Improving Spectrum Efficiency with ACKs Jiansong Zhang # , Haichen - - PowerPoint PPT Presentation

▶
improving spectrum efficiency with
SMART_READER_LITE
LIVE PREVIEW

Improving Spectrum Efficiency with ACKs Jiansong Zhang # , Haichen - - PowerPoint PPT Presentation

Jan 30, 2023 354 likes •771 views

Improving Spectrum Efficiency with ACKs Jiansong Zhang # , Haichen Shen , Kun Tan , Ranveer Chandra * , Yongguang Zhang and Qian Zhang # Microsoft Research Asia * Microsoft Research Redmond # HKUST Feedback in Wireless

slide-1
SLIDE 1

Improving Spectrum Efficiency with μACKs

Jiansong Zhang†#, Haichen Shen†, Kun Tan†, Ranveer Chandra*, Yongguang Zhang† and Qian Zhang#

†Microsoft Research Asia *Microsoft Research Redmond #HKUST

slide-2
SLIDE 2

Feedback in Wireless Networks

  • Feedback is critical for network protocols

2

t DATA ACK

 Confirm reception / detect loss (i.e. ACKs)

  • Current network protocols are primarily based
  • n frame level feedback
slide-3
SLIDE 3

Frame-level Feedback Considered Harmful in Wireless

  • May be too late

 Feedback received after all damage has been done

3

t

𝑼𝟐 𝑼𝟑

ACK Timeout

Example 1: Collision detection based on ACK

slide-4
SLIDE 4

Frame-level Feedback Considered Harmful in Wireless

  • May contain limited information

4

Example 2: Frame retransmission is inefficient

Medium Access Preamble & Header Data ACK

DIFS SIFS

Retransmission: 𝑺𝒇𝒆𝒗𝒐𝒆𝒃𝒐𝒅𝒛

slide-5
SLIDE 5

Frame-level Feedback Considered Harmful in Wireless

5 Medium Access Preamble & Header Data ACK

DIFS SIFS

Retransmission: 𝑫𝒑𝒐𝒖𝒇𝒐𝒖𝒋𝒑𝒐 𝑰𝒇𝒃𝒆𝒇𝒔𝒕

  • May be costly to re-establish transmission context
  • May contain limited information

Example 2: Frame retransmission is inefficient

slide-6
SLIDE 6

We should do symbol level feedback

6

slide-7
SLIDE 7

µACK Towards Symbol-level Feedback

7

t Data Frame

uACK uACK … uACK uACK

f

  • Two Tightly synchronized radio chains

 Wide-band forward channel  Narrow-band feedback channel

  • Tiny acknowledgement symbols
slide-8
SLIDE 8

µACK Application 1 – Collision Detection and Early Backoff

  • Early collision detection by feedback timeout

8

Preamble Few symbols

Feedback Timeout

Collision

slide-9
SLIDE 9

µACK Application 2 – Hidden & Exposed Terminal Mitigation

9

𝑼 𝑺 𝑰

𝜈𝐵𝐷𝐿 from R prevents H from colliding

Hidden Terminal:

slide-10
SLIDE 10

10

𝑺𝟐 𝑼 𝑺𝟑 Exposed Terminal:

  • 𝜈𝐵𝐷𝐿 is an extended busy tone

𝑭

𝐹 can detect it is under exposure

µACK Application 2 – Hidden & Exposed Terminal Mitigation

slide-11
SLIDE 11

µACK Application 3 – In Frame Retransmission

  • Retransmission appends to original frame

11

t

Preamble GOS 1 GOS 2 GOS 3 GOS 4 GOS 2 Preamble

uACK uNACK uACK uACK EOS GOS: group of symbols EOS: end of stream

slide-12
SLIDE 12

µACK Benefits Wireless in Various Ways

  • Application 1:

 Collision Detection and Early Backoff

  • Application 2 (extended):

 Hidden & Exposed Terminal Mitigation

  • Application 3:

 In-frame Retransmission

12

slide-13
SLIDE 13

13

µACK Benefits Wireless in Various Ways

  • Application 1:

 Collision Detection and Early Backoff

  • Application 2 (extended):

 Hidden & Exposed Terminal Mitigation

  • Application 3:

 In-frame Retransmission

slide-14
SLIDE 14

In-frame Retransmission Details

  • Design questions

 What is the symbol group size?  What is 𝜈𝐵𝐷𝐿 physical layer?  How to determines a group of symbol is correct?

14

t

Preamble GOS 1 GOS 2 GOS 3 GOS 4 GOS 2 Preamble

uACK uNACK uACK uACK EOS GOS: group of symbols EOS: end of stream

slide-15
SLIDE 15

Data Symbol Group Size

  • Symbols in a group are fate-sharing

 GOS length < coherent time of the channel

  • Tradeoff between redundant bits and feedback

channel requirement

 Larger GOS  more redundant bits, and less feedback bandwidth

  • Design choice

 20𝜈𝑡 GOS  5 OFDM symbols  1MHz feedback channel ~ 5% for 20MHz data channel

15

slide-16
SLIDE 16

µACK PHY

  • Simple spectrum spreading PHY

 Feedback symbol time is 20𝜈𝑡 (the length of GOS)  Four bits per symbol (encode 3 states)  Channel width is 1MHz (50% guard band)  Bandwidth 500KHz  Chip rate is 500Kcps  Ten chips per symbol

16

slide-17
SLIDE 17

Error Detection

  • Two methods

 Segment CRC (additional overhead)  PHY hints

17

We found PHY hints becomes less reliable in some cases …

slide-18
SLIDE 18

PHY hints become unreliable on marginal SNR

24Mbps, 10dB (marginal) 24Mbps, 12dB (higher)

18

slide-19
SLIDE 19

PHY hints become unreliable on marginal SNR

24Mbps, 10dB (marginal) 24Mbps, 12dB (higher)

19

slide-20
SLIDE 20

PHY hints become unreliable on marginal SNR

24Mbps, 10dB (marginal) 24Mbps, 12dB (higher)

20

slide-21
SLIDE 21

PHY hints become unreliable on marginal SNR

24Mbps, 10dB (marginal)

21

False negative 24Mbps, 12dB (higher)

slide-22
SLIDE 22

PHY hints become unreliable on marginal SNR

24Mbps, 10dB (marginal)

22

False negative False positive 24Mbps, 12dB (higher)

We explicitly embed CRC in each GOS

slide-23
SLIDE 23

Segment CRCs add additional overhead

23

Can we avoid the overhead?

slide-24
SLIDE 24

Pilot Side-Channel

24

Dummy-bit Pilots

  • Encode information in

the pilots

 Embed 16 bits in a GOS  Hamming (16, 11) code  CRC-10

slide-25
SLIDE 25

Pilot Side-Channel

  • How?

 Differential BPSK (similar to 802.11b)

25

I Q 𝑬𝒗𝒏𝒏𝒛𝒄𝒋𝒖 = (𝟐, 𝟏)

Example:

Symbol Encoded (I, Q) 𝑻𝟏 (𝟐, 𝟏)

slide-26
SLIDE 26

Pilot Side-Channel

26

I Q 𝑻𝟐 = (𝟐, 𝟏)

Example:

Symbol Encoded (I, Q) 𝑻𝟏 (𝟐, 𝟏) 𝑻𝟐 𝟏 (𝟐, 𝟏)

  • How?

 Differential BPSK (similar to 802.11b)

slide-27
SLIDE 27

Pilot Side-Channel

27

I Q 𝑻𝟑 = (−𝟐, 𝟏)

Example:

Symbol Encoded (I, Q) 𝑻𝟏 (𝟐, 𝟏) 𝑻𝟐 𝟏 (𝟐, 𝟏) 𝑻𝟑 𝟐 (−𝟐, 𝟏)

  • How?

 Differential BPSK (similar to 802.11b)

slide-28
SLIDE 28

Pilot Side-Channel

28

I Q 𝑻𝟒 = (−𝟐, 𝟏)

Example:

Symbol Encoded (I, Q) 𝑻𝟏 (𝟐, 𝟏) 𝑻𝟐 𝟏 (𝟐, 𝟏) 𝑻𝟑 𝟐 (−𝟐, 𝟏) 𝑻𝟒 𝟏 (−𝟐, 𝟏)

  • How?

 Differential BPSK (similar to 802.11b)

slide-29
SLIDE 29

Pilot Side-Channel

29

I Q

Example:

Symbol Encoded (I, Q) 𝑻𝟏 (𝟐, 𝟏) 𝑻𝟐 𝟏 (𝟐, 𝟏) 𝑻𝟑 𝟐 (−𝟐, 𝟏) 𝑻𝟒 𝟏 (−𝟐, 𝟏) 𝑻𝟓 𝟐 (𝟐, 𝟏)

𝑻𝟓 = (𝟐, 𝟏)

  • How?

 Differential BPSK (similar to 802.11b)

slide-30
SLIDE 30

Pilot Side-Channel

30

I Q

Example:

Symbol Encoded (I, Q) 𝑻𝟏 (𝟐, 𝟏) 𝑻𝟐 𝟏 (𝟐, 𝟏) 𝑻𝟑 𝟐 (−𝟐, 𝟏) 𝑻𝟒 𝟏 (−𝟐, 𝟏) 𝑻𝟓 𝟐 (𝟐, 𝟏) … … …

𝑻𝟓 = (𝟐, 𝟏)

  • How?

 Differential BPSK (similar to 802.11b)

slide-31
SLIDE 31

Decision Directed Pilot Tracking

  • Pilots should be decoded first before used for

channel tracking

 No performance loss if pilots are correctly decoded  No performance loss even if pilots are not correctly decoded

  • Normal pilots are inserted at beginning of an GOS

 Pilot decision error will not propagate to next GOS

31

slide-32
SLIDE 32

Sora Based Implementation

  • Extend Sora

 Multi-radio board  Direct symbol transmission to radio

32

slide-33
SLIDE 33

Performance Evaluation

  • Is µACK feasible?

 Micro-benchmarks

  • What is the benefit of µACK?

 Wired single link  9 node real network

33

slide-34
SLIDE 34

End-to-end Latency of μACK

Viterbi Decoding µACK modulation Hardware 7.5µs 1.96µs 9.103µs

34

Breakdown:

17.5µs

slide-35
SLIDE 35

μACK PHY Performance

35

  • µACK vs. 802.11 6Mbps
slide-36
SLIDE 36

DDPT Performance

36

  • DDPT vs. Normal
slide-37
SLIDE 37

μACK on Wired Single Link

  • 𝜈𝐵𝐷𝐿 sender aggressively use higher data rates.
  • Up to 220% over 802.11a, up to 30% over PPR

37

slide-38
SLIDE 38

Trace-based Emulation

38

Throughput Latency

slide-39
SLIDE 39

Related Work

39

  • Hybrid ARQs

 Complementary to 𝜈𝐵𝐷𝐿

  • Partial Packet Recovery
  • CSMA/CN
  • Rate adaptation

 𝜈𝐵𝐷𝐿 shows by reducing loss recovery overhead, one can use more aggressive rates  𝜈𝐵𝐷𝐿 also enables in-frame rate adaptation

  • Busy-tone schemes (DBTMA)

 𝜈𝐵𝐷𝐿 can serve as an extended busy tone

slide-40
SLIDE 40

Conclusion

40

  • 𝜈𝐵𝐷𝐿 enables sending fine-grained feedback

 Collision detection  Mitigation of hidden & exposed terminal problem  In-frame loss recovery

  • 𝜈𝐵𝐷𝐿 is feasible & significantly improves

spectrum efficiency

 Reduces retransmission overhead  Increases transmission rate  Improves collision management