eRAN8.1 Deep Dive - ACH Feature I Ntroduction: Huawei Technologies Co., LTD
eRAN8.1 Deep Dive - ACH Feature I Ntroduction: Huawei Technologies Co., LTD
eRAN8.1 Deep Dive - ACH Feature I Ntroduction: Huawei Technologies Co., LTD
2. Solution Introduction
3. Impact Analysis
4. Activation Guide
5. Verification
HUAWEI TECHNOLOGIES
HISILICON CO., LTD.
SEMICONDUCTOR Page 2
ACH (Automatic Congestion Handling)
Scenarios:
Planned Event(Periodic busy hour, New year Eve, Sports meeting, football match, concert,)
Unplanned Event (congregation, parade, carnival…)
Acceptance
Value:
Improve network capacity by seeking for new balance between system reliability and capacity.
Save much manpower for guarantee activities in big event , and improve the efficiency.
HUAWEI TECHNOLOGIES
HISILICON CO., LTD.
SEMICONDUCTOR Page 3
Contents
1. Background
2. Solution Introduction
3. Impact Analysis
4. Activation Guide
5. Verification
HUAWEI TECHNOLOGIES
HISILICON CO., LTD.
SEMICONDUCTOR Page 4
Architecture of ACH
• Bottleneck:
Resource Utilization
Radio resource will be the
bottleneck before the
system hardware resource PDCCH CCE UL/DL PRB
User Number
• Trigger :
The User Number is the root cause of
Traffic Mode User Distribution network loading, including the traffic
mode and user distribution.
Actions List in ACH for Congestion
Category Actions Theory
In high loaded network scenarios, reduce the resource overhead for TA measurement based on
TA Optimization DMRS, and enlarger the period of the TA adjust command, consequently, reduce the PDCCH CCE
and UL/DL PRB overhead.
Detect the abnormal UEs: stop the UL scheduling according to continuous 16 PUSCH CRC errors,
Overhead Reducing stop the DL scheduling according to continuous 16 UL DTX; And recover the UL scheduling when
Abnormal UEs Detection
receiving SR from UE, recover the DL scheduling when PUSCH CRC OK or PUCCH CQI response is
normal.
Switch off DL Frequency Switch off DL FSS to reduce the mass aperiodic al CQI report, and then decrease the resource
Selection Scheduling(FSS) overhead including CCE and UL PRB, and help to control the UL interference.
Try to schedule all of the data for user in one TTI even a little more PRB resource is allocated. For
Adjust the RBG Allocate example, if the UE require 1.5 RBG to schedule the data, then eNodeB assign 2 RBGs to this UE,
Strategy instead of scheduling the user in two TTI s and consume more PDCCH CCE and scheduling
Efficiency
changes. Then improve the PDCCH CCE efficiency and the DL PRB utilization rate.
Improvement
When a user require CCE resource with a high CCE aggregation level, the PDCCH CCE fragment
PDCCH Capacity Improve can't be used. then CCE aggregation level degraded with power boosting, will help to improve the
CCE utilization and schedule more users.
By default, PUSCH PRBs are allocated from high to low in RB index for all cells. In high loaded
UL PRB Allocate
Mitigate Interference network scenario, PUSCH PRBs are allocated at random position in the whole PUSCH bandwidth, in
Randomization
order to reduce the interference from neighbor cells.
MOD
User_Spec_Occupy_Rate > 50% User_Spec_Occupy_Rate < 40%
Overhead Abnormal UEs CELLALGOSWITCH:
And or
Reducing Detecting CellSchStrategySwitch=AbnUeSchS
PDCCH_CCE_Utilization_Rate > 60% PDCCH_CCE_Utilization_Rate < 25%
witch-1
MOD
UL PRB Allocate User_Spec_Occupy_Rate > 50% User_Spec_Occupy_Rate < 40%
Mitigate CELLULSCHALGO:ULRBALLOCAT
Strategy And or
Interference IONSTRATEGY=FS_INRANDOM_A
Optimization UL_PRB_Utilization_Rate > 20% UL_PRB_Utilization_Rate < 10%
DAPTIVE
Note: Rollback is to recover the original configured value no matter what was configured.
Page 8 HUAWEI TECHNOLOGIES CO., LTD.
Overhead Reducing - TA Optimization
• Issue:
TA command and TA measurement cost much resource, including PDCCH CCE and DL/UL PRB resource:
The default TATimer value is short, which bring too many TA commands in big event.
Too many UL PRB are scheduled just for measuring TA through DMRS.
• Solution
Enlarge the TATimer value (Infinity), then reduce the frequency of TA command.
Reduce the UL PRB used for measuring TA.
In light traffic load, the eNodeB have sufficient channel resources to maintain uplink timing for UEs, ensuring TA accuracy
and service quality. If network congestion occurs, reducing the uplink scheduling resources used for TA measurement thr
ough DMRS to reduce UL PRB.
Category Action Action Condition Parameters Rollback Condition
Here, User_Spec_Occupy_Rate : Cell User Number divided by Cell user number specification;
The condition threshold can be modified by MOD LIOptAtomRule command.
Page 9 HUAWEI TECHNOLOGIES CO., LTD.
Overhead Reducing - Abnormal UEs Detecting
• Abnormal UEs Detecting:
In the uplink, if the eNodeB detects 16 consecutive uplink CRC errors on a UE, the eNodeB stops uplink sche
duling for the UE and resumes uplink scheduling when it receives a scheduling request (SR) from the UE.
In the downlink, if the eNodeB detects 16 consecutive uplink discontinuous transmissions (DTXs) on a UE, th
e eNodeB stops downlink scheduling for the UE and resumes it when the uplink CRC result is correct or whe
n periodic CQI reporting becomes normal on the PUCCH.
• Gain:
eNodeB can schedule more services in heavy traffic scenarios, improving cell capacity.
MOD
User_Spec_Occupy_Rate > 50% User_Spec_Occupy_Rate < 40%
Overhead Abnormal UEs CELLALGOSWITCH:
And or
Reducing Detecting CellSchStrategySwitch=AbnUeSch
PDCCH_CCE_Utilization_Rate > 60% PDCCH_CCE_Utilization_Rate < 25%
Switch-1
• Downlink frequency selective scheduling triggers large-scale aperiodic CQI reporting, consuming
many uplink PRBs and CCEs and increasing uplink interference. In event of network congestion,
disabling downlink frequency selective scheduling helps reduce the interference and increase the
uplink cell capacity. However, disabling downlink frequency selective scheduling slightly decrease
s downlink cell throughput.
10MHz
1RBG=3RB
MOD
UL PRB Allocate User_Spec_Occupy_Rate > 50% User_Spec_Occupy_Rate < 40%
Mitigate CELLULSCHALGO:ULRBALLOCAT
Strategy And or
Interference IONSTRATEGY=FS_INRANDOM_A
Optimization UL_PRB_Utilization_Rate > 20% UL_PRB_Utilization_Rate < 10%
DAPTIVE
CCE agg.
No available
+
Initial Value (for Set close-loop adjustment initial value to -10 for Set close-loop adjustment initial value to -10 for Set close-loop adjustment initial
parameter) SRB and DRB SRB only. value to 0 for DRB.
8
4 Initial value:0
Initial value:-10
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
MCS
2. Solution Introduction
3. Impact Analysis
4. Activation Guide
5. Verification
Negative Impact
“Abnormal UEs Detecting” strategy may increase the call drop probability a little.
2. Solution Introduction
3. Impact Analysis
4. Activation Guide
5. Verification
2. Solution Introduction
3. Impact Analysis
4. Activation Guide
5. Verification
1526728998 L.Thrp.Time.Cell.UL.HighPrecision Total duration of downlink data transmission in a cell (with the precision of 1 ms)
1526729005 L.Thrp.bits.DL.LastTTI Downlink traffic volume sent in the last TTI for PDCP SDUs before the buffer is empty
1526729015 L.Thrp.Time.DL.RmvLastTTI Data transmit duration except the last TTI before the downlink buffer is empty
1526729415 L.Thrp.bits.UE.UL.SmallPkt Uplink traffic volume of PDCP PDUs scheduled for small packets
1526729416 L.Thrp.Time.UE.UL.RmvSmallPkt Uplink data transmission duration except that for small packets
1526726659 L.RRC.ConnReq.Succ Number of RRC Connection Setup Complete messages received from UEs in a cell
1526726658 L.RRC.ConnReq.Att Number of RRC connection setup attempts performed by UEs in a cell (retransmission excluded)
1526728269 L.RRC.SetupFail.Rej Number of RRC Connection Reject messages sent to UEs in a cell
1526728490 L.RRC.SetupFail.Rej.FlowCtrl Number of times the eNodeB sends an RRC Connection Reject message to the