November 2001 doc IEEE 802 11 01646 r

  • Slides: 10
Download presentation
November 2001 doc. : IEEE 802. 11 -01/646 r 0 Why the TGg compromise

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Why the TGg compromise will work a MAC perspective Ronald Brockmann (Intersil) Wim Diepstraten (Agere) Maarten Hoeben (Intersil) Sid Schrum (Texas Instruments) Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Goals • Use OFDM

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Goals • Use OFDM concurrently with CCK in the presence of 802. 11 b – Efficient mixed 802. 11 b/802. 11 g operation – PBCC and CCK-OFDM options • Don’t break 802. 11 b • 802. 11 g product achieves in a mixed environment same or higher performance than 802. 11 b • Works for current MAC with or without TGe enhancements Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Assumptions • 802. 11

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Assumptions • 802. 11 b PHY may not correctly see a pure OFDM frame as a busy channel • 802. 11 g PHY can acquire on both Barker and OFDM preamble on every packet • 802. 11 b rates are basic rates – Control frames, multicasts and broadcasts are sent using Barker/CCK Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 How to protect OFDM

November 2001 doc. : IEEE 802. 11 -01/646 r 0 How to protect OFDM frames? • Use of RTS/CTS and protections of NAV – An existing MAC mechanism to deal with hidden node situations – The NAV of the RTS or CTS protects the OFDM Data Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 How about the performance?

November 2001 doc. : IEEE 802. 11 -01/646 r 0 How about the performance? • Example Data frame of 1500 octets + additional MAC overhead: – 802. 11 b – { Data – Ack } • 1213 цs + 106 цs = 1330 цs – 802. 11 g – { RTS – CTS – Data (24 Mbps) – Ack } • 124 цs + 106 цs + 16 цs + 532 цs + 16 цs + 106 цs = 910 цs Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Further enhancements • Don’t

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Further enhancements • Don’t use RTS/CTS for shorter frames – The OFDM Data frame + preamble may be shorter than the RTS itself – A TCP ACK frame 24 Mbit/s is just 2 slot times long! • Use CCK rates for shorter frames • Allow Acknowledgements to be sent on rates other than a basic rate. – 11 Mbps RTS/CTS + 24 Mbps Data – Ack is as efficient as 11 Mbps Data – Ack – MAC enhancements to increase efficiency of the MAC for certain PHYs may be incorporated by TGe or standardized by a separate TG Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Centralized solutions • The

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Centralized solutions • The PCF allows efficient OFDM frame exchanges protected by the Contention Free Period – No additional overhead is necessary to protect the OFDM frame exchanges from 802. 11 b • How about the HCF – HCF is based on the same mechanisms – No known issues with current HCF draft – Corner cases may need further evaluation Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 How to solve timing

November 2001 doc. : IEEE 802. 11 -01/646 r 0 How to solve timing issues • General rules of thumb: – Use OFDM SIFS after an OFDM frame on which a response follows. • Use the SIFS associated with the received frame modulation – Use 802. 11 b IFS and slottimes otherwise Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Some timing examples •

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Some timing examples • <DIFS>[slots…] Data<SIFS>Ack • <DIFS>[slots…] RTS<SIFS>CTS<SIFS>Data<SIFS>Ack • CFPoll<SIFS>Data<SIFS>CFAck+CFPoll… • CFData+CFPoll<PIFS>CFPoll<SIFS>Null… Submission Ronald Brockmann, et. al.

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Conclusions • There is

November 2001 doc. : IEEE 802. 11 -01/646 r 0 Conclusions • There is enough evidence to believe that the proposed compromise TGg PHY can be made to work efficiently with the existing MAC and the proposed TGe enhancements • Both PBCC and CCK-OFDM use existing 802. 11 b preamble so there is no CCA issue – Works smoothly alongside existing 802. 11 b equipment Submission Ronald Brockmann, et. al.