1. A Topology Control Approach for Utilizing Multiple Channels in Multi-Radio Wireless Mesh Networks (broadnet2005) Mahesh K. Marina, Samir R. Das
2006/9/14
Kim Young Hoon

2. Contents Introduction Problem Formulation Channel Assignment Algorithm
Simulations Results
Conclusions

3. Contents Introduction Problem Formulation Channel Assignment Algorithm
Simulations Results
Conclusions

4. Introduction Wireless mesh networks
Wired router nodes/wireless nodes
No need of infrastructure
Wider coverage
Mesh networks with multi-hop extension of standard
Different from LANs: risk of disconnection
All nodes use same channel in mesh for connectivity
Inefficient Utilization of Available Channels
Need to use Multiples Channels

5. Introduction Single radio for multiple channels?
Possible. But need of ...
Dynamically switch between channels
Tight time synchronization among nodes
Slow switching
 reducing synchronization requirements and overhead,
 increasing end-to-end delay
Require MAC or hardware modification
Therefore, Multiple radios per node
Effective use of given channels
Overcoming the deficiencies of single radio
interface를 싼 값에 장비할 수 있다는 점.
주어진 채널을 효과적으로 사용할 수 있다는 점
하나의 radio를 사용할 때의 단점들을 극복할 수 있다는 점.

6. Introduction What happens in multi-radio mesh network?
Disconnection between nodes can happen
Node’s transmission is interfered by other nodes’
So, a key issue in multi-radio mesh network architecture is Channel Assignment Problem.

7. Introduction Channel Assignment Problem
has to balance between connectivity and interference
is viewed as topology control problem (adjustable links between nodes in wireless)
In this paper, authors proposed base channel assignment
to obtain an initial, well-connected topology.

8. Contents Introduction Problem Formulation Channel Assignment Algorithm
Simulations Results
Conclusions

9. Problem Formulation
Channel Assignment Problem belongs to the class of NP-complete
Proof flow)
Channel Assignment Problem
 Topology Control Problem
Topology Control’s Target
 Reducing Interference
Channel Assignment Problem
 Optimization Problem
Optimization Problem
 Decision Problem
Showing that decision problem is in NP-complete (using minimum edge coloring)
For more detail, see 2nd part of the paper

10. Channel assignment algorithm
CLICA (Connected Low Interference Channel Assignment)
Polynomial time heuristic
Order nodes by their degree of flexibility
degree of flexibility: amount of freedom when choosing channel
Greedily assign channel between nodes

11. Channel assignment algorithm
Coloring uncolored links with a common color to already assigned radios
Coloring nodes which have no available radios
Greedily coloring uncolored links

12. Channel assignment algorithm - case 1
Initial order: a-d-c-b
Starting from a, assign channel a-b, b’s priority bumps up
Assign channel b-c, and c-d in similar manner
Node a and d have a common channel, so assign that to a-d
<5>
<1> b a c
<4>
<6>
<2> d
<7>
<3>

13. Channel assignment algorithm - case 2
Node a and d has two radios
Starting from a, assign channel a-b, b’s priority bumps up
Assign channel b-c, and c-d in similar manner
Node a and d have additional radios, so assign different channel to a-d
<5>
<1> b a c
<4>
<6>
<2> d
<7>
<3>

14. Channel assignment algorithm
Each coloring decision is made in a greedy fashion
Locally optimal choice
Theorem 2: CLICA algorithm yields a connectivity preserving color assignment

15. Contents Introduction Problem Formulation Channel Assignment Algorithm
Simulations Results
Conclusions

16. Simulation Results – part 1
Graph-based simulations
Interference and capacity properties of topologies generated by different channel assignment algorithms
Compared with CCA (Common Channel Assignment – assign same set of channelr to all nodes)
Measure:
Maximum link conflict weight – network wide interference
Maximum number of concurrent transmissions – total one-hop capacity
802.11b와 a를 본따 3개의 channel과 12개의 channel에서 실험하였으나 발표에서는 12개만을 인용하겠다.

17. Simulation Results – part 1
CCA
CCA interference performance is unaffected by the number of channels
CCA capacity performance shows a linear growth
CLICA
As the number of radios increases, interference goes up and capacity shows marginal perfromance
Minimum interference doesn’t match maximum capacity
DUE TO HEURISTIC NATURE

18. Simulation Results – part 2
Ns-2 simulations
Evaluating the performance of CLICA
Aggregate throughput and average delay
50 nodes with 250m TX range in 1000m x 1000m
550m interference range
physical layer model in ns-2
Fixed data rate of 2Mbps

19. Simulation Results – part 2: Single hop performance

20. Simulation Results – part 2: Multihop performance

21. Conclusions The authors
have formulated base channel assignment as a topology control optimization problem
solved the channel assignment (radio-channel mapping) problem in greedy way (called CLICA)
shows the interference-reducing results by simulations

22. Q&A