Drop HARQ feedback for aggressive HARQ transmission

IEEE 802.16 Presentation Submission Template (Rev. 9) Document Number:

IEEE S802.16m-09/0048Date Submitted:2009-1-5

Source: Zheng Yan-Xiu, Yu-Chuan Fang, Chang-Lan Tsai, Chung-Lien Ho, Hsi-Min Hsiao, Ren-Jr Chen, Richard Li, E-mail: zhengyanxiu@itri.org.tw. ITRI

Venue: IEEE Session #59, San Diego.

Base Contribution: N/A Re: 802.16m-08/052, Call for Comments on 802.16m SDD (802.16m-08/003r6), Section 11.13.2.6 HARQ Feedback Purpose: To be discussed and approval by IEEE 802.16m TG

Notice:This document does not represent the agreed views of the IEEE 802.16 Working Group or any of its subgroups. It represents only the views of the participants listed in the “Source(s)” field above. It is offered as a basis for discussion. It is not binding on the contributor(s), who reserve(s) the right to add, amend or withdraw material contained herein.

Release:The contributor grants a free, irrevocable license to the IEEE to incorporate material contained in this contribution, and any modifications thereof, in the creation of an IEEE Standards publication; to copyright in the IEEE’s name any IEEE Standards publication even though it may include portions of this contribution; and at the IEEE’s sole discretion to permit others to reproduce in whole or in part the resulting IEEE Standards publication. The contributor also acknowledges and accepts that

this contribution may be made public by IEEE 802.16.

Patent Policy:The contributor is familiar with the IEEE-SA Patent Policy and Procedures:

<http://standards.ieee.org/guides/bylaws/sect6-7.html#6> and <http://standards.ieee.org/guides/opman/sect6.html#6.3>.Further information is located at <http://standards.ieee.org/board/pat/pat-material.html> and <http://standards.ieee.org/board/pat >.

Aggressive HARQ Transmission• Aggressive HARQ transmission removes HARQ soft buffer

constraint to achieve higher MS transmission rate– Less HARQ soft buffer implies lower MS complexity– Higher transmission rate improves user experience

• However, aggressive HARQ transmission introduces buffer overflow– Received samples are stored in the HARQ buffer if they

are not decodable– Buffer overflow occurs when MS reception status becomes

bad temporarily due to channel variation and erroneous RSSI report, e.g. deep fade, shadowing, channel estimation error, etc• And then, packets drop

– Packet dropping results in unreliable HARQ transmission

Packet Dropping Rate Analysis

• 4 HARQ buffers are assumed• 4 ~ 24 HARQ processes are analyzed

– Multi-carrier may support up to 32 HARQ processes

• Packet error rate=0.01, 0.1, 0.2, 0.3, 0.5• As packet error rate is high, the packet

dropping becomes significant• The error rate also increases with number

of HARQ processes

Error rate= 0.01

Error rate= 0.1

Error rate= 0.2

Error rate= 0.3

Error rate= 0.5

HARQ Processes=4

0 0 0 0 0

HARQ Processes=8

6.9e-10 5.7e-05 0.0015 0.0088 0.0684

HARQ Processes=12

6.3e-09 4.1e-04 0.0080 0.0370 0.1746

HARQ Processes=16

2.5e-08 0.0013 0.0199 0.0734 0.2508

HARQ Processes=20

7.0e-08 0.0029 0.0349 0.1076 0.3001

HARQ Processes=24

1.6e-07 0.0051 0.0507 0.1357 0.3333

The Bit Error Rate Performance of Lost of Packet in the First Transmission

• Observation: disaster occurs if the first transmission is lost– Turbo decoder can not

correctly decode packet if complete systematic part is not received

– If HARQ-IR is used and the first packet is dropped, the next redundancy version (RV1) can not carry complete systematic part and the redundancy version is not decodable

An Example for Aggressive HARQ Transmission

• 4 HARQ soft buffers• A/X denotes ACK with X

buffer used• Red N/X denotes NACK with

X buffer used• Blue N/X denotes NACK

with drop and X buffer used• 12 HARQ processes• 17 Packets pass

– At least 5 more packets compared to conventional HARQ scheduling

• 9 Packets are dropped – Dropped packets may

deteriorate reception performance and cause large latency

– It may further worsen HARQ-IR performance

Solution: HARQ Buffer Management for Aggressive HARQ Transmission

• MS-assisted HARQ flow control– Avoid HARQ failure due to buffer overflow

• HARQ buffer synchronization– Acknowledge BS MS buffer state by an extra HARQ

feedback – Clean up the obsolete HARQ soft buffer while necessary

(C802.16m-08/1328)

MS-assisted HARQ Flow Control

• Concept: MS feedbacks current buffer status to BS to facilitate HARQ scheduling– Drop feedback (DROP): inform BS the overflow of MS

HARQ soft buffer– DROP is only applied for aggressive HARQ transmission

• Normal HARQ transmission only applies 1-bit ACK/NACK feedback

• BS schedules aggressive HARQ transmission based on feedback– Retransmit if NACK is received– Reinitiate a dedicated HARQ process if DROP is received

• MS sends DROP feedback to notify BS of overflow dropping

Example Revisited with DROP Feedback

• 4 HARQ soft buffers• 12 HARQ processes• A/X denotes ACK with X buffer used• Red N/X denotes NACK with X buffer used• Blue D/X denotes DROP with X buffer used• DROP signal can reinitiate HARQ

transmission to avoid transmission failure due to the 2nd transmission

• 22 Packets pass – At least 10 more packets transmitted

compared to conventional HARQ scheduling– 5 packets transmitted compared to aggressive

HARQ transmission • 4 Packets are dropped

– Less packets are dropped– It maintains HARQ-IR performance– HARQ reliability can be maintained for

aggressive HARQ transmission

Performance Evaluation for HARQ-CCThroughput and Block Error Rate

• Round Trip Delay=10 ms • MCS selection is based on

predetermined SNR• No feedback error• When 10K soft bits are used in soft

buffer, – drop feedback does not throughput

gain– drop feedback maintains HARQ fail

rate at 1%-0.1% instead of more than 0.1 HARQ fail rate

• Conventional scheduling can only provide maximum throughput 500Kbps with 10K soft bits

• We push throughput to 2X~3X• If MCS selection can be better, more

throughput and less HARQ fail rate are achievable

Performance Evaluation for HARQ-IRThroughput and Block Error Rate

• Round Trip Delay=10 ms • MCS selection is based on predetermined

SNR• No feedback error• When 10K soft bits are used in soft

buffer, – drop feedback drives extra 33%

throughput gain from retransmission of the first packet

– drop feedback maintains HARQ fail rate at 1% instead of more than 0.1 HARQ fail rate

• Conventional scheduling can only provide maximum throughput 500Kbps with 10K soft bits

• We push throughput to 2X~3X• If MCS selection can be better, more

throughput and less HARQ fail rate are achievable

Example of Feedback Channel Design

• Conventional bi-state feedback: 12 orthogonal sequences for 6 for ACK and 6 for NACK

• Tri-state HARQ feedback: – Method: 12 orthogonal sequences for 4 ACK and 4 NACK and 4 DROP– Pros:

• Compatible with conventional design– Cons:

• Increased overhead• Nice-State two channel HARQ feedback:

– Method: two feedback channel joint coding• Choose four sequence from 12 orthogonal sequences • Construct nine sequences to represent two feedback

channels– Pros:

• Compatible with conventional design• Similar error rate performance

– Cons:• Even feedback channels

41

9/2 ][exp][i i

ikjk pSpC

Error Rate Performance for Tri-State Feedback

• The proposed design provides similar performance comparing to 1-bit ACK/NACK

Channel Bandwidth 10MHz

Over-sampling Factor 28/25

FFT Size 1024

Cyclic prefix (CP) ratio 1/8

Channel condition PB3, VA120, VA350

The number of antennas Tx:1, Rx:2

Modulation BPSK

FMT size 6x2

Block size 6x6

Receiver HARQCH: non-coherent detection, MLD

Overhead issue

• Tri-state feedback channel is only applied for aggressive HARQ transmission mode, e.g. long burst service– Less resource shared by other users– More concurrent HARQ feedback channels

• Bi-state feedback channel is applied for normal HARQ transmission mode, e.g. VoIP

• Configuration example: 4 LRUs is configured for UL feedback channels– Tri-State HARQ feedback :

• 24 tri-state HARQ feedbacks (2 LRU)+36 1-bit HARQ feedbacks (2LRU)• Transmitting large packet ( 50Kbits) with limited overhead is reasonable≧

– Nice-State two channel HARQ feedbacks: • 36 tri-state HARQ feedbacks (2LRU)+36 1-bit HARQ feedbacks (2LRU)• Transmitting more HARQ processes to increase throughput is reasonable

An Example of Overhead Calculation

• 96 LRU per subframe @ 20MHz• 100Mbps @ Downlink• 500Kbps per 5ms frame

– 16 concurrent HARQ processes per 5ms• 31250 bps per HARQ process• 16 HARQ feedbacks per 5ms

– 16 Tri-State feedback occupies 4/3 LRU => 2LRU = 2.08% per UL subframe– 8 Nice-State feedback occupies 8/9 LRU => 1LRU = 1.04% per UL subframe– Transmission (bits)/HARQ overhead (LRU) > 250K bits/LRU (high efficiency)

– 8 concurrent HARQ processes per 5ms• 62500 bps per HARQ process• 8 HARQ feedbacks per 5ms

– 8 Tri-State feedback occupies 2/3 LRU => 1LRU = 1.04% per UL subframe– 4 Nice-State feedback occupies 4/9 LRU => 1LRU = 1.04% per UL subframe– Transmission (bits)/HARQ overhead (LRU) > 500K bits/ LRU (high efficiency)

• VoIP: – 26Kbps per user and 18 concurrent users– 520bits per 20ms frame– 18 concurrent users share one LRU for HARQ feedback (Intel, Samsung, LGe,…etc)– Transmission (bits)/HARQ overhead (LRU)= 9.36K bits/LRU (Low efficiency)

• Feedback overhead is very minor for high throughput transmission

Conclusions

• Drop HARQ feedback alleviates buffer overflow issue due to lack of soft buffer– Drop feedback is introduced for HARQ flow control in case of buffer overflow – BS can reinitiate HARQ process to ensure MS receiving maximum 4

transmissions if MS discards HARQ packets– Drop HARQ feedback maintains link reliability for aggressive HARQ

transmission– Drop HARQ feedback further drives extra throughput gain for HARQ-IR

• Drop HARQ feedback only introduces less HARQ feedback overhead for high throughput scenario

• Two exemplary 1-bit feedback compatible HARQ feedback designs are introduced– Two channel joint coding further provide similar error rate performance to 1-

bit feedback design

Proposed text

11.13.2.7 Aggressive HARQ Transmission

16m BS can transmit coded bits more 16m MS soft buffer capability.

DROP feedback is used to notify BS the MS buffer overflow when aggressive HARQ transmission is used. BS can restart the HARQ process to maintain identical link error rate performance to the normal HARQ transmission.

Backup Slides

Example of HARQ Flow Chart

• When most HARQ processes are incorrect received concurrently, buffer outage still occurs.

• When the case occurs, dropping HARQ feedback indicates the process is dropped to assist BS to reschedule

• Extra Advantages:– The successful transmission would

not be dropped– The failed transmission could be

safely reinitiated– ARQ-introduced latency is further

reduced– Dropping signaling can further

reinitiate HARQ process if DL allocation signal might be missed by MS, or MS ACK/NACK signal is missed

HARQ Feedback Errors and Response

• ACK => DROP : BS/MS restarts the process in the next transmission (no significant influence)

• ACK => NACK : send ACK at the next transmission (same as conventional HARQ feedback)

• NACK => DROP : BS/MS restarts the process in the next transmission (no significant influence)

• NACK => ACK : apply ARQ mechanism to restore the lost packet (same as conventional HARQ feedback)

• DROP => ACK : apply ARQ mechanism to restore the lost packet (same as conventional HARQ feedback)

• DROP => NACK : Send DROP in the next transmission to reinitiate the process (no significant influence)