Song Wei

1. Song Wei Enabling Distributed Throughput Maximization inWireless Mesh Networks APartitioning Approach

2. Problem • Channel Scheduling • Centralized • Distributed

3. Model • Graph theory • Conflict graph • Graph coloring • Throughput Constrains • Arrival Process • Stable Region

4. A Simple Example A Simple Example 6 6 5 5 1 1 4 4 2 2 3 3

5. A Simple Example • Conflict Graph

6. A Simple Example • Stability Region :Arrival process on link (i, j) lambda: arrival rate Stability Region: For single radio, b=1. For two radios, b=2. Etc

7. A Simple Example • Scheduling • Centralized: Yes • Distributed: ?

8. A Simple Example • Partition: 100% Distributed Scheduling 6.1 6.1 5,6 5,6 1,2 1,2 4,5 4,5 3,4 2,3 3,4 2,3

9. A Simple Example • Capability Increase 6 6 6 5 5 5 1 1 1 4 4 4 3 3 3 2 2 2

10. A Simple Example • Insight • Topology study • Divide and conquer • Better the dividing

11. Topology for 100% DS • Local Pooling

12. Topology for 100% DS • Local Pooling Building Blocks • 1-5 • 6 • 7

13. Topology for 100% DS • Construction of Local Pooling network • Clique • Clique Connection • Tree of Cliques • Segments connected Cliques • Tree of Blocks

14. Topology for 100% DS • Tree

15. Partition • Problem • Multi Radio – Multi Channel Model • Formulation

16. Partition • Algorithm • BFS : O(k(m+n)) • MCI :

17. Improve Capacity

18. Improve Capacity • How to define Capacity here? • Average • Max • What’s the bottle neck? • Degrees of vertexes in first partitions are too high • How to solve it? • Balance degree of vertex

19. Improve Capacity • R-Greedy

20. Improve Capacity • R-MaxD • Alter path • Consider Max Degree Vertex

21. Improve Capacity • R-AvgD • Alter path • Consider All Vertex with no TF degradation

22. Performance Evaluation • Partitioning

23. Performance Evaluation • Capability Expansion

24. Questions • Thanks