Greedy Routing with Guaranteed Delivery Using Ricci Flow Jie Gao Stony Brook University Rik Sarkar, Xiaotian Yin, Feng Luo, Xianfeng David Gu.

Slides:



Advertisements
Similar presentations
Steady-state heat conduction on triangulated planar domain May, 2002
Advertisements

Differential Forms for Target Tracking and Aggregate Queries in Distributed Networks Rik Sarkar Jie Gao Stony Brook University 1.
Orthogonal Drawing Kees Visser. Overview  Introduction  Orthogonal representation  Flow network  Bend optimal drawing.
Yang Yang, Miao Jin, Hongyi Wu Presenter: Buri Ban The Center for Advanced Computer Studies (CACS) University of Louisiana at Lafayette 3D Surface Localization.
Surface Reconstruction From Unorganized Point Sets
Discrete Differential Geometry Planar Curves 2D/3D Shape Manipulation, 3D Printing March 13, 2013 Slides from Olga Sorkine, Eitan Grinspun.
A Unified View to Greedy Routing Algorithms in Ad-Hoc Networks
Spherical Representation & Polyhedron Routing for Load Balancing in Wireless Sensor Networks Xiaokang Yu Xiaomeng Ban Wei Zeng Rik Sarkar Xianfeng David.
Sorce: Suggestive Contours for Conveying Shape. (SIGGRAPH 2003) Doug DeCarlo, Adam Finkelstein, Szymon Rusinkiewicz, Anthony Santella. 1 Suggestive Contours.
2/14/13CMPS 3120 Computational Geometry1 CMPS 3120: Computational Geometry Spring 2013 Planar Subdivisions and Point Location Carola Wenk Based on: Computational.
Compact and Low Delay Routing Labeling Scheme for Unit Disk Graphs Chenyu Yan, Yang Xiang, and Feodor F. Dragan (WADS 2009) Kent State University, Kent,
Surface Flattening in Garment Design Zhao Hongyan Sep. 13, 2006.
Nonlinear methods in discrete optimization László Lovász Eötvös Loránd University, Budapest
Discrete Geometry Tutorial 2 1
An Introduction to Polyhedral Geometry Feng Luo Rutgers undergraduate math club Thursday, Sept 18, 2014 New Brunswick, NJ.
Xianfeng Gu, Yaling Wang, Tony Chan, Paul Thompson, Shing-Tung Yau
Rüdiger Westermann Lehrstuhl für Computer Graphik und Visualisierung
Triangulated 3-manifolds: from Haken to Thurston Feng Luo Rutgers University May, 19, 2010 Barrett Memorial Lecture University of Tennessee.
Image Segmentation and Active Contour
Siggraph Course Mesh Parameterization: Theory and Practice Barycentric Mappings.
many branches of mathematics
Ad-Hoc Networks Beyond Unit Disk Graphs
1cs542g-term Notes. 2 Meshing goals  Robust: doesn’t fail on reasonable geometry  Efficient: as few triangles as possible Easy to refine later.
Exploration of Path Space using Sensor Network Geometry Ruirui Jiang, Xiaomeng Ban, Mayank Goswami, Wei Zeng, Jie Gao, Xianfeng David Gu Stony Brook University.
Mesh Simplification Global and Local Methods:
CENG 789 – Digital Geometry Processing 05- Smoothing and Remeshing
Graph Drawing Introduction 2005/2006. Graph Drawing: Introduction2 Contents Applications of graph drawing Planar graphs: some theory Different types of.
Cutting a surface into a Disk Jie Gao Nov. 27, 2002.
Jie Gao Joint work with Amitabh Basu*, Joseph Mitchell, Girishkumar Stony Brook Distributed Localization using Noisy Distance and Angle Information.
Computational Geometry Seminar Lecture 4 More on straight-line embeddings Gennadiy Korol.
Variational Principles and Rigidity on Triangulated Surfaces Feng Luo Rutgers University Geometry & Topology Down Under The Hyamfest Melbourne, Australia.
Parametrizing Triangulated Meshes Chalana Bezawada Kernel Group PRISM 3DK – 3DK – September 15, 2000.
Geometric Spanners for Routing in Mobile Networks Jie Gao, Leonidas Guibas, John Hershberger, Li Zhang, An Zhu.
Visualization and graphics research group CIPIC January 30, 2003Multiresolution (ECS 289L) - Winter MAPS – Multiresolution Adaptive Parameterization.
Mesh Parameterization: Theory and Practice Barycentric Mappings.
Dept. of Computer Science Distributed Computing Group Asymptotically Optimal Mobile Ad-Hoc Routing Fabian Kuhn Roger Wattenhofer Aaron Zollinger.
1 University of Denver Department of Mathematics Department of Computer Science.
Volume and Angle Structures on closed 3-manifolds Feng Luo Rutgers University Oct. 28, 2006 Texas Geometry/Topology conference Rice University.
Complex Model Construction Mortenson Chapter 11 Geometric Modeling
Part Two Multiresolution Analysis of Arbitrary Meshes M. Eck, T. DeRose, T. Duchamp, H. Hoppe, M. Lounsbery, W. Stuetzle SIGGRAPH 95.
Range Queries in Distributed Networks Jie Gao Stony Brook University Dagstuhl Seminar, March 14 th,
Delaunay Triangulations Presented by Glenn Eguchi Computational Geometry October 11, 2001.
Domain decomposition in parallel computing Ashok Srinivasan Florida State University COT 5410 – Spring 2004.
Intrinsic Parameterization for Surface Meshes Mathieu Desbrun, Mark Meyer, Pierre Alliez CS598MJG Presented by Wei-Wen Feng 2004/10/5.
A D V A N C E D C O M P U T E R G R A P H I C S CMSC 635 January 15, 2013 Quadric Error Metrics 1/20 Quadric Error Metrics.
Algorithms for Triangulations of a 3D Point Set Géza Kós Computer and Automation Research Institute Hungarian Academy of Sciences Budapest, Kende u
Scalable and Fully Distributed Localization With Mere Connectivity.
Boundary Recognition in Sensor Networks by Topology Methods Yue Wang, Jie Gao Dept. of Computer Science Stony Brook University Stony Brook, NY Joseph S.B.
Routing Considerations for Sensor Networks Lecture 12 October 12, 2004 EENG 460a / CPSC 436 / ENAS 960 Networked Embedded Systems & Sensor Networks Andreas.
Computer Graphics Some slides courtesy of Pierre Alliez and Craig Gotsman Texture mapping and parameterization.
Mesh Coarsening zhenyu shu Mesh Coarsening Large meshes are commonly used in numerous application area Modern range scanning devices are used.
1 11. Polygons Polygons 2D polygons ( 다각형 ) –Polygon sides are all straight lines lying in the same plane 3D polyhedra ( 다면체 )  chap. 12 –Polyhedra.
How to tell the differences between a Cat and a Dog Masoud Alipour Ali Farhadi IPM – Scientific Computing Center Vision.
UNC Chapel Hill M. C. Lin Delaunay Triangulations Reading: Chapter 9 of the Textbook Driving Applications –Height Interpolation –Constrained Triangulation.
Mesh Resampling Wolfgang Knoll, Reinhard Russ, Cornelia Hasil 1 Institute of Computer Graphics and Algorithms Vienna University of Technology.
Maximal Independent Set and Connected Dominating Set Xiaofeng Gao Research Group on Mobile Computing and Wireless Networking Univ. of Texas at Dallas.
Rigidity of the hexagonal triangulation of the plane and its applications Feng Luo, Rutgers University Joint with Jian Sun (Tsinghua Univ. China) and Tianqi.
Recent Progress in Mesh Parameterization Speaker : ZhangLei.
Reverse Engineering of Point Clouds to Obtain Trimmed NURBS Lavanya Sita Tekumalla Advisor: Prof. Elaine Cohen School of Computing University of Utah Masters.
CDS 301 Fall, 2008 Domain-Modeling Techniques Chap. 8 November 04, 2008 Jie Zhang Copyright ©
Decimation of Triangle Meshes Paper by W.J.Schroeder et.al Presented by Guangfeng Ji.
Algorithms and Networks
Morphing and Shape Processing
Jie Gao Stony Brook University CG Week 2012
Why Compare Surfaces? 1. Every object we see is a surface (almost).
Domain-Modeling Techniques
The Art Gallery Problem
The Art Gallery Problem
Volume and Angle Structures on closed 3-manifolds
Presentation transcript:

Greedy Routing with Guaranteed Delivery Using Ricci Flow Jie Gao Stony Brook University Rik Sarkar, Xiaotian Yin, Feng Luo, Xianfeng David Gu

Greedy Routing Assign coordinates to nodes Message moves to neighbor closest to destination Simple, easy, compact 2

Greedy Routing Can get stuck 3 x y z

Use Ricci flow to make all holes circular Greedy routing does not get stuck at holes. 4 

Talk Overview Theory on Ricci flow --smoothing out surface curvature Distributed algorithm on sensor network – Extract a triangulation from unit disk / quasi unit disk graphs – Ricci flow: distributed iterative algorithm 5

Part I : Curvature and Ricci Flow 6 Curvature: κ = 1/R Flatter curves have smaller curvature. R

Changing the Curvature 7 Flatten a curve: κ  0 Make a cycle round κ  1/R R

Discrete Curvature 8 Turning angle Sum of curvature = 2π

Discrete Curvature in 2D Triangulated Surface 9

10 Deviation from straight line For an interior vertex For a vertex on the boundary Deviation from the plane corner angle

Total Discrete Curvature Gauss Bonnet Theorem: total curvature of a surface M is a topological invariant: Ricci flow: diffuse uneven curvatures to be uniform 11 Euler Characteristic: 2 – 2(# handles) – (# holes)

Sanity check: total curvature h holes, Euler characteristics = 2-(h+1) Outer boundary: curvature 2π Each inner hole: -2π Interior vertex: 0 Total curvature: (1-h)2π 12

What they look like 13 Negative CurvaturePositive Curvature

Ricci Flow: diffuse curvature Riemannian metric g on M: curve length We modify g by curvature Curvature evolves: Same equation as heat diffusion 14 Δ: Laplace operator

Ricci Flow: diffuse curvature “Ricci energy” is strictly convex  unique surface with the same surface area s.t. there is constant curvature everywhere. Conformal map: angle preserving 15

Ricci Flow in our case Target: a metric with pre-specified curvature 16 Boundary cycle

Discrete Ricci Flow Metric: edge length of the triangulation – Satisfies triangle inequality Metric determines the curvature Discrete Ricci flow: conformal map that modifies edge length to smooth out curvature. 17

Discrete Flow : Use Circle Packing Metric Circle packing metric: circle of radius γ i at each vertex & intersection angle φ ij. Remark: no need of an embedding (vertex location) 18 With γ, φ one can calculate the edge length l and corner angle θ, by cosine law Discrete Ricci flow modifies γ, preserves φ

Discrete Flow : Use Circle Packing Metric Circle packing metric: circle of radius γ i at each vertex & intersection angle φ ij. Take Discrete Ricci flow: 19 Target curvature Current curvature

Background [Hamilton 82, Chow 91 ] : Smooth curvature flow flattens the metric [Thurston 85, Sullivan and Rodin 87 ] : Circle packing – discrete conformal maps [He and Schramm 93, 96 ] : Discrete flow with non-uniform triangulations [Chow and Luo 03 ] : Discrete flow, existence of solutions, criteria, fast convergence 20

Part II : Ricci Flow in Sensor Networks Build a triangulation from network graph – Requirement: triangulation of a 2D manifold Compute virtual coordinates – Apply distributed Ricci flow algorithm to compute edge lengths d(u, v) with the target curvature (0 at interior vertices and 2π/|B| at boundary vertices B) – Calculate the virtual coordinates with the edge lengths d(u, v) 21

Issues with Network Graph We need a triangulation s.t. locally we have 2D patches. 22 Crossing Edges degeneracies

Location-based Triangulation Local delaunay triangulations with locations [Gao, Guibas, Hershberger, Zhang, Zhu - Mobihoc 01] Compute Delaunay triangulations in each node’s neighborhood Glue neighboring local triangulations Remove crossing edges We now extend it to quasi-UDG 23

Location-based triangulation Handling Degeneracies : Insert virtual nodes Locations not required for Ricci flow algorithm Set all initial edge lengths = 1 24

Location Free Triangulation Landmark-based scheme : planar triangulation. [Funke, Milosavljevic SODA ’07] Set all edge lengths = 1 25

Ricci Flow in Sensor Networks All edge length = 1 initially – Introduces curvature in the surface Use Ricci flow to reach target curvature κ’ – Take tangent circle packing metric: γ=1/2, φ=0 initially – Interior nodes: target curvature = 0 – Nodes on boundary C: target curvature = - 2π/|C| – Modify u i = logγ i by δ (κ’-κ) – Until the curvature difference < ε |C|: Length of the boundary cycle

Ricci Flow in Sensor Networks Ricci flow is a distributed algorithm – Each node modifies its own circle radius. Compute virtual coordinates – Start from an arbitrary triangle – Iteratively flatten the graph by triangulation. 27

Examples 28

Routing 29

Experiments and Comparison NoGeo : Fix locations of boundary, replace edges by tight rubber bands : Produces convex holes [Rao, Papadimitriou, Shenker, Stoica – Mobicom 03] Does not guarantee delivery: cannot handle concave holes well MethodDeliveryAvg StretchMax Stretch Ricci Flow100% NoGeo83.66%

Convergence rate Curvature error bound ε Step size δ # steps = O(log(1/ε)/δ) 31 Iterations Vs error

Theoretical guarantee of delivery Theoretically, Ricci flow is a numerical algorithm. Triangles can be skinny. In our simulations, we have not encountered any delivery problems In theory, one can refine the triangulation to deal with skinny triangles 32

Summary Deformation of network metric by smoothing out curvatures. All holes are made circular. Greedy routing always works. Future work: – Guarantee of stretch? – Load balancing? 33

Mobius Ambiguity Many different deformations are conformal – Equivalent under mobius transform Local operation : need to fix target deformation 34

Convergence rate Curvature error bound ε Step size δ # steps = O(log(1/ε)/δ) Experiments: ε=1e Iterations Vs Number of Nodes