1. Cross Layer Design in Wireless Mesh Networks
指導教授：吳和庭教授
報告：江昀庭
2012/5/8

2. Source Reference
[1] N.M. Salleh, M. Muhammad, M. S. Zakaria, V.R. Gannapathy, M.K. Suaidi, I. M.Ibrahim, M. Z. A. AbdulAziz, M.S. Johar, M.R. Ahmad “Wireless mesh network - Cross layer design challenge!!” Applied Electromagnetics, APACE Asia-Pacific Conference on Digital Object Identifier: /APACE Publication Year: 2007 , Page(s): [2] Arianpoo, N.; Jokar, P.; Leung, V.C.M. “Enhancing TCP Performance in Wireless Mesh Networks by Cross Layer Design” Computing, Networking and Communications (ICNC), 2012 International Conference on Digital Object Identifier: /ICCNC Publication Year: 2012 , Page(s):

3. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

4. Introduction
Cross layer networking in the physical layer, network layer and transport layer.
Enhancing TCP performance by Cross Layer Design. (MHHP)

5. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

6. Purpose
Improve bit error rate and noise between neighbor in WMNs.

7. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

8. Approach (1) Can be performed in two ways:
The first approach is to improve the performance of a protocol layer by taking into account parameters in other protocol layers. Typically, parameters in the lower protocol layers are reported to higher layers.
The second approach of is to merge several protocols into one component.

9. Approach (2)

10. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

11. Challenge (1) Physical Layer
It’s not fundamentally different from other wireless technologies.
Network Layer
the wireless link where WMNs are radically different from 3G systems, WLANs and WMANs. All these technologies use a single wireless link and hence have no need for a network layer.

12. Challenge (2) Transport Layer
Transport protocol is used TCP on the Internet. Unfortunately, TCP was designed and fine-tuned for wired networks where most packet losses are due to buffer overflows in the routers.
Bit-error rate in wireless
Move and Work off

13. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

14. Fair End-to-end Bandwidth Allocation
FEBA bandwidth allocation is based on the weight of each flow.
FEBA assigns larger bandwidth to the nodes that relay a larger number of flows.

15. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

16. More Hops Higher Priority (1)
To solve the unfairness problem of the nodes with a larger number of hops.
Propose an algorithm that gives a higher priority to the flows coming from the farther nodes.
FlowPriority
= PacketPriority × DistancetoDestination (1)

17. More Hops Higher Priority (2)
Giving priority to the nodes based on the number of traversed hops might increase the computational load; but that is not an issue in WMN.

18. More Hops Higher Priority (3)
It’s an example.
Assume same priority for all flows.
Throughput increase 115%

19. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

20. Analytical Justification (1)
X is the transmit rate in bytes per second
s is packet size in bytes
RTT is the Round Trip Time
p is the loss event rate,
RTO is the TCP retransmission time out value in seconds
b is the number of packets acknowledged by a single TCP ACK packet.

21. Analytical Justification (2)
This method is in favore of the nodes with larger number of hops; MHHP decreases RTT of farther nodes and increases RTT of closer nodes. Balancing RTT helps to make balance the throughput in the network.

22. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

23. Performance Evaluation (1)
Fig. 2. End to end throughput of the longest flow

24. Performance Evaluation (2)
Decreases RTT of farther nodes and increases RTT of closer nodes ?
Fig. 3. Round Trip Time of the longest flow

25. Performance Evaluation (3)
Fig. 4. End to end throughput of the shortest flow

26. Performance Evaluation (4)
Fig. 5. Round Trip Time of the shortest flow

27. Performance Evaluation (5)
Fig. 6. End to end throughput of the longest flow - ring topology

28. Performance Evaluation (6)
Fig. 7. End to end throughput of the longest flow - star topology

29. Performance Evaluation (7)
Fig. 8. End to end throughput of the longest flow - triangular topology

30. Performance Evaluation (8)
Fig. 9. Round Trip Time of the longest flow - ring topology

31. Performance Evaluation (9)
Fig. 10. Round Trip Time of the longest flow - star topology

32. Performance Evaluation (10)

33. Outline Introduction Cross Layer Design
Purpose
Approach
Challenge
Fair End-to-end Bandwidth Allocation(FEBA)
More Hops Higher Priority (MHHP)
Analytical Justification
Performance Evaluation
Conclusion

34. Conclusion (1)
Cross Layer Design improve bit error rate and noise between neighbor in WMNs.
In order to improve protocol efficiency, cross layer design becomes indispensable.
Cross-layer design have risks due to loss of protocol layer abstraction

35. Conclusion (2)
The impact of using MHHP on throughput in WMN is significant, especially when the number of hops is less than 5.

36. Q&A
Thanks for your listening