has not obtained the scheduling
chance. The increase in retransmission times causes the increase in the probability of
reaching the maximum number of RLC retransmissions, which causes call drop. If the
length of the timer for prohibiting status reporting from the receiver is smaller than the
scheduling interval at the peer end, the timer times out and status reports are sent to the
peer end again when the retransmitted data at the peer end has not obtained the
scheduling chance. As a result, resources are wasted or retransmitted data may be sent
from the peer end for the second time.
Therefore, it is more appropriate if the lengths of the two timers are set according to the
number of UEs on the network. However, the timer settings are fixed currently, failing to
meet actual network environment requirements.
To solve the preceding problem, RLC parameter adaptation configurations are required.
Parameters values are selected according to the number of UEs on the network.
Parameter Configurations
MOD RLCPDCPPARAGROUP: RlcParaAdaptSwitch=1;
In addition, at some sites (for example, the ones in Canada and Australia), after DRX is
enabled, the CQI_ONLY scheduling proportion remains high. This is because UEs do not
report periodic CQIs during the sleep period after DRX is enabled and aperiodic CQI
reporting may be easily triggered because no valid periodic CQI is reported. False SR
detection causes extension in the DRX activation period. As a result, CQI_ONLY
scheduling is triggered.
1. Add restrictions on aperiodic CQI triggering during initial access: "Aperiodic CQI
reporting is triggered 200 ms after DLMAC instances are established" is
changed to "Aperiodic CQI reporting is triggered 200 ms after DLMAC instances
are established and MSG5 is received." Layer 2 identifies MSG5 using
DMAC_ACT_IND sent from layer 3.
2. If CRC check fails after aperiodic CQI reporting is triggered in the downlink,
aperiodic CQI reporting is not triggered after 8 TTIs but until aperiodic CQI
reporting triggering conditions are met.
Parameter Configurations
MOD CELLALGOSWITCH: LocalCellId=0, DlSchSwitch=AperiodicCqiTrigOptSwitch-1;
Parameter Configurations
MOD CELLALGOSWITCH: LocalCellId=0, CellSchStrategySwitch=AbnUeSchSwitch-1;
DL RBLER =
(L.Traffic.DL.SCH.QPSK.ErrTB.Rbler + L.Traffic.DL.SCH.16QAM.ErrTB.Rbler +
L.Traffic.DL.SCH.64QAM.ErrTB.Rbler – L.Traffic.DL.SCH.HOCmd.ErrTB.Rbler –
L.Traffic.DL.SCH.RRCRel.ErrTB.Rbler)/(L.Traffic.DL.SCH.QPSK.TB +
L.Traffic.DL.SCH.16QAM.TB + L.Traffic.DL.SCH.64QAM.TB –
L.Traffic.DL.SCH.HOCmd.TB – L.Traffic.DL.SCH.RRCRel.TB)
UL RBLER =
(L.Traffic.UL.SCH.QPSK.ErrTB.Rbler + L.Traffic.UL.SCH.16QAM.ErrTB.Rbler +
L.Traffic.UL.SCH.64QAM.ErrTB.Rbler)/(L.Traffic.UL.SCH.QPSK.TB +
L.Traffic.UL.SCH.16QAM.TB + L.Traffic.UL.SCH.64QAM.TB)
Parameter Configurations
MOD ENODEBALGOSWITCH: HighLoadNetOptSwitch=SPECSIGRETRANSOPTSWITCH-1;
After the scheme takes effect, the values of the two indicators decrease.
When OuterLoopPucchSwitch (PUCCH outer loop power control switch) is set to ON,
the eNodeB dynamically adjusts SINRTarget of PUCCH format 1a/1b based on the
difference between the BER and BERTarget of PUCCH format 2/2a/2b so that PUCCH
power control changes with the channel status, thereby improving PUCCH demodulation
performance, reducing the false detection rate of format 2/2a/2b, and increasing downlink
throughput.
When OuterLoopPucchSwitch (PUCCH outer loop power control switch) is set to OFF,
the fixed SINRTarget is used for PUCCH inner loop power control.
Parameter Configurations
MOD CELLALGOSWITCH: UlPcAlgoSwitch = OuterLoopPucchSwitch-1;
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=DlPacketLenAwareSchS-1;
Off:
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=DlPacketLenAwareSchSw-0;
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Parameter Configurations
Run the following command to enable this function:
(L.Thrp.bits.UL – L.Thrp.bits.UE.UL.SmallPkt)/L.Thrp.Time.UE.UL.RmvSmallPkt
The following formula is used to determine whether the downlink user throughput
increases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
3.3.12 Delayed Transmission of Status Reports
Introduction
When CCE resources are limited on the live network in big events, the eNodeB judges
downlink connections to be scheduled in each TTI at the MAC layer to save CCE
resources. If a connection to be scheduled contains only a downlink RLC ACK status
report (without downlink data to be sent) and the status report scheduling wait time is less
than the preset threshold, status report transmission is delayed in the current TTI. After
delayed transmission of status reports takes effect when CCE resources are limited on
the live network in big events, user-perceived downlink throughput can be increased.
Parameter Configurations
To enable this function, run the following command:
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=DlRLCStateReportSchDelaySw-1;
MOD CELLALGOSWITCH:LOCALCELLID=0,DLSCHSWITCH=DlRLCStateReportSchDelaySw-0;
You can observe the counters in the following formula to determine whether user-
perceived downlink throughput increases:
(L.Thrp.bits.DL – L.Thrp.bits.DL.LastTTI)/L.Thrp.Time.DL.RmvLastTTI
Parameter Configurations
To enable enhancement of adaptive downlink HARQ retransmission:
Performance Prediction
The user-perceived downlink rate increases by 1% to 5% in heavy-traffic scenarios where
the downlink PRB usage is greater than 50% and the downlink IBLER is between 9% and
30%, the proportion of UEs at the cell center is lower than or equal to 25%. The downlink
RBLER is expected to possibly increase.
You can observe the counters in the following formula to determine whether the number
of downlink RB retransmissions reduces:
Related features:
FDD:
1) AMC optimization in burst
Downlink throughput and SINR curve at the cell edge
DT interference scenarios
or at a medium distance from the cell center: 5–10%
2) Increase of the shutdown
in 4T scenarios and 5% in 2T scenario
ratio
FDD:
User Related features:
Heavy The PRB usage is greater than 50%, the CFI is
plane 1) CFI adaptation optimization
load greater than 2, and the downlink UE throughput
KPI (FDD)
increases by 5%.
5 Cases
The following web page lists the gains of some optimization measures taken in offices on
the live network:
http://support.huawei.com/support/pages/kbcenter/view/product.do?
actionFlag=detailProductSimple&web_doc_id=SC0000822317&doc_type=12
3-2&doc_type=123-2&saveBrowserLoged=true
6 Reference Documents