Adaptive Modulation and Coding

Outer Loop Quality Control

OLQC CQI Adaptation

Outer Link Quality Control (OLQC) adapts the channel quality information
that is used by the scheduler and link adaptation to achieve target block
error ratio (BLER) for the first transmission of a Transport Block

OLQC compensates any non-idealities of the link adaptation :

CQI estimation error of the UE
CQI quantization error
CQI reporting error
Time delay between CQI measurement and the reception of the subsequent data
block
CQI interpolation error dlOlqcEnable
Enable/disable OLQC
Errors due to CQI averaging of PRBs LNCEL; false/true; true

dlOlqcEnable parameter is used to enable/disable the outer link quality

control. When outer link quality control is disabled then the corrected CQI
values correspond to the reported CQI values.
OLQC CQI Adaptation

The picture below shows the principle of CQI adaptation

OLQC CQI Adaptation

OLQC adds a CQI offset to the CQI reports that are provided by the UE via the UL L1/L2 signaling.

The corrected CQI report will then be provided to the DL link adaptation and MIMO mode control for
selection between Transmit diversity and Spatial Multiplexing

Starting from the initial value the CQI offset will be adjusted in response to the ACK/NACK for the first
transmission of a transport block that is provided by the UE via L1/L2 control signaling

The averaging period for the CQI values reported by the UE are set by the MIMO mode Control
parameters mimoClCqiAvg & mimoOlCqiAvg
OLQC CQI Adaptation
Single Code Word
For a correct received transport block the CQI report will be increased by a value
CQIstepup
For an incorrect transport block the value will be decreased by CQIstepdown.
No change will be done when no ACK/NACK is available or when it is a
retransmission of the corresponding transport block.
A maximum and a minimum CQI offset is defined (called DCQImax and DCQImin) in
order to suppress very large fluctuations that may arise in extreme situations

CQI corrected ( x, t ) = CQI reported ( x, t ) + CQI(t ) .

min(CQI(t 1) + CQIstepup , CQI max ), for first HARQ feedback = ACK,

CQI(t ) = max(CQI(t 1) CQIstepdown , CQI min ), for first HARQ feedback = NACK,
CQI(t 1), for first HARQ feedback = N/A.

dlOlqcDeltaCqiIni, dlOlqcDeltaCqiMax,
dlOlqcDeltaCqiMin, dlOlqcDeltaCqiStepUp
All are vendor specific parameters
 OLQC CQI Adaptation
Two Code Words

CQI corrected ( x, t ) = CQI reported ( x, t ) + CQI(t ) .

min(CQI(t 1) + CQI stepup , CQI max ),
max(CQI(t 1) CQI
stepdown , CQI min ),
min(max(CQI(t 1) + (CQI stepup CQI stepdown ) / 2,
CQI(t ) = CQI min ), CQI max ),
min(CQI(t 1) + CQI stepup , CQI max ),
max(CQI(t 1) CQI stepdown , CQI min ),
CQI(t 1),
for new transmisssion HARQ feedbacks = ACK * + ACK*,
for new transmission HARQ feedbacks = NACK * * + NACK * *,
for new transmission HARQ feedbacks = ACK * + NACK * *,
for new transmission HARQ feedbacks = ACK * + N/A,
for new transmission HARQ feedbacks = NACK * * + N/A,
for new transmission HARQ feedbacks = N/A + N/A.

* the ACK feedback for a new transmission shall be mapped to N/A for the considered code word in the following two cases in order to supress an upgrade of the CQI offset:

The considered new transmission of the corresponding code word has been done with the maximum possible MCS that is possible for the considered UE (UE capabilities) and eNodeB (enabling/disabling of certain
modulation of coding formats). In this case only a downgrade of the MCS and the CQI offset shall be allowed since the maximum possible MCS is reached and the channel conditions might be much better than indicated by
the MCS.

MCS downgrade has been done for the considered new transmission of the considered code word (due to the fact that the selected transport block exceeds the peak data rate or the amount of buffered data of the considered
UE). In this case only a downgrade of the MCS and the CQI offset shall be allowed since the channel conditions might be better than indicated by the MCS.

** the NACK feedback for a new transmission shall be mapped to N/A for the considered code word in order to suppress a downgrade of the CQI offset:

The considered new transmission of the corresponding code word has been done with the minimum possible MCS that is possible for the considered UE and eNodeB. In this case only an upgrade of the MCS and the CQI
offset shall be allowed since the channel conditions might be worse than indicated by the MCS.
OLQC CQI Adaptation

OLQC algorithm targets to achieve a Target DL BLER (dlTargetBler)

Based on the Target BLER, OLQC will determine whether the reported CQI is increased or
decreased
CQI offset is defined by considering the CQI increase and decrease steps which should be
balanced.
Assuming that a CQI decrease will occur with a probability BLERtarget for the first transmission
whereas a CQI increase occurs with the complementary probability (1-BLERtarget) the balance
equation can be formulated as:

(1 BLER target ) CQI stepup = BLER target CQI stepdown .

Therefore, CQIstepdown can be calculated from the parameters CQIstepup and the dlTargetBler as:

dlTargetBler
1 - BLER target Target BLER DL
CQI stepdown = CQI stepup .
BLER target LNCEL;0.0010.999;0.001; 0.1
DL Adaptive Modulation and Coding (AMC)
(AMC)
DL AMC

The target of the DL Adaptive Modulation and Coding (AMC) algorithm is to improve system capacity, peak
data rate and coverage reliability

The transmitted signal by a particular user is modified to account for the signal quality variation through link
adaptation.

The aim of the link adaptation is to estimate the transport block size for a UE and a certain set of allowed
resource blocks (frequency resources) for transmission

For the Downlink Data Channel a fast Adaptive Modulation and Coding (AMC) functionality based on UE
reported CQI is performed by the AMC algorithm

AMC selects a suitable Modulation and Coding Scheme (MCS) for the PRBs/RBGs assigned to a UE as
indicated by the downlink scheduler.
DL AMC

DL AMC is enabled via

dlamcEnable

For retransmissions the

same MCS as for the
original transmission is
applied

For new transmissions the

MCS is decided based on
CQI reports from the UE

dlamcEnable
Enable Adaptive Modulation and
Coding in DL
LNCEL; true, false ; true
DL AMC
MCS Selection for new Transmissions

For the first transmissions where no previous CQI information is available

from the UE the DL AMC will provide the initial MCS to be used for the UE

Initial MCS is specified by iniMcsDl

If DL AMC is disabled via dlamcEnable, then no link adaptation will be

performed and a fixed MCS shall be applied according to iniMcsDl (Initial
MCS for the DL) and the applied MCS shall never be changed over the
time

dlamcEnable iniMcsDl
Enable Adaptive Modulation and Initial MCS in DL
Coding in DL LNCEL; 028; 1 ; 4
LNCEL; true, false ; true
DL AMC
MCS Selection for new Transmissions
An average channel state is determined from the CQI information corresponding to the
PRBs/RBGs having been assigned (or are considered for being assigned) by the scheduler.
For this the CQI values have to be mapped to CIR values via link level simulation results.
The CIR values for the allocated PRBs/RBGs are averaged linearly. The resulting averaged
CIR value is converted back into an averaged CQI value (but not quantized to full CQI steps;
interpolation between full CQI steps needs to be applied).
The target for this mapping is to choose the transport block out of the group of possible
transport blocks such that the difference of the code rate as derived from the averaged CQI
value (according to the CQI table using interpolation to account for the non-quantized value of
the averaged CQI) and the effective code rate corresponding to the considered transport block
is minimized.
The MCS index as corresponding to the transport block as determined above shall be used
and signaled accordingly.

The mapping of the averaged CQI to an MCS level shall be performed by mapping the
modulation scheme and code rate of the CQI table according to [3GPP-36.213]
DL AMC

MCS Selection for new Transmissions

In case of the averaged CQI value falling in-between two CQI indices with different corresponding modulation
scheme, the scheme with the lower modulation order will be chosen

For dual codeword transmission link adaptation has to be performed per codeword if CQI information per
codeword is available (i.e., for closed loop MIMO transmission mode).

If only wideband CQI information is available for a UE the corresponding MCS level can be mapped directly
(without a preceding averaging step).

If no new CQI values were received for a UE, and the UE is scheduled nevertheless, the MCS shall be
determined as described above provided the latest available CQI information is not older than dlamcTHistCqi
If dlamcTHistCqi is exceeded (or CQI values are not yet available) the initial MCS (iniMcsDl) shall be applied.

iniMcsDl dlamcThistCqi
Initial MCS in DL Time in TTIs for which historical
CQI is remembered in AMC
LNCEL; 028; 1 ; 4
Vendor-specific parameter
DL Adaptive Modulation and Coding (AMC)
for PDCCH
 enableAmcpdcch
Main target of DL-AMC-CCH Enable/disable CQI
based AMC for PDCCH
LNCEL; true; true

Similar to data transmission, it is necessary to make a signaling (PDCCH) robust enough for poor
UEs (low SINR, e.g. at the cell-edge)
Transmission with low ECR (Effective Coding Rate) leads to increased resource utilization which
reduces the number of scheduled UEs; thus good UEs should occupy less PDCCH resources and
operate with lower number of CCEs (higher ECR)
7 UEs (5 MHz), 10 UEs (10 MHz), 20 UEs (20 MHz)
Any Link Adaptation technique must deal with a trade-off between signaling robustness (coverage)
and signaling capacity
Macro cell case #1
Uniform UE distribution

4-CCE
8-CCE 2-CCE 1-CCE
CQI-to-Aggregation Mapping unit
DLS_INPUT_LIST =
{
CQI from DL-AMC/DL-OLQC
CQI filtering/processing
}
Broadcast, Tag, DCI-format, CSS, Prio-A;
Paging, Tag, DCI-format, CSS, Prio-B;
RACH Response, Tag, DCI-format, CSS, Prio-C;
Preamble Assignment, Tag, DCI-format, CSS, Prio-D;
Message 4 Assignment, Tag, DCI-format, CSS, Prio-E;
UE-1, Tag, DCI-format, USS, Prio-X;
UE-2, Tag, DCI-format, USS, Prio-Y;
;
UE-k: ...;

ULS_IINPUT_LIST =
Filtered, compensated {
and shifted CQI UE-1, Tag, USS, Prio-X;
UE-2, Tag, USS, Prio-Y;
;
All DCI
formats UE-k: ...;
}
1
CQI-to-Aggregation Mapping
1a

CQI-to-Aggregation Mapping unit relies on UE-specific CQI
information to build the list of required AGG levels for all possible DCI
formats for every active UE (UE which appears on the DL/UL scheduling
REQUIRED_AGG_LIST = list).
{ REQUIRED_AGG_LIST must refer to all active UEs so that the
UE-1: pdcchCQI, AGG-DCI0, AGG-DCI1, ; schedulers know how many resources are needed to allocate them.
UE-2: pdcchCQI, AGG-DCI0, AGG-DCI1, ;
Common signaling (e.g. Broadcast, Paging, etc.) is not considered at
; this step; the mapping affects UE Search Space (USS) only.
UE-k: ...;
}
CQI-to-Aggregation Mapping unit rdPdcchAggTables

SINR-vs-BLER tables have been

obtained from 4GMax LL
simulator (EPA05, 2x2MIMO) for
two representative payload sizes
of 45 bits and 60 bits.
The PDCCH performance
should aim at 1% target BLER.
SINR targets have to be
translated to CQI thresholds.
CQI = 0.51*SINR + 5.3

The mapping table is not

sufficient. R&D in-built table
must consist of thresholds for all
possible DCI formats (various
payload size). A scaling factor
(SF) is applied.

SFsmallDCI = 10*log10(DCI_size/45)

SFlargeDCI = 10*log10(DCI_size/60)
Mapping table for 45/60 bits payload composed
based on CQI-to-SINR formula (4GMax)
CQI-to-Aggregation Mapping unit
pdcchAggDefUE
rdPdcchAggTables PDCCH LA UE default aggregation;
LNCEL; 1(0), 2(1), 4(2), 8(3); 4(2)

After post-processing of 4GMax output, the

table is ready to be used by the CQI-to-
Aggregation Mapping unit.
The table is valid for 10MHz bandwidth,
however the operator can adjust the Mapping table for 45/60 bits payload composed
thresholds by using O&M parameter based on CQI-to-SINR formula (4GMax)
pdcchCqiShift (if OLLA PDCCH is disabled,
see below). CQI = 0.51*SINR + 5.3
If PDCCH AMC is disabled or CQI is outdated, SFsmallDCI = 10*log10(DCI_size/45)
pdcchAggDefUe will be applied to all DCI
pdcchCqiShift SFlargeDCI = 10*log10(DCI_size/60)
formats of all UEs.
LNCEL; -1010;0.1; 0

Full rdPdcchAggTables for all available DCI formats (10MHz system bandwidth)
OLLA for PDCCH - Motivation

PDCCH carries information about the resources assignments for both Uplink (UL) and Downlink (DL) data channels.
Scheduling
Downlink scheduling grant
request
(MCS, PRBs, ..) Uplink scheduling grant
(MCS, PRBs, ..)

TTI n TTI n PUCCH/PUSCH

PDCCH
TTI n+x PDCC
PDSCH
H
UE e-NB TTI n+y PUSCH UE
e-NB Data Data

If a PDCCH payload is missed the User Equipment (UE) cannot know whether it has been scheduled and on which
time/frequency resources.

? TTI n PUCCH/PUSCH
TTI n PDCCH
TTI n+x PDCCH
PDSCH
PUSCH UE
UE e-NB TTI n+y ?
e-NB
Waste of Resources!
OLLA for PDCCH Principle

The PDCCH OLLA can be based on the PDSCH OLLA as follows:

Offset_PDCCH = deltaCQI + pdcchCqiShift,

deltaCQI from OLQC is used to control the PDSCH and PDCCH inner loop link (ILLA) adaptation. It is the PDSCH
OLQC offset available and calculated based on the Ack/Nack/DTX feedback from previous PDSCH transmission

and pdcchCqiShift is a term needed to compensate for the difference in BLER target for the PDSCH (e.g. 10%) and
PDCCH (e.g. 1%).

pdcchCqiShift is in use to fine tune the PDCCH BLER

The value is controlled statically by O&M: pdcchCqiShift or
dynamically by the feature.
In the latter case its computation is based on a similar algorithm as used for PDSCH OLLA and
the target BLER PDCCH is O&M defined by pdcchHarqTargetBler
OLLA for PDCCH Functional Model

Decides #CCEs and trans-mission

power per UE (more CCEs/ power to
low SINR UEs)
PDCCH Link Adaptation
PDCCH Inner New
WideBand CQI PDCCH Power and
Loop Link Existing
Adaptation Aggregation Level

deltaCQIShift actOlLaPdcch
deltaCQI PDCCH Outer OLLA for PDCCH: activation
Loop Link
True, false
Adaptation
Adjusts the dynamic
correction of UE SINR HARQ
estimate per user
PDSCH Outer Link Feedback
Quality Control Ack/Nack/DTX (for initial DL
transmission)
deltaCQI
PDSCH Inner pdcchHarqTargetBler
Frequency Loop Link BLER target for PDCCH outer
Selective CQI Adaptation loop link adaptation
0.1 %, 0.2 % 3.0 %
PDSCH Link Adaptation
/
PDCCH Scheduling
PDCCH carries DCI (Downlink Control Information) to inform UE
about UL and DL Resource Block allocation for user data transmission

After UL and DL scheduling list was generated by UL and DL scheduler,

this algorithm selects UEs for UL and DL physical resource block allocation
Share between UL and DL can be set with parameter LNCEL: pdcchUlDlBal
pdcchUlDlBal
PDCCH allocation balance constant between UL and DL. The
PDCCH UE-specific search space capacity is divided into UL
and DL based on the parameter before dynamic schedulings
LNCEL; 00.9; 0.05; 0.5

ZIG-ZAG is not used anymore (trial approach)

only if entries have same priority level
UE specific search space can be limited by LNCEL: PDCCHAlpha
[0,5 .. 2], step 0,05, default = 0,8 (PDCCH UE search space capacity
is multiplied with this parameter)
Usage based PDCCH adaptation (LTE616)
Number of OFDM symbols for PDCCH is adapted to the required amount of CCEs
Balance between DL and for UL may be adapted basedon CCE blocking and used CCE
Activation: by actLdPdcch (true, false)
symbols for PDCCH in a TTI is between the minimum reasonable (bandwidth-dependent) and maxNrSymPdcch
The parameter actLdPdcch can be set to True if all 3 conditions are fulfilled:
- phichDur is set to Normal
- maxNrSymPdcch value is greater than 1
- dlChBw value is greater than 25
The Load Adaptive PDCCH algorithm optimizes the value depending on the actual load distribution between DL
and UL,
pdcchUlDlBal parameter defines the initial value.
PDCCH scheduling

TOTAL_INPUT_LIST_DL_
AMC may be composed ZIG-ZAG approach Prio-oriented approach
Joint
based on different joint list list
Uplink
list
creation techniques High
priority
High
Joint list
Downlink priority
however as mentioned on list
Highest
priority
Common
the previous slide, Prio- High
priority signalling (CSS),
oriented solution has been Msg4
chosen for RL10
ZIG-ZAG solution is valid UL HARQ
only in case of entries with
the same priorities
Dedicated
signalling (USS)
Low
priority Scheduling
Low requests
priority

Other DL/UL
Lowest
Low priority
priority
PDCCH scheduling

pdcchUlDlPrio
PDCCH LA priorities
LNCEL; 099; 1; -
Parameter is vendor specific Structured Parameter: 15 single parameters not configurable
UL Adaptive Modulation and Coding (AMC)
UL AMC and ATB
Used to adapt PUSCH to different
link conditions by variable
modulation and coding scheme,
and variable bandwidth
Inputs
Ack/Nack information
SINR Measurements
Power Headroom reports
Outputs
modulation
code rate
maximum amount of PRBs
NSN Implementation
Simple BLER based algorithm to select best
MCS according to UE specific Radio
Conditions
SINR Measurements are not required, only
HARQ Ack/Nak output signalling used
Target BLER adjustable
Emergency Downgrade/Upgarde Feature
combatting very high or low BLER peaks
Adaptive Transmission Bandwidth (ATB) for
controlling maximum PRB amount based on
power headroom reports
UL AMC
ulamcAllTbEn
O&M switch for enabling/disabling the counting of
all TBs instead of the 1st transmission TB for
defining UL AMC inner loop factor.
LNCEL; true; true

The uplink AMC shall consist of 2 main components:

1. An outer loop LA (OLLA) acting on the 1st Transmission Errors on TBs
2. An inner loop LA based on BLER acting on either 1st Transmission Errors of TBs or on all TBs
transmission errors derived from HARQ process.
UL inner loop LA is a slow LA which will be performed every ulamcSwitchPer
OLLA is an event based driven LA and will provide the capability to adjust to fast
changing radio conditions by performing emergency downgrades or fast upgrades
of the MCS
ulamcSwitchPer
Period in sent Transport Blocks
TBs when UL Inner Loop LA
should be executed
LNCEL; 10500; 10; 30 TBs
UL AMC
ulamcEnable
Enable Adaptive Modulation and
Coding in UL
LNCEL; true, false ; true
UL AMC can be enabled/disabled with ulamcEnable
If UL AMC is disabled, then no LA will be performed and a fixed MCS shall be applied according to iniMcsUl
(Initial MCS for the UL) and the applied MCS shall never be changed over the time
If UL AMC is enabled then the data transfer of every UE shall start with iniMcsUl and the MCS shall change over
time according to radio conditions.
iniMcsUl
UL AMC shall provide the following functions: Initial MCS in UL
LNCEL; 020; 1; 5
BLER averaging
OLLA which provides Emergency Downgrade (EDG) and Fast Upgrade (FUG) Events
MCS selection based on BLER providing the optimum MCS depending on radio link conditions
UL ATB derived from selected MCS according to radio conditions. UL ATB process results in an upper PRB
allocation limit submitted to the UL scheduler.
ulamcEdgFugEn
O&M switch for enabling/disabling the 1st
transmission BLER based Emergency Downgrade
and Fast Upgrade functionality included in the UL
AMC Outer Loop Link Adaptation.
LNCEL; true; true
UL AMC

UL AMC shall select the MCS to be employed from the table below according to
the radio conditions
MCSIndex Modulation TBS Redundancy
IMCS Order Index Version

0
Qm'
2
ITBS
0
rvidx
0
In RL09/RL10 , 64 QAM modulation is not available in
1
2
2
2
1
2
0
0
the UL, therefore selected MCSs will be restricted from
3
4
2
2
3
4
0
0
MCS 0 to MCS 20
5 2 5 0
6 2 6 0
7 2 7 0
8 2 8 0
9 2 9 0
10 2 10 0
11 4 10 0
12 4 11 0
13 4 12 0
14 4 13 0
15 4 14 0
16 4 15 0
17 4 16 0
18 4 17 0
19 4 18 0
20 4 19 0
21 6 19 0
22 6 20 0
23 6 21 0
24 6 22 0
25 6 23 0
26 6 24 0
27 6 25 0
28 6 26 0
29 1
30 reserved 2
31 3
UL AMC Increased UL MCS Range

UL peak throughput can be enhanced (see table)

Additionally supported: MCS21-MCS24 (LTE829)
Applicable to 16QAM UEs (cat1-cat4)

actModulationSchemeUL
Enable 16QAM high MCS
LNCEL;
QPSK (0), 16QAM (1), 16QAMHighMCS (2)

UL Peak
Max # Improve-
BW / MHz Rate /
PRB ment / %
Mbps
3
5
10
15
20 96 of 100 51.0 26

Exercise: Please complete the table.

Inner Loop Link Adaptation (ILLA)
(ILLA)
Inner Loop LA
ulTargetBler
LNCEL; 1050%; 1%; 10%

The target of the inner loop LA is to maintain a UEs BLER close to the established target BLER ,
which is established by the parameter: ulTargetBler
Note that user data and L3 signaling are multiplexed together on PUSCH a will therefore have a
common BLER Target
Inner Loop LA is based on BLER measurements which are calculated based on the ack/nack
feedback obtained from L1/L2
Inner loop LA will be performed every time the timer ulamcSwitchPer expires
Based on the UEs actual BLER compared to the desired target BLER AM will make a decision
whether to upgrade or downgrade the MCS

ulamcSwitchPer
Period in sent Transport Blocks
TBs when UL Inner Loop LA
should be executed
LNCEL; 10500; 10; 30 TBs
Inner Loop LA
ulTargetBler
LNCEL; 1050%; 1%; 10%

Thresholds for upgrade and downgrade of MCS are established by the following
parameters:
ulTargetBler : Target BLER for the Uplink ulamcUpdowngrF
ulamcUpdowngrF: Upgrade/Downgrade Factor LNBTS; 13; 0.05; 1.2
Vendor specific parameter

High BLER Threshold (Downgrade) = Round(ulTargetBler * ulamcUpdowngrF)

Low BLER Threshold (Upgrade) = Round(ulTargetBler / ulamcUpdowngrF)

Note Upgrade and downgrade are always performed by a single MCS step
ensuring that the maximum and minimum possible MCSs aren't surpassed

Target BLER

Low BLER High BLER

Upgrade MCS Maintain MCS Downgrade MCS

Inner Loop LA
Example:
ulTargetBler : 10%
ulamcUpdowngrF: 1.5

High BLER Threshold (Downgrade) = Round(ulTargetBler * ulamcUpdowngrF) = 15%

Low BLER Threshold (Upgrade)= Round(ulTargetBler / ulamcUpdowngrF) =7%

7% 15 %

Target BLER

Low BLER High BLER

Upgrade MCS Maintain MCS Downgrade MCS

Inner Loop LA
ulamcAllTbEn
O&M switch for enabling/disabling the counting of
all TBs instead of the 1st transmission TB for
defining UL AMC inner loop factor.
LNCEL; true; true
There are 2 possibilities to calculate the BLER for the Inner Loop LA (ulamcAllTbEn):
Consider only 1st TB Transmissions: BLER does not take into account any HARQ gains achieved by soft
combining
Consider all Transmissions: the HARQ gain is included leading to small decision errors
Performance of the inner loop AMC is going to be highly dependent on UL AMC switch period (ulamcSwitchPer
):
High Values of this parameter will ensure more stability in the LA process but worst reaction to fast radio
condition variations
Low values of this parameter will provide faster reactions of LA and additionally decrease the resolution the
BLER measurements causing possibly more instability
Example: ulamcSwitchPer = 100 TBs ~ 100 ms there are 100 TBs and by that a BLER resolution of 1%, within
50 ms there are 50 TBs and the BLER resolution is 2% and for 20 ms there are 20 TBs and BLER resolution is
5%. This should be considered when setting ulamcUpdowngrF

ulamcUpdowngrF ulamcSwitchPer
LNCEL; 13; 0.05; 1.2 LNCEL; 10500; 10; 30 TBs
Outer Loop Link Adaptation (OLLA)
OLLA

OLLA is based on the 1st transmission ACK/NACK information provided by L1/L2 HARQ.
OLLA provides a quicker adaptation to radio conditions compared to the inner loop LA which typically will
act every 100-500ms defined by ulamcSwitchPer
OLLA basically counts the BLER based on 1st transmissions (C)
min( C (t 1) + C stepup , C max ), for first HARQ feedback = ACK,

C ( t ) = max( C ( t 1) C stepdown , C min ), for first HARQ feedback = NACK,
C ( t 1), for first HARQ feedback = N/A.

Where:
Cmax and Cmin give upper and lower limits on the compensation defined by parameters
(ulamcDeltaCmax, ulamcDeltaCmin)
Cstepup and Cstepdown are incremental compensation steps sizes, which obey to the following formula:
1 - BLER target
Cstepdown = Cstepup .
BLER target

Where Cstepup and BLERtarget are parameters: ulamcCStepUp , ulTargetBler

OLLA

Every time OLLA is initialized or reset C is set to ulamcDeltaCini

OLLA compensation value C is reset at each AMC period, EDG and FUG event.

Emergency Downgrade (EDG) shall be triggered, whenever the compensation value C is equal to Cmin.
AMC shall switch immediately to the next lower (i.e. more robust) MCS

Fast Upgrade (FUG) shall be triggered, whenever the compensation value C is equal to C max. AMC shall
switch immediately to the next higher (i.e. less robust) MCS
Note that ulamcDeltaCmin and ulamcDeltaCmax as well as ulamcCStepUp have to be configured carefully
depending on adjusted ulTargetBler, for example a certain number of consecutive Nacks has to be assumed,
which shall trigger the EDG

ulamcDeltaCini ulamcDeltaCmin ulamcDeltaCmax ulamcCStepUp

LNBTS; -10..10; 0.1; 0 LNBTS; -10..10; 0.1; -5 LNBTS; -10..10; 0.1; 5 LNBTS; 0..2; 0.05; 0.2

Vendor specific parameter Vendor specific parameter Vendor specific parameter Vendor specific parameter
OLLA ulamcEdgFugEn
O&M switch for enabling/disabling the 1st
transmission BLER based Emergency Downgrade
and Fast Upgrade functionality included in the UL
AMC Outer Loop Link Adaptation.
LNCEL; true; true

Emergency downgrades and fast upgrades functionality of the OLLA can be disabled with the parameter:
ulamcEdgFugEn
If OLLA is disabled only the slow BLER based inner loop AMC shall control the MCS selection.
Gain of EDG is shown from the simulation results below:
UL Throughput vs. SNR; 20 MHz LTE System, RX-Div, AWGN Channel

50000
45000
10% BLER
40000 Targets of 1st
Throughput [kbps]

35000 Transmission
30000
25000
20000
15000
10000
5000
0
-16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16
SNR [dB]

QPSK 1/6 QPSK 1/3 QPSK 1/2

QPSK 2/3 16QAM 1/2 16QAM 2/3
16QAM 5/6 Thrp. Envelope w/o EDG Thrp. Envelope with EDG
Inner Loop and Outer Loop Interaction
OLLA
Comp. C

Max FUG
Cmax
Reduced
Event
AMC Period

ulamcSwitchPer
0 Time
Period in sent Transport Blocks
TBs when UL Inner Loop LA
AMC Switching Period Reduced should be executed
EDG AMC Period
Min Event LNCEL; 10500; 10; 30 TBs
Cmin

Inner loop LA is periodical based on the parameter

ulamcSwitchPer
OLLA is event based
Every time a EDG or FUG event takes place the Inner loop
LA is reset too, therefore the periodicity of the inner loop LA
can be shortened by the OLLA events
Inner Loop and Outer Loop Interaction

Example:
ulTargetBler =10% ulamcDeltaCini =0
ulamcCStepUp = 0.1 1- BLERtarget
ulamcCStepDown=0.1*(1-0.1)/0.1 = 0.9. Cstepdown = Cstepup .
BLERtarget

So setting ulamcDeltaCmin=-9 means, that an EDG will be triggered after

10 unsuccessfully received consecutive 1st transmission TBs after every
MCS upgrade/downgrade event
min( C (t 1) + C stepup , C max ), for first HARQ feedback = ACK,

C ( t ) = max( C ( t 1) C stepdown , C min ), for first HARQ feedback = NACK,
C ( t 1), for first HARQ feedback = N/A.

By such an adjustment of the UL AMC with OLLA will be able to switch

down the MCS after 10ms (10 TTIs) even if the ulamcSwitchPer is rather
high with e.g. 50 TBs (equals to ~50ms).
Note that the BLER calculation is reset after each AMC period
UL AMC During DRX/DTX

At the end of data transfer the currently selected MCS shall be stored and a Timer
for historical MCS shall be started.

If the same UE proceeds with a data transfer within the time period ulamcHistMcsT,
then the historical MCS shall be reloaded from memory and applied instead of the
iniMcsUl iniMcsUl
LNCEL; 020; 1; 5

By setting ulamcHistMcsT = 0 the functionality of historical MCS can be switched

off.

Before starting an UE specific DTX period or entering an Inactivity period the actual
MCS shall be stored and a Timer for Inactivity shall be started. With every
ulamcInactT period the MCS shall be decreased, but the selected MCS shall not go
below the initial MCS iniMcsUl.

If the currently selected MCS is below iniMcsUl then no action during DRX/DTX
and/or Inactivity period shall be required.
Adaptive Transmission Bandwidth (ATB)
Adaptive Transmission Bandwidth (ATB)
Besides selecting the most appropriate MCS according to radio conditions, the UL AMC shall also perform slow ATB in
parallel. (i.e. fast means every TTI)

ATB is necessary in case of lack of UE power to concentrate the remaining power on less PRBs, thus allowing a regular
data transmission in UL even up to the cell edge.

ATB will inform the scheduler about the maximum Number of PRBs per TTI that can be assigned to a UE based on the
UEs power headroom reports

The periodicity of ATB is defined by the parameter ulatbEventPer which defines a multiple of AMC events (periodic
changes, EDG, FUG) after which ATB will be carried out
ATB functionality can be enabled/disabled with ulatbEnabled
Note that after every EDG and FUG event the slow ATB limits have to be recalculated since the MCS might have
changed.

ulatbEventPer
Period in MCS increase/decrease
events when UL ATB functionality
should be performed.
LNCEL; 150; 1; 1
ulatbEnable
O&M switch for enabling/disabling
the UL ATB functionalities.
LNCEL; true ; true
Adaptive Transmission Bandwidth (ATB)

Trigger conditions for UE to send Power headroom reports:

dlPathlossChg : When UE surpasses a defined threshold of power headroom it
shall report it to the eNodeB. This event driven report will handle fast variations of
the path loss

tPeriodicPhr: Parameter to set periodic reporting of the power headroom

tProhibitPhr: Parameter to define minimum interval between power headroom

reports sent to eNodeB

dlPathlossChg tPeriodicPhr tProhibitPhr

This is a trigger condition for power
headroom submission due to
pathloss change
LNCEL; 1 db (0), 3 db (1), 6 db (2),
infinite (3); 3dB (1)
Period for periodic Power Headroom
Reports
LNCEL; 10sf (0), 20sf (1), 50sf (2),
100sf (3), 200sf (4), 500sf (5),
1000sf (6), infinity (7); 20sf (1)
20sf = 20 ms
Minimum intermediate time between
two consecutive Power Headroom
Reports
LNCEL; 0sf (0), 10sf (1), 20sf (2),
50sf (3), 100sf (4), 200sf (5), 500sf
(6), 1000sf (7); 0sf (0)
0sf = 0 ms
Adaptive Transmission Bandwidth (ATB)

ATB Algorithm:

1. At call setup the maximum number of PRBs that can be allocated to a

single UE shall be limited by the parameter iniPrbsul
iniPrbsul
Initial amount of PRBs in UL
LNCEL; 1100; 1; 10

2. ATB events shall act synchronously with the slow AMC, based on
ulatbEventPer
ulatbEventPer
Period in MCS increase/decrease
events when UL ATB functionality
should be performed.
LNCEL; 150; 1; 1
Adaptive Transmission Bandwidth (ATB)
ATB Algorithm:
3.At call ATB calculates a running average filter acting continuously on all of the incoming power headroom reports
of a certain UE. The averaging period is defined by means of ulatbPhrAvgF

Power head room reports depend on the number of PRBs which were scheduled to the UE. Information on the number of scheduled
PRBs is obtained from the UL Scheduler

The equivalent possible PRBs derived from PWR_HEADR_UL and UE_PRBs_UL for a certain time instance t shall be given by:

PWR_HEADR_PRBs(t) = UE_PRBs_UL(t) * PWR_HEADR_UL(t).

For this PWR_HEADR_UL has to be linearized (converted from dB into linear scale), e.g. 3 dB is factor 2 and -3 dB is factor 1/2 and 0 dB
is a factor 1.

The running average filter output is given by

RUNAVG_PRBs(0) = iniPrbsul.
RUNAVG_PRBs(n) = (1 - ulatbPhrAvgF)*RUNAVG_PRBs(n-1) + ulatbPhrAvgF *PWR_HEADR_PRBs(n).

ulatbPhrAvgF
Parameter used for time averaging of
power headroom reports
LNBTS; 0.9
Vendor specific parameter
Adaptive Transmission Bandwidth (ATB)
Note:
with ulatbPhrAvgF = 1 always the last power headroom report is used
with ulatbPhrAvgF = 0 the ATB is disabled and always the initial setting is employed (this is a second possibility to switch the algorithm off).

4. At any ATB decision the present value of the running average filter is read and the max number of PRBs is set
to a rounded integer value by:
MAX_NUM_PRBs = floor( RUNAVG_PRBs ).

5.Ensure that PRBs are within and upper and lower limit boundaries:
UPPER_LIMIT_PRBs = MAX_BITRATE_UL (given by Admission Control and QoS) / (MCS_THROUGHPUT_per_PRB*(1-
ULAMC_TARGET_BLER))
The upper Limit shall not exceed #PRBs_UL given by the Carrier Bandwidth.

The lower Limit is given by:

LOWER_LIMIT_PRBs = MIN_BITRATE_UL (given by Admission Control and QoS) / (MCS_THROUGHPUT_per_PRB*(1-
ULAMC_TARGET_BLER))
MCS_THROUGHPUT_per_PRB is the MCS dependent UE throughput under ideal radio conditions (0% BLER) assuming a fictive
allocation of 1 PRB per TTI.
ATB during DTX/DRX

If no Power headroom indications are received during the

previous reporting period, then ATB running average filter stays
with the previous settings (no change).
If no Power headroom indications are received during the
whole call at all, then ATB running average filter stays with the
initial static setup iniPrbsul.

iniPrbsul
Initial amount of PRBs in UL
LNCEL; 1100; 1; 10
Extended UL Link Adaptation (E-
(E-ULA)
Uplink Link Adaptation entities

The purpose UL LA is to improve system capacity, peak data rate and coverage reliability by the
adaptation of of transmission settings to the radio channel conditions

UL Adaptive Modulation and Coding (UL AMC) which selects appropriate ILLA
MCS for UL transmission taking actual transmission reliability (BLER). UL-AMC OLLA

is split into: ATB

Inner Loop Link Adaptation (ILLA) slow periodic AMC

Periodic ACK/NACK information is used for calculating BLER (Block Error Rate) after 1st
transmission or nth retransmission
Outer Loop Link Adaptation (OLLA) event-triggered aperiodic AMC
Periodic ACK/NACK information is used for calculating BLER after 1st transmission of a Sync OLLA
Transport Block in order to derive a compensation factor
Slow
Adaptive Transmission Bandwidth (ATB) ATB

responsible for defining maximum number of PRBs that can be assigned

to a particular UE by UL SCH
Uplink Link Adaptation history

New Uplink Link Adaptation concept has been introduced in RL30

RL10 UL LA RL30 E-ULA

OLLA OLLA
Remains unchanged
ILLA ILLA
Not used when E-ULA is
active
SlowATB New PHR and BLER
PHR based based ATB algorithm
OLLA and ATB
synchronization algorithm
Extended-Uplink Link Adaptation motivation

It is more efficient to distribute the power over a wider bandwidth (more

PRBs) using lower MCS

If a UE is power limited (corresponding to bad RF conditions)

This fact is due to Shannons formula for the channel capacity of a bandwidth
and power limited channel.

S
C = B w log 2 1 +
N
More efficient
Wider bandwidth Less efficient
(more PRBs)
Few PRBs
Lower MCS
Higher MCS
RL30 E-ULA concept
With LTE1034 the 3 processes (UL AMC, UL ATB and UL OLLA) that rule the
UL Link Adaptation, work synchronized but independently to each other.

Eliminate any possibility of BLER target drifting by:

stopping the SLOW AMC algorithm (ILLA)
leaving the MCS regulation the OLLA algorithm Therefore OLLA algorithm is unchanged
and become the only one ruling the MCS
OLLA reacts relatively fast when it comes to reduce index up and down
MCS index and slowly enough when it comes to
upgrade MCS index Sl
ow
AT
B
The main OLLA
idea
AMC

ATB is no longer PHR based but BLER
based (with PHR correction).
It will become active only when the OLLA has
already reached the lower possible limit for
the MCSindex
Most of all SlowATB is coordinated with
OLLA.
This means that SlowATB acts only when
OLLA has no longer margin left in term of
reaction.
E-ULA algorithm overview
START

OLLA verifies BLER conditions and triggers FUG or

OLLA
EDG events when necessary as in former releases
Counter is incremented in every TTI when user is
actively scheduled
Increment ATB is triggered for:
ttiEventCounter
ttiEventCounter threshold for periodical ATB triggering
ATB
No (eUlLaAtbPeriod)
Triggering ? OR
Yes
EDG trigger - is sent by OLLA when EDG event happens
and the lowest MCS Index has been already reached.
Therefore EDG cannot further decrease this MCS index. In
this case OLLA triggers the earlier activation of the
SlowATB process.

Abbreviated Range Default

Name (Stepsize/Granularity)
Next
slide END eUlLaAtbPeriod {10, 15, 20, 30, 35, 40, 45, 50} 30
BLER based ATB routines enhanced in E-ULA
When UE being in TRIGGER When UE being in
bad RF goes to bad RF goes to
better RF conditions worst RF conditions
No Yes
BLER > blerTarget ?
When BLER is higher than the given
When BLER is lower than the given
target and OLLA has already set the
target and OLLA has already set the
MCS Index to the eUlLaLowMcsTh,
MCS Index to the
No No while MAX_NUM_PRB is still over the
eUlLaLowMcsTh+ eUlLaDeltaMcs
lowest PRB threshold
(value defined by the Operator).
(eUlLaLowPrbThr)
Yes
Yes

The number of PRBs is increased Amount of PRBs is decreased by

by factor eUlLaPrbIncDecFactor factor eUlLaPrbIncDecFactor

provide MAX_NUM_PRB and

END NewMCS to other functions

Abbreviated Name Range Default Remark

1
eUlLaLowPrbThr {1, 2, 3, 4, 5} This shall be always bigger than or equal to redBwMinRbUl
1
eUlLaLowMcsThr {1, 2, 3, 4} -
0.8
eUlLaPrbIncDecFactor {0.5, 0.6, 0.7, 0.75, 0.8, 0.85, 0.9} -
Defines how many MCS indexes above the minimum MCS index are required
eUlLaDeltaMcs {1, 2, 3, 4, 5, 6} 3
before ATB may increase the amount of allocable PRBs
Resetting the algorithm after long pause between active-
scheduled TTIs 1/3
It is necessary to define how to react to long pauses between TTIs where UE
is actively scheduled.

UL-AMC defines already the parameter/timer ulamcInactT for the purpose of resetting the MCS-index after
the expiration of this timer.
To avoid parameter multiplications those parameters are utilized with a similar function in E-ULA but in E-
ULA instead the algorithm acts as well on the number of allocable PRBs instead of the MCS Index only.

Start decreasing When PRBs

ulamcInactT
PRBs decreased to initial,
reset MCS

1 2 3 4 12 13 14 15
TTIs
User User actively
scheduled No transmission scheduled
- long pause again

Abbreviated Range(Stepsize/Granularity) Default Description

Name
100 ms
ulamcInactT 10 ms, 20 ms, ., 1000 ms / 10 ms Timer for Inactivity and DRX/DTX Periods
Resetting the algorithm after long pause between active-
scheduled TTIs 2/3
If user is still not active after expiration of timer, its resource assigments
should be gradually decreased

If meantime user was scheduled

No transmission long pause again, the counter is reset and
no further LA adjustment is
User is not Wait until counter ulamcInactT is reached done
scheduled

When counter reached and PRB > iniPrbsUl > start decreasing PRBs by
Start eUlLaPrbIncDecFactor
decreasing
PRBs

When PRBs already at iniPrbsUl set MCS to iniMcsUl

Reset MCS Reset counters and
re-initialize OLLA in
case of any
recalculation

Abbreviated Name Range(Stepsize/Granularity) Default

iniPrbsUl 1, 2, ..., 100 / 1 10

iniMcsUl MCS0, MCS2, , MCS20 MCS5

 Resetting the algorithm after long Start when user get
pause between active-scheduled TTIs START initial assignment
in E-ULA 3/3

Yes No
Reset counter User scheduled ? Increment counter

For every
Reset MCS to TTI Counter reached No
initail if PRBs ulamcInactT ? Decrease
already resetted PRBs if not
Yes at initial yet

No No
CurrentMCS > iniMcsUl MAX_NUM_PRB > iniPrbsUl

Yes Yes

MAX_NUM_PRB = max (
NewMCS = iniMcsUl iniPrbsUl,
(MAX_NUM_PRB * eUlLaPrbIncDecFactor))
We shouldnt
decrease
more, because
PRB and MCS
are already at
initial default
Re-initialize OLLA
Reset ttiEventCounter
Leave In case of re-
MAX_NUM_PRB
unchanged calculation

END
 Uplink Link Adaptation for PUSCH (RL30)

UE Power Headroom reporting

The UL AMC/ATB delivers to Telecom C-Plane the relevant configuration data for the
UE Power Headroom Configuration
The UE shall be configured to report PERIODIC PWR_HEADR_UL reports. For this the period shall be configured by
RRC - if applicable. The parameter tPeriodicPhr is available from O&M. Also the minimum time period in-between two
reports is available by O&M with tProhibitPhr. Event driven Parameter such as rapid change in path loss shall be set
according to dlPathlossChg. Also the latter data is taken from O&M parameter settings.

So the UE shall be configured as follows by:

event triggered reporting
DL_PATHLOSS_CHANGE = dlPathlossChg with {dB1, dB3, dB6, infinity}

periodical reporting
PERIODIC_PHR_TIMER = tPeriodicPhr with {sf10, sf20, sf50, sf100, sf200, sf500, sf1000, infinity}

report prohibit timer

PROHIBIT_PHR_TIMER = tProhibitPhr with {sf0, sf10, sf20, sf50, sf100, sf200, sf500, sf1000}

If the periodic PHR Timer is set to infinity and the DL path loss change, too, then no
PWR_HEADR_UL indications are received during the whole call. Then in this case the
ATB stays with the initial static setup ULATB_INIPRBs.
E-ULA activation

Parameter actUlLnkAdp activates Link Adaptation and defines its mode

ATB
actUlLnkAdp ILLA OLLA
PHR based BLER based

off

eUlLa

slowAmc

slowAmcATB

slowAmcOlla

slowAmcOllaATB
Controlled uplink packet segmentation
Controlled UL Packet Segmentation

UL coverage improvement algorithm as an extension to the UL LA.

PDCCH is better utilized by balancing new transmissions and retransmissions.
The enhancement in coverage throughput comes at the cost of cell throughput.
For a given modulation code scheme (MCS) as determined by the uplink link adaptation and
configured transport block size (TBS), the uplink scheduler determines the PRB allocation.

ulsMinTbs
Defines the minimum UL TBS
(segment size).
LNCEL, 16... 1544, step 8, (bit)
104

ulsMinRbPerUe
Minimum PRB allocation for
UEs which are power
Limited.
LNCEL; 1100, step 1, (1)
3
THANK YOU