1 An approach based on shortest path and connectivity consistency for sensor network localization problems Makoto Yamashita (Tokyo Institute of Technology) I-Lin Wang (National Cheng Kung University) Zih-Cin Lin (National Cheng Kung University) 2012/08/22 ISMP 2012 (TU Berlin, Berlin, Germerny)
22
Embed
Makoto Yamashita (Tokyo Institute of Technology) I-Lin Wang (National Cheng Kung University)
An approach based on shortest path and connectivity consistency for sensor network localization problems. Makoto Yamashita (Tokyo Institute of Technology) I-Lin Wang (National Cheng Kung University) Zih-Cin Lin (National Cheng Kung University). ISMP 2012 (TU Berlin, Berlin, Germerny ). - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
An approach based on shortest path and connectivity consistency for sensor network localization problems
Makoto Yamashita (Tokyo Institute of Technology)I-Lin Wang (National Cheng Kung University)Zih-Cin Lin (National Cheng Kung University)
Framework of our approach2• Shortest Path• Gradient Method• Connectivity Consistency
Numerical Results3Multiple Start4
• Starting Point Selection• Combination of Location Results
Conclusion and Future works5
2012/08/22
3
SNL(Sensor Network Localization Problem)
We want to infer locationsfrom distance information
System of Equation
2012/08/22
4
Protein Structure
We can use distances between atoms measured by NOE effect.
We want to infer whole structure.
Structure determines chemical property of protein.
There are many other applications.
2012/08/221AX8, 1003 atoms
5
Existing Methods
Multidimensional Scaling[Merit] Low computation cost[Demerit] All distances are necessary
SDP relaxation (Biswas & Ye 2004)[Merit] High accuracy[Demerit] High compuation cost
We combine some heuristicsMiddle accuracy & Middle computation cost
2012/08/22
6
Our Approach
Combination of heuristicsShortest pathGradient methodConnectivity consistency
2012/08/22
7
Trilateration
Three anchors determine the location uniquely.
2012/08/22
1a
2a
3a
1d
2d
3d
8
Shortest Path
Propagation from anchors
Moredistance information⇒Shortest Path
Rough estimate Gradient method⇒
2012/08/22
9
Minimization of difference
Instead of solving the system
minimize
Effective for noisy distance input
2012/08/22
input true noise(e.g.:20% ~ 30%)
10
Gradient Method
Repeat
until
2012/08/22Shortest Path result
11
Connectivity Consistency
Given distance is usually less than radio range.
2012/08/22
Repulsion
Attraction
Adjustment
12
Framework of our heuristics
1. Select initial anchors2. Estimate roughly with Shortest Path3. Apply Gradient Method with estimate distance4. Apply Gradient Method with original distance5. Adjust sensors by Connectivity Consistency6. Go to Step 4