TGac Preamble Auto-detection Comparisons Month Year Month Year doc.: IEEE 802.11-yy/xxxxr0 doc.: IEEE 802.11-yy/xxxxr0 TGac Preamble Auto-detection Comparisons Date: 2010-05-17 Authors: Slide 1 Hongyuan Zhang et al. Page 1 John Doe, Some Company John Doe, Some Company

Month Year doc.: IEEE 802.11-yy/xxxxr0 Background Motion passes on the preamble structure in spec framework as: First VHTSIG symbol modulated with BPSK 2nd VHTSIG symbol modulation TBD Two preamble designs were proposed: Scheme 1: 10-70r5 [1] Scheme 2: 10-359r0 [2] Difference of the two schemes are mainly on the 2nd VHTSIG symbol: [2] assumes existing 11n devices that use both HT-SIG symbols to do HT detection, and claimed that the proposed scheme will reduce the chance of 11n false detection at those existing devices. Hongyuan Zhang et al. John Doe, Some Company

Scheme 1 T Rate=6Mbps Length determined by T 2 symbols 1 symbol L-STF L-LTF L-SIG VHTSIGA VHT-STF VHT-LTFs VHTSIGB VHTData T VHT auto-detection Hongyuan Zhang et al.

(+45/-45 deg alternative BPSK) Month Year Month Year doc.: IEEE 802.11-yy/xxxxr0 doc.: IEEE 802.11-yy/xxxxr0 Scheme 2-A L-SIG (BPSK) VHT-SIGA1 (BPSK) VHT-SIGA2 (+45/-45 deg alternative BPSK) Slide 4 Hongyuan Zhang et al. Il-Gu Lee et al. Page 4 Il-Gu Lee, ETRI John Doe, Some Company

(Alternative Q-BPSK/BPSK) Month Year Month Year doc.: IEEE 802.11-yy/xxxxr0 doc.: IEEE 802.11-yy/xxxxr0 Scheme 2-B L-SIG (BPSK) VHT-SIGA1 (BPSK) VHT-SIGA2 (Alternative Q-BPSK/BPSK) Slide 5 Hongyuan Zhang et al. Il-Gu Lee et al. Page 5 Il-Gu Lee, ETRI John Doe, Some Company

(BPSK data & BPSK pilot) Month Year Month Year doc.: IEEE 802.11-yy/xxxxr0 doc.: IEEE 802.11-yy/xxxxr0 Scheme 2-C L-SIG (BPSK data & BPSK pilot) VHT-SIGA1 (BPSK data & BPSK pilot) VHT-SIGA2 (Alternative Q-BPSK/BPSK data & Q-BPSK pilots) Q-BPSK data BPSK pilot Q-BPSK pilot BPSK data Slide 6 Hongyuan Zhang et al. Il-Gu Lee et al. Page 6 Il-Gu Lee, ETRI John Doe, Some Company

About the 11n Auto-Det Assumption in [2] Month Year doc.: IEEE 802.11-yy/xxxxr0 About the 11n Auto-Det Assumption in [2] We believe it is not practical for 11n devices to perform HT detection using two HTSIG symbols for the following reasons: Reason 1: HT detection by HTSIG1 alone is already very reliable. It is equivalent to 48-repetition code of 1 bit across 48 data tones, versus rate ½ BCC encoding in the HTSIG bits. Therefore HT detection by 2 HTSIG symbols will not improve the overall performance—which is limited by other functions such as packet detection, timing, SIG field and Data field decoding. Reason 2: HT detection using HTSIG2 requires additional buffering of the samples in one whole OFDM symbol (i.e. the symbol at the position of HTSIG1), as well as delay of the equalizer and decoder processing  Not a valid design from engineering point of view. Hongyuan Zhang et al. John Doe, Some Company

Cont’d Reason 3: 11n devices has to rely on the detection over single HTSIG symbol, for HT Greenfield auto-detection  If it already “trusts” HT-GF detection by one symbol, why bother to implement a more “advanced” HT-MM detection over two symbols? Legacy BPSK …… L-STF L-LTF L-SIG L-Data L-Data HT-MM BPSK QBPSK QBPSK L-STF L-LTF L-SIG HTSIG1 HTSIG2 HT-STF HT-LTFs HT-Data HT-GF QBPSK QBPSK HTSTF HTLTF1 HTSIG1 HTSIG2 HT-LTF2, 3…. HT-Data Differentiate from HT-MM and Legacy packet by QBPSK in one OFDM symbol Hongyuan Zhang et al.

Issues with Scheme 2-C Rotate the pilot may impact the VHT detection or phase tracking performance: Need to do VHT detection before correcting the phase drift. Otherwise, correct the phase with two hypothesis. Q-BPSK data Q-BPSK pilot BPSK data Hongyuan Zhang et al.

Simulations on Scheme 2-B v.s. Scheme 1 Channel DNLOS 1x1 20MHz (detection in one sub-band) Three cases are compared: False VHT detection, when 11a packet is sent to 11ac receiver. Miss VHT detection, when 11ac packet is sent to 11ac receiver. False HT detection, when 11ac packet is sent to 11n receiver (assuming HT detection by 2 symbols). Assumptions for 11n HT detection: Case 1: HT detection Symbol by Symbol  HT detection on HTSIG2 only if HT detection on HTSIG1 passes. Case 2: HT detection jointly on two symbols  Compare QBPSK vs BPSK on all the 96 tones across two symbols VHT-SIGA2 (Alternative Q-BPSK/BPSK) Hongyuan Zhang et al.

1. 11ac False Detection 4 dB worse Hongyuan Zhang et al.

2. 11ac Miss Detection Hongyuan Zhang et al.

3. 11n False Detection For joint 2 symbol HT det, Scheme 2-B has limited gain over Scheme 1. Both are close to HT Detection on 1 symbol. Scheme 1 and 2-B does not make much difference If 11n Rx uses Symbol-by-Symbol HT detection Hongyuan Zhang et al.

Further Discussions The previous figure assumes the following auto-det algorithm at 11n receiver: Per-tone Q and I energy comparison and count the number of tones that Q is stronger than I. We believe this is a simple and practical auto-det approach. When 11n receiver conducts a joint energy comparison across tones (this is a complicated approach, but let’s assume there exists such an 11n device), Two HTSIG symbol detection will have higher HT false detection probability on receiving VHT preamble in Scheme 1, than Scheme 2B. However, when 11n false detection happens at such a device, it is almost sure that HTSIG CRC will fail, so it will go into CCA stage. This will NOT affect the network performance  because we believe that such 11n devices are scarce or even not exist. Hongyuan Zhang et al.

Advantages of Scheme 1 over Scheme 2 More reliable VHT detection (smaller VHT false detection probability) Easier implementation: unchanged from 11n HT detection (which is also implemented in the same device to support 11n). Hongyuan Zhang et al.

Conclusions The underlying assumption in [2] supporting Scheme 2 might not be practical. Even with the assumption above, the improvement by Scheme 2 is limited. Higher false 11n detection prob only for a particular non-practical detection method. False 11n detection in very few devices (i.e. 2-symbol HT detection using joint energy approach) will not affect the network performance. Scheme 1 is more reliable for VHT detection and easier to implement. Propose TGac to accept the auto-detection method as Scheme 1 [1]. Hongyuan Zhang et al.

Strawpoll : Do you support to do the following changes to the specification frame work document (IEEE 802.11-09/0992r9) R3.2.1.G: The 1st symbol of VHT-SIG-A shall be BPSK modulated. The subsequent symbol(s) second symbol of VHT-SIG-A shall be 90-degree rotated BPSK (QBPSK) modulated on its data tones for VHT auto-detect. Yes No Abstain Hongyuan Zhang et al.

Motion Move to make the following changes in the specification frame work document (IEEE 802.11-09/0992r9) R3.2.1.G: The 1st symbol of VHT-SIG-A shall be BPSK modulated. The subsequent symbol(s) second symbol of VHT-SIG-A shall be 90-degree rotated BPSK (QBPSK) modulated on its data tones for VHT auto-detect. Yes No Abstain Hongyuan Zhang et al.

References [1] 10-0070r5 “802.11ac Preamble” [2] 10-0359r0 “802.11ac preamble for VHT auto-detection” Hongyuan Zhang et al.