1 / 38

SELECT: Self-Learning Collision Avoidance for Wireless Networks

SELECT: Self-Learning Collision Avoidance for Wireless Networks. Chun-Cheng Chen, Eunsoo , Seo , Hwangnam Kim, and Haiyun Luo Department of Computer Science, University of Illinois, Urbana-Champaign IEEE Transactions on Mobile Computing, Vol. 7, No.3, 2008. Outline. Introduction

Download Presentation

SELECT: Self-Learning Collision Avoidance for Wireless Networks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. SELECT: Self-Learning Collision Avoidance for Wireless Networks Chun-Cheng Chen, Eunsoo, Seo,Hwangnam Kim, and HaiyunLuoDepartment of Computer Science, University of Illinois, Urbana-ChampaignIEEE Transactions on Mobile Computing,Vol. 7, No.3, 2008

  2. Outline • Introduction • Hidden/exposed terminal problem in 802.11 networks • Motivation • SELECT • a self-learning collision avoidance mechanism • Performance evaluation • Conclusion

  3. Introduction • Limited number of orthogonal channels restricts the deployment of 802.11 APs. • 3 channels for 802.11b/g, 12 for 802.11a • Interference range is long compared with communication range

  4. Introduction • Recent published data shows 40% of 802.11 APs are operating on channel 6 • In Boston, a max number of 85 APs are detected in the interference range • At least 30 APs are directly interfering with each other

  5. Hidden/exposed terminal problem Restrain by B’s CTS, Cannot reply E’s RTS C’s RTS collidewith A->B Restrain by RTS Restrain by CTS

  6. Drawbacks of hidden/ exposed receiver problem • Sender drops the head-of-line data packet • Resulting in a contention-induced packet loss • Unsuccessful RTS transmission, misled the sender to conclude • Receiver is unavailable (false link breakage is triggered) • Channel quality at the receiver side is low (Using low data transfer rate)

  7. Drawbacks of hidden/ exposed receiver problem • Unsuccessful RTS attempts inflate sender’s contention window • Repeated RTS attempts prevent the sender’s neighbor from transmitting • Low channel utilization • Hidden/exposed terminal problem will persists until the clients move and contention relation changes

  8. Motivation Potentialsender Exposedreceiver • Use MICA2 CC1000 to simulate the operation of 802.11 devices

  9. RSS at motes C and D while A is transmitting to B

  10. RSS vs. SR (successful ratio) • C→D, G →H are active • E →F serves as an additional interference • A →B, A records the RTS successful ratio

  11. RSS vs. RTS SR at mote A

  12. RSS vs. RTS SR over time

  13. Summary of RSS vs. RS • The RSS at the sender and the receiver has strong correlation • To estimate the RSS at the receiver from the sender is complex • The sender can use its RSS as an indicator of the status at receiver

  14. Overview of SELECT • Sender uses the detected RSS to map the receiver’s condition (successful ratio) • RSS is divided into several intervals, each interval has a corresponding SR • RSS ≧ CSthred → channel busy • SR ≧ threshold → transmit the data • SR < threshold → pretend the transmission is failed

  15. SELECT: self-learning collision avoidance • RSS-SR mapping maintenance • RSS-SR mapping lookup • Integration with 802.11 DCF • Intelligent SR threshold setup

  16. RSS-SR mapping maintenance • To update the SR within an interval Twin • Using a variable α (from 0 to 1)to indicate the weight of old data • α~1: the stored data is very new • α~0: the stored data is almost useless Current time Last update time

  17. RSS-SR mapping algorithm Calculate α Set updatevariable Update variable & timestamp

  18. RSS-SR mapping lookup • When a sender wants to send data to a receiver, the sender lookup the corresponding SR under current RSS • Remove out-of-date data first

  19. RSS-SR mapping lookup Channel Busy Return SR

  20. Integration with 802.11 DCF • When MAC module access the channel and the result is determined • Udp_RSS_SR • RSS_SR_Look-UP

  21. Integration with 802.11 DCF: when backoff expired • RSS ≧ CSthred → channel busy • Performs random backoff • RSS < CSthred → channel idle • SR ≧ threshold → transmit the data • SR < threshold → pretend the transmission is failed, also performs random Backoff

  22. Intelligent SR threshold setup (1) • The authors assume the successful ratio (SR) of each RSS is distributed according to the measured RSS distribution • When can a station measure RSS? • During random backoff

  23. Intelligent SR threshold setup (2) • Crssi = number of measured signal strength falls within interval RSSi • T=update interval • Trssi= the time that channel quality falls within interval RSSi

  24. Intelligent SR threshold setup (3) • If SRi < threshold, station won’t transmit during period T • The lose of throughput • △rssj= time spend to transmit a packet within interval RSSj interval

  25. Intelligent SR threshold setup (4) • Try to maximize the expected throughput Total spend time Time saved by a node at the low-SR rssi Available throughput

  26. Simulation setup • Ns-2 2.28 • Two-Ray Ground model • Communication range: 115m • RSSmin=-100dBm • RSS validation windows= 2 second • CBR/UDP traffic

  27. Exposed receiver • Station 3 is an exposed receiver

  28. Result of exposed receiver(w/o RTS/CTS) # of drop packets at Node 2 Throughput gain at Node 2

  29. Result of exposed receiver (w/o RTS/CTS) Successful ratio at Node 2 Throughput profile

  30. Result of exposed receiver(with RTS/CTS) # of drop packets at Node 2 Throughput gain at Node 2

  31. Result of exposed receiver (with RTS/CTS) Successful ratio at Node 2 Throughput profile

  32. Hidden receiver • Station 0 and 3 are hidden receivers to each other

  33. Normalized throughputx: with △: w/o RTS/CTS • 802.11 DCF • SELECT

  34. Normalized instantaneous throughput: 0->1

  35. Random topology • # of drop packets • Throughput gain

  36. Real experiment results by using MICA2 Throughput (pkt/second) RTS successful ratio

  37. Conclusion • The paper proposes SELECT • An effective and efficient self-learning collision mechanism • SELECT improves throughput by up to 140 % and the successful ratio by 302 percent

  38. Thank you!!

More Related