1 / 29

Comparison of Multi-Channel MAC Protocols ——An Approach by Theory Analyze & Simulation

Comparison of Multi-Channel MAC Protocols ——An Approach by Theory Analyze & Simulation. 2013-11-19. Thesis Authors: J. Mo, H. W. So, J. Walrand. Lecture by Y. Yang EECS, York University. ★ Introduction : Goals & Methodolog y ★ Protocol-Categories : What to Be Compared?

damali
Download Presentation

Comparison of Multi-Channel MAC Protocols ——An Approach by Theory Analyze & Simulation

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. Comparison of Multi-Channel MAC Protocols——An Approach by Theory Analyze & Simulation 2013-11-19 Thesis Authors: J. Mo, H. W. So, J. Walrand Lecture by Y. Yang EECS, York University

  2. ★ Introduction :Goals & Methodology ★Protocol-Categories :What to Be Compared? ★Theoretical Analysis :Models & Results ★Simulation Analysis : Models & Results ★Conclusion :Is There a “Best” Protocol? ★Discussion :What Can We Learn?

  3. Background: Various Multi-Channel Mac Protocols Dynamic Channel Allocation (DCA) Channel Hoping Multiple Access (CHMA) Multichannel Access Protocol (MAP) 上海地 Modern Wireless Protocols (802.11 etc.) provides several channels for us. Therefore, Many MAC protocols are proposed exploiting this mechanism to improve the throughput. Slotted Seeded Channel Hopping (SSCH) Key Difference How Devices Agree On The Channel To Transmit? How To Resolve Potential Contentions? Channel Hopping multiple Access with packet Trains (CHAT) Channel Hoping Multiple Access (CHMA) Dynamic Private Channel (DPC) DCA With Power Control (DCA-PC) Multi-Channel MAC for Ad-Hoc (McMAC)

  4. Our Focus & Questions • Are their performancesdifferent? — Considering both 802.11 a and 802.11 b. • What issues may effect their performance? — Packet Length, Channel Switching Penalty, …… • Which one is better in specified cases? 上海地

  5. Execute a more complex simulation & analyze the results Research Methods Analyze the numerical results drawn from Math-Models Build Mathematical Models for each category, based on Markov Chain 上海地 Findings 2 Findings 1 ClassifyAll Protocols Into 4 Categories, for simplify the following analysis. Conclusions

  6. ★Introduction :Goals & Methodology ★Protocol-Categories :What to Be Compared? ★Theoretical Analysis :Models & Results ★Simulation Analysis : Models & Results ★Conclusion :Is There a “Best” Protocol? ★Discussion :What Can We Learn?

  7. What is the essential difference among these protocols? • Consider the generalized model of Multi-Channel transfer (by my understanding) A send the data to B via Channel CD B Replies A and negotiate a transfer channelCD, in the (control) Channel CC A Wants To Send Data To B A Sends RTS to B in a (control) Channel CC F I N 上海地 How is the control Channel CC allocated, so that A and B can meet each other before negotiating?

  8. Class #1 : Dedicated Control Channel Key Characters: Two Radios per device A fixed channel assigned for all control messages. 上海地 A B C D E F Advantages: Need notimesynchronization Limits: Cost of two radios Cost of channel waste

  9. Class #2 : Common Hopping Key Characters: One Radio per device Cyclethrough all channels synchronously 上海地 A B C D E Advantages: Use all channels for data transmitting Only one antenna needed Disadvantages: Unaware of other nodes’ status Switching cost Require tight synchronization

  10. Class #3 : Split Phase Key Characters: One radio per device A fixed channel carrying all control messages and some data. Split time into control phases and data phases 上海地 A B C D E Advantages: Use all channels for data transmitting Only one antenna needed Disadvantages: Require synchronization (looser than Common Hopping)

  11. Class #4 : Multiple Rendezvous ( McMAC ) Key Characters: Several negotiations can be processed in several channels at same time Pseudo-random sequence is used for each node to locate its “home channel” and its destination’s “home channel”, via exchanging random seeds in packets. 上海地 Advantages: Only One Radio Needed Use all channels Disadvantages: Switching Cost Require synchronization (looser than Common Hopping)

  12. A Classification Based On The Control-Channel-Assignment 上海地

  13. ★ Introduction :Goals & Methodology ★Protocol-Categories :What to Be Compared? ★Theoretical Analysis :Models & Results ★Simulation Analysis : Models & Results ★Conclusion :Is There a “Best” Protocol? ★Discussion :What Can We Learn?

  14. Assumptions 1. Single-Collision Model —— for simplifying the analysis 2. Small time slots, perfect synchronized. —— for discrete analysis 3. Only one data packet per channel agreement —— for standardizing all protocols 4. Packet size are independent Geometricallydistributed —— for simple Markov model 5. Devices always have data to send with a probability —— for throughput computing 上海地

  15. Basic Principles of the Markov Chain The Transition Probability is 上海地 i new comers join in with the probability Sk(i) jtransmitions terminate with the probability Tk(i) The average utilization is T is easy to be computed in all models K communications involved 2K devices Lcommunications involved 2L devices …… The Markov Chain Model

  16. Numerical Results (1) : Throughput in 802.11a vs 802.11 b 802.11 a---12Orthogonal Channels 802.11 b --- 3Orthogonal Channels 802.11 g---3 Orthogonal Channels

  17. Numerical Results (2) Date Rate By different parameters Small Packet 上海地 Large Packet

  18. ★Introduction :Goals & Methodology ★Protocol-Categories :What to Be Compared? ★Theoretical Analysis :Models & Results ★Simulation Analysis : Models & Results ★Conclusion :Is There a “Best” Protocol? ★Discussion :What Can We Learn?

  19. Difference From Numerical Model 1. CSMA/CA considered 2. Queuing of packets 3. Both CBR (constant bit rate) and stream (file transfer) tested 4. Allows multiple pairs sharing one channel during one data transmitting phase. 上海地

  20. Packet Delay in CBR Traffic , 802.11b Delta = 1 , one sender links to only 1 receiver, means lower contentions. Avg Packet Delay (ms) Delta = 5 , one sender links To 5 receivers, means higher contentions. Offered Load (Mbps)

  21. Packet Delay in CBR Traffic , 802.11a Delta = 1 , one sender links to only 1 receiver, means lower contentions. Avg Packet Delay (ms) Delta = 5 , one sender links To 5 receivers, means higher contentions. Offered Load (Mbps)

  22. Impacts of Medium Occupancy Time Limit in 802.11 a Senders have to relinquish the medium after 3.3 ms. Avg Packet Delay (ms) Senders have to relinquish the medium after 10.5 ms. Offered Load (Mbps)

  23. ★Introduction:Goals & Methodology ★Protocol-Categories :What to Be Compared? ★Theoretical Analysis :Models & Results ★Simulation Analysis :Models & Results ★Conclusion :Is There a “Best” Protocol? ★Discussion :What Can We Learn?

  24. Comparison of All The 4 Classes

  25. “different protocols are preferable depending on the operating conditions.” Dedicated Control Channel OR McMac ?

  26. ★ Introduction :Goals & Methodology ★Protocol-Categories :What to Be Compared? ★Theoretical Analysis :Models & Results ★Simulation Analysis :Models & Results ★Conclusion :Is There a “Best” Protocol? ★Discussion :What Can We Learn?

  27. Advantages Perfect proving, from theoretical math analysis to simple monte-carlo simulations, then a complex simulations. Reasonable explanations on the findings, including the gap to the conclusions in previous research. 上海地

  28. Maybe we can think further… If the geometric distribution is the best? From single-collision to a more actual scenario? Markov chain doesn’t consider the history effect, which caused some discussable hypothesis (for example in 3.2.1) in the paper. What if we consider the “memory effect”? …… 上海地

  29. Question Time Thank You

More Related