1. 1 CARA: Collision-Aware Rate Adaptation for IEEE 802.11 WLANs Jongseok Kim, Seongkwan Kim, Sunghyun Choi and Daji Qiao* School of Electrical Engineering and INMC Seoul National University *Department of Electrical and Computer Engineering Iowa State University Infocom 2006

2. 2 Outline Introduction Related work Collision-aware rate adaptation (CARA) –RTS Probing –CCA Detection Performance Evaluation Conclusion

3. 3 Introduction Using different modulation and channel coding schemes, IEEE 802.11 PHYs provide multiple transmission rates –11(2), 11b(4), 11a(8), 11g(12) IEEE 802.11 standard does not specify that how to utilize the multiple rates, many rate adaptation schemes have been proposed –RBAR, OAR, ARF …

4. 4 Introduction (cont.) Correctly determine the transmission rates can maximize the system performance –Increase the medium utilization –Prevent from the channel error The effectiveness of a rate adaptation scheme depends on how fast it can respond to the variation of wireless channel

5. 5 Introduction (cont.) The proposed rate adaptation schemes can be classified into two categories depending on the usage of the feedback from the receiver –Closed-loop –Open-loop

6. 6 Closed-loop scheme The receiver specifies its desired transmission rate –RTS/CTS exchange is needed Costly solution Rarely used in the practical –The control frames may be modified that it is not compliant to IEEE 802.11 standard

7. 7 Open-loop scheme The transmitter makes the rate adaptation decision solely based on its local information –Interaction is not required –ACK frame, SINR –Standard-compliant

8. 8 Motivation Due to the contention nature of IEEE 802.11 DCF, frame collisions are inevitable Unfortunately, Most open-loop rate adaptation schemes do not consider the collision effect –Determine the channel condition erroneously –Decrease the trans. rate over-aggressively

9. 9 Related work Automatic Rate Fallback (ARF) –Originally developed for Lucent Technologies’ WaveLAN-II WLAN devices –It is the most widely implemented rate adaptation scheme in IEEE 802.11 devices –A open-loop scheme

10. 10 ARF If two consecutive ACKs are not received correctly by the sender –Lower the transmission –Start the timer Either the timer expires or the succefully- received ACKs reach 10 –Raise the transmission rate –Timer cancelled

11. 11 ARF (cont.) ARF cannot react quickly when the channel condition fluctuates Periodically rate increase attempts may be failed mostly –Waste the channel utilization Do not work properly when multiple users contend for the shared wireless medium –Frame collision is not concerned

12. 12 CARA: Collision-Aware Rate Adaptation Transmitter combines adaptively RTS/CTS exchange Use the Clear Channel Assessment (CCA) functionality –Differentiate frame collisions from channel errors –Make the correct rate decisions Compatible with existing IEEE 802.11 devices

13. 13 CARA RTS probing CCA detection

14. 14 RTS probing - Assumptions Transmission error probability of an RTS frame is negligible –Small size –Base rate –RTS transmission failures are due to collisions After successful RTS/CTS, if data transmission failures –Must be due to channel errors

15. 15 RTS probing (cont.) Enables RTS/CTS exchange only when a data frame transmission fails

16. 16 RTS probing (cont.) P th = 1 N th = 2 M th = 10 P th = 1 N th = 2 M th = 10

17. 17 Threshold settings P th = 0 –RTS/CTS is always activated P th >= 1, N th = 1 –RTS/CTS is never activated P th >= 2, N th = 2 –If M th = 10. It is equivalent to ARF P th = 1, N th = 2 –CARA default values

18. 18 Example

19. 19 CCA detection DSSS PHY should provide capability to perform CCA detection –Energy Detection threshold –Ex: –80 dBm CCA Mode 1 –Energy above ED threshold CCA Mode 2 –CS only, CCA shall report a busy medium only upon detection of a DSSS signal CCA Mode 3 –CS with energy above ED threshold

20. 20 CCA detection (cont.) CCA detection is optional in CARA –It is a supplement to RTS Probing Transmitter station would retransmit without increasing n and lowering the transmission rate

21. 21 Performance Evaluation NS-2 Frame size : 1500 Bytes (unless specified) Infrastructure based

22. 22 One-to-one topology No contention

23. 23 Star-topology 10 meters

24. 24 Line-topology Performance anomaly problem Performance anomaly problem 70 meters Random frame size

25. 25 Random-topology 10 Stations 40 meters Random frame size

26. 26 Random-topology 40 meters Random frame size

27. 27 Adaptability in 1.6 secs. 5 Stations 40 meters Random frame size Star topology 5 Stations 40 meters Random frame size Star topology

28. 28 Comparison for the 30-sec. simulation run

29. 29 Conclusion RTS probing is very effective in differentiating collisions from channel errors CCA detection is useful when data durations are different among contending stations

30. 30 Conclusion (cont.) ARF is the most widely-deployed rate adaptation scheme in the commercial 802.11 devices –It is critical to have a well-designed link adaptation scheme with collision-awareness feature to replace ARF

31. 31 Thank You !!