September 2012 doc IEEE 802 11 121104 r

September 2012 doc. : IEEE 802. 11 -12/1104 r 1 11 ah Interframe Spacing Values Date: 2012 -09 -17 Authors: Submission Slide 1 Tevfik Yucek, Qualcomm Inc.

September 2012 doc. : IEEE 802. 11 -12/1104 r 1 Authors: Submission Slide 2 Tevfik Yucek, Qualcomm Inc.

September 2012 doc. : IEEE 802. 11 -12/1104 r 1 Authors: Submission Slide 3 Tevfik Yucek, Qualcomm Inc.

September 2012 doc. : IEEE 802. 11 -12/1104 r 1 Introduction • SIFS time and slot duration are important keeping the overhead small [1] • We investigate the components that makes up SIFS and Slot durations and propose values that are suitable for 802. 11 ah Submission Slide 4 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 Definitions Submission Slide 5 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 Definitions • SIFSTime = a. Rx. RFDelay + a. Rx. PLCPDelay + a. MACProcessing. Delay + a. Rx. Turnaround. Time. • Slot. Time = a. CCATime + a. Rx. Turnaround. Time + a. Air. Propagation. Time+ a. MACProcessing. Delay Submission Slide 6 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 Values • a. Air. Propagation. Time – Twice the propagation time (in microseconds) for a signal to cross the maximum distance between the most distant allowable STAs that are slot synchronized. – Assuming 300 and 900 meters range for 11 n/ac and 11 ah respectively, corresponding values are 2 us and 6 us. • a. Rx. PLCPDelay, a. MACProcessing. Delay, a. Tx. PLCPDelay – These are digital processing delays and depend on the digital clock rate. Assuming 11 n/ac baseband is downclocked by 10, values need to be increased by 10. – MAC processing time / Tx. PLCPDelay for Slot is set to zero since the MAC calculation can be done before transmission. • a. Tx. Ramp. On. Time and a. Rx. Tx. Switch. Time are assumed to be the same for 11 n/ac and 11 ah • a. Tx. RFDelay and a. Rx. RFDelay – Assuming the same filter order, delay increases 10 x due to 10 x bandwidth scaling (2 MHz vs. 20 MHz). – We assume 1 MHz will use the same filter with 2 MHz mode. If narrower filters are used, we assume smaller digital delay of 1 MHz will compensate the longer filter delay. Submission Slide 7 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 CCA Time: Theoretical considerations • Due to 20 x BW reduction, CCA time needs to be increased 20 x compared to 11 n/ac – For same SNR, 1 MHz STF requires 2 x the integration time of 2 MHz (less number of samples per symbol). • Furthermore, to improve the detection to an SNR supported by Rep 2 packets, CCA time needs to be increased by 2 – Required SNR is 3 d. B lower for MCS 0 rep 2 • However, STFs are power boosted by 3 d. B for rep 2 • This results in theoretical value of 4*20*2/2 = 80 us CCA duration Submission Slide 8 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 CCA Time: Practical requirements • 10% PER point for MCS 0 -Rep 2 requires >8 d. B SNR at Urban Macro channel model • Figures in the next page shows the ROC curves for detecting valid PPDU transmissions in – 11 d. B STF-SNR in SCM Urban Macro channel model – 0 d. B STF-SNR in AWGN channel – 3 d. B STF-SNR in AWGN channel • Simple correlation algorithm based on STFs is used • It is possible to meet the 90% detection requirement using 40 us CCA duration for both bases (fading and AWGN). – Values less then 40 us can be used. However, 40 us gives some room in implementation error. For example, a single detection threshold is used across different channel conditions. Submission Slide 9 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 CCA Results – Urban Macro Submission Slide 10 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 CCA Results – AWGN Submission Slide 11 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 CCA Results – AWGN Submission Slide 12 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 Slot Time 11 ac 11 ah CCA Time 4 40 Air Propagation Time 2 6 a. MACProcessing. Delay 1 0 a. Rx. Turnaround. Time (a. Tx. PLCPDelay+a. Rx. Tx. Switch. Ti me+a. Tx. Ramp. On. Time+a. Tx. RFDel ay 2 (1, 0. 25, 0. 5) 5. 5 (0+0. 25+5) Slot. Time 9 51. 5 52 Submission Slide 13 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 SIFS 11 n/ac 11 ah a. Rx. RFDelay . 5 5 a. Rx. PLCPDelay 12. 5 125 a. MACProcessing. Delay 1 10 a. Rx. Turnaround. Time (a. Tx. PLCPDelay+a. Rx. Tx. Switch. Tim e+a. Tx. Ramp. On. Time+a. Tx. RFDelay 2 (1, 0. 25, 0. 5) 15. 5 (10+0. 25+5) SIFS 16 Submission Slide 14 155. 5 156 Tevfik Yucek, Qualcomm Inc.

September 2012 doc. : IEEE 802. 11 -12/1104 r 1 SIFS: Multiple of symbol duration • We propose to use 160 us SIFS time which is a multiple of 40 us symbol duration (normal GI) • This is less efficient but it would allow to the NAV to be expressed in terms of symbol duration, allowing for compressed representation of Duration/NAV field in MAC/PHY fields. • If any transmission in a TXOP uses short GI, the duration defined by 40 us unit might be slightly longer than the actual transmission sequence, but the difference is less than 1 symbol; which should not cause big impact • Note that 11 ac short GI has similar issue Submission Slide 15 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 Straw Poll • Do you support the following proposed interframe spacing related values – – SIFS=160 us, Slot=52 us a. CCATime = 40 us a. Air. Propagation. Time = 6 us – Yes: – No: – ABS: Submission Slide 16 Tevfik Yucek, Qualcomm Inc.

doc. : IEEE 802. 11 -12/1104 r 1 References • [1] IEEE 11 -12 -0362, On the 802. 11 ah Performance Submission Slide 17 Tevfik Yucek, Qualcomm Inc.