university of denver department of mathematics department of computer science

62
1 University of Denver Department of Mathematics Department of Computer Science

Upload: joseph-rosario

Post on 03-Jan-2016

26 views

Category:

Documents


0 download

DESCRIPTION

University of Denver Department of Mathematics Department of Computer Science. Applications Ad hoc Wireless networks Robot Route Planning in a terrain of varied types (ex: grassland, brush land, forest, water etc.) Geometric graphs Planar graph Unit disk graph. Geometric Routing. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: University of Denver Department of Mathematics Department of Computer Science

1

University of Denver

Department of Mathematics

Department of Computer Science

Page 2: University of Denver Department of Mathematics Department of Computer Science

2

Geometric RoutingGeometric Routing

Applications Ad hoc Wireless networks Robot Route Planning in a terrain of varied types (ex:

grassland, brush land, forest, water etc.)

Geometric graphs Planar graph Unit disk graph

Page 3: University of Denver Department of Mathematics Department of Computer Science

3

General graphGeneral graph

A graph (network) consists of nodes and edges represented as G(V, E, W)

a b

c d

ee1(1)

e3(5)

e4(2)

e2(2)

e6(2)

e5(2)

Page 4: University of Denver Department of Mathematics Department of Computer Science

4

Planar Graphs Planar Graphs

A Planar graph is a graph that can be drawn in the plane such that edges do not intersect

a b

c d

e

Examples: Voronoi diagram and Delaunay triangulation

Page 5: University of Denver Department of Mathematics Department of Computer Science

5

AGENDAAGENDA

Topics:1. Minimum Disk Covering Problem (MDC)

2. Minimum Forwarding Set Problem (MFS)

3. Two-Hop Realizability (THP)

4. Exact Solution to Weighted Region Problem (WRP)

5. Raster and Vector based solutions to WRP Conclusion Questions?

Page 6: University of Denver Department of Mathematics Department of Computer Science

6

Topics:1. Minimum Disk covering Problem (MDC)

2. Minimum Forwarding Set Problem (MFS)

3. Two-Hop realizability (THP)

4. Exact solution to Weighted Region Problem (WRP)

5. Raster and vector based solutions to WRP Conclusion Questions?

Page 7: University of Denver Department of Mathematics Department of Computer Science

7

1 . Minimum Disk Covering Problem (MDC)

Cover Blue points with unit disks centered at Red points !! Use Minimum red disks!!

1

Page 8: University of Denver Department of Mathematics Department of Computer Science

8

Other VariationOther Variation

Cover all Blues with unit disks centered at blue points !! Using Minimum Number of disks

1

Page 9: University of Denver Department of Mathematics Department of Computer Science

9

ComplexityComplexity

MDC is known to be NP-complete Reference “Unit Disk Graphs”

Discrete Mathematics 86 (1990) 165–177, B.N. Clark, C.J. Colbourn and D.S. Johnson.

Page 10: University of Denver Department of Mathematics Department of Computer Science

10

Previous work (Cont…)Previous work (Cont…)

A 108-approximation factor algorithm for MDC is known

“Selecting Forwarding Neighbors in Wireless Ad-Hoc Networks”

Jrnl: Mobile Networks and Applications(2004)Gruia Calinescu ,Ion I. Mandoiu ,Peng-Jun Wan Alexander Z. Zelikovsky

Page 11: University of Denver Department of Mathematics Department of Computer Science

11

Previous methodPrevious method

Tile the plane with equilateral triangles of unit side

Cover Each triangle by solving a Linear program (LP)

Round the solution to LP to obtain a factor of 6 for each triangle

Page 12: University of Denver Department of Mathematics Department of Computer Science

12

The method to cover triangleThe method to cover triangle

1

Page 13: University of Denver Department of Mathematics Department of Computer Science

13

Covering a triangleCovering a triangle

IF No blue points in a triangle- NOTHING TO DO!!

IF ∆ contains RED + BLUE THEN Unit disk centered

at RED Covers the ∆

Assume BLUE + RED do not share a ∆

Page 14: University of Denver Department of Mathematics Department of Computer Science

14

Covering a triangle cont…Covering a triangle cont…

A

CB

T1

T2

T3

Page 15: University of Denver Department of Mathematics Department of Computer Science

15

Covering a triangle cont…Covering a triangle cont…

1. Using Skyline of disks

2. cover each of the 3 sides with 2-approximation

3. combine the result to get:

6-approximation for each ∆

Page 16: University of Denver Department of Mathematics Department of Computer Science

16

Desired PropertyDesired Property P

Skyline gives an approximation factor of 2

No two discs intersect more than once inside a triangle

No Two discs are tangent inside the triangle

Page 17: University of Denver Department of Mathematics Department of Computer Science

17

Unit disk intersects at most 18 trianglesUnit disk intersects at most 18 triangles

It can be easily verified that a Unit disk intersects at most 18 equilateral triangles in a tiling of a plane

Page 18: University of Denver Department of Mathematics Department of Computer Science

18

Result 108-approximationResult 108-approximation

Covered each triangle with approximation factor of 6

Optimal cover can intersect at most 18 triangles

Hence, 6 *18 = 108 - approximation

Page 19: University of Denver Department of Mathematics Department of Computer Science

19

ImprovementsImprovements

CAN WE use a larger tile? split the tile into two regions? get better than 6-approximation by different

tiling? cover the plane instead of tiling?

Page 20: University of Denver Department of Mathematics Department of Computer Science

20

Can we use a larger tile?Can we use a larger tile?

If tile is larger than a unit diameter !!

Unit disc inside Tile cannot cover the tile

Hence we cannot use previous method

Page 21: University of Denver Department of Mathematics Department of Computer Science

21

Split the tile into two regionsSplit the tile into two regions

v0

v1

v2

v3

v4

n = 2m +1

n = 5; m = 2

Page 22: University of Denver Department of Mathematics Department of Computer Science

22

Different shape Tile?Different shape Tile?

Each side with 2-approx. factor

Hence 8 for a square Unit disk can intersect

14 such squares 14 * 8 =112 No Gain by such

method

2

1

Page 23: University of Denver Department of Mathematics Department of Computer Science

23

Different shape Tile?Different shape Tile?

Each side with 2-approx. factor

Hence 12 for a hexagon Unit disk can intersect

12 such hexagons 12 * 12 =144 No Gain by such

method

Page 24: University of Denver Department of Mathematics Department of Computer Science

24

Our ApproachOur Approach

How about using a unit diameter hexagon as a tile

Split the tile into 3 regions around the hexagon

Does this give a better bound?

Page 25: University of Denver Department of Mathematics Department of Computer Science

25

Hexagon- split it into 3 regionsHexagon- split it into 3 regions

Partition Hexagon into 3 regions (Similar to triangle)

Obtain 2-approximation for each side 6-approximation for hexagon

Unit disk intersects 12 hexagons

Hence, 6 * 12 = 72-approximation

T1

T2

T3

Page 26: University of Denver Department of Mathematics Department of Computer Science

26

CoveringCovering

Instead of tiling the plane, how about covering the plane?

Page 27: University of Denver Department of Mathematics Department of Computer Science

27

Conclusion of MDCConclusion of MDC

Conjecture: A unit disk will intersect at least 12 tiles of any covering of R2 by unit diameter tiles

Each tile has an approximation of 6 by the known method

Cannot do better than 72 by the method used

Page 28: University of Denver Department of Mathematics Department of Computer Science

28

Topics:1. Minimum Disk covering Problem (MDC)

2. Minimum Forwarding Set Problem (MFS)

3. Two-Hop realizability (THP)

4. Exact solution to Weighted Region Problem (WRP)

5. Raster and vector based solutions to WRP Conclusion Questions?

Page 29: University of Denver Department of Mathematics Department of Computer Science

29

2. Minimum Forwarding Set Problem (MFS)2. Minimum Forwarding Set Problem (MFS)

Cover blue points with unit disks centered at red points, now all red points are inside a unit disk

s

A

ONE-HOP REGION

Page 30: University of Denver Department of Mathematics Department of Computer Science

30

Previous work (MFS)Previous work (MFS) Despite its simplicity, complexity is

unknown

3- and 6-approximation algorithms known Algorithm is based on property P

Page 31: University of Denver Department of Mathematics Department of Computer Science

31

Desired Property Desired Property P P AgainAgain

1. No two discs intersect more than once along their border inside a region Q

2. No Two discs are tangent inside a region Q

3. A disk intersect exactly twice along their border with Q

P1

P3

Q

Page 32: University of Denver Department of Mathematics Department of Computer Science

32

PropertyProperty P

Property P applies if the region is outside of disk radius 2

Unit disk 2

Q

s

A

Page 33: University of Denver Department of Mathematics Department of Computer Science

33

Redundant pointsRedundant points Remove redundant points

s

Redundant point

xy

Page 34: University of Denver Department of Mathematics Department of Computer Science

34

Bell and Cover of node Bell and Cover of node xx Remove points inside the Bell- Bell

Elimination Algorithm (BEA)

Page 35: University of Denver Department of Mathematics Department of Computer Science

35

AnalysisAnalysis

Assume points to be uniformly distributed BEA eliminates all the points inside the

disk of radius Need about 75 points Therefore exact solution

2

Page 36: University of Denver Department of Mathematics Department of Computer Science

36

80828486889092949698

100

50 60 75 85 97 140 200Number of Points

% success

Empirical result

Page 37: University of Denver Department of Mathematics Department of Computer Science

37

Distance of one-hop neighbors

Extra region

Page 38: University of Denver Department of Mathematics Department of Computer Science

38

Approximation factorApproximation factor

Page 39: University of Denver Department of Mathematics Department of Computer Science

39

Topics:1. Minimum Disk covering Problem (MDC)

2. Minimum Forwarding Set Problem (MFS)

3. Two-Hop realizability (THP)

4. Exact solution to Weighted Region Problem (WRP)

5. Raster and vector based solutions to WRP Conclusion Questions?

Page 40: University of Denver Department of Mathematics Department of Computer Science

40

Degree of at most 2Degree of at most 2

Two-hop to bipartite graph

s

ab

c

1 23

4

1

2

3

a

b

c

4

Page 41: University of Denver Department of Mathematics Department of Computer Science

41

3 . Two-hop realizability3 . Two-hop realizability

Result:A bipartite graph having a degree of at most 2 is two-hop realizable

1

2

3

a

b

c

4d

5one-hop neighbors two-hop neighbors

Page 42: University of Denver Department of Mathematics Department of Computer Science

42

Topics:1. Minimum Disk covering Problem (MDC)

2. Minimum Forwarding Set Problem (MFS)

3. Two-Hop realizability (THP)

4. Exact solution to Weighted Region Problem (WRP)

5. Raster and vector based solutions to WRP Conclusion Questions?

Page 43: University of Denver Department of Mathematics Department of Computer Science

43

4. Weighted region problem (WRP)4. Weighted region problem (WRP)

Objective - Find an optimal path from START to GOAL Complexity of WRP is unknown

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Page 44: University of Denver Department of Mathematics Department of Computer Science

44

Planar GraphsPlanar Graphs

Planar sub-division considered as planar graph

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Page 45: University of Denver Department of Mathematics Department of Computer Science

45

Shortest path G(V, E, W)

Dijkstra algorithm finds a shortest path from a source vertex to all other vertices

Running time O(|V| log |V| + |E|) Linear time for planar graphs

Page 46: University of Denver Department of Mathematics Department of Computer Science

46

WRP - General case WRP - General case

Notations

f = weight of face f

e = weight of edge e, where e = f f’ ≤ min {f, f’}

A weight of implies A path cannot cross that face or edge

Note that all optimal paths must be piecewise linear!!

Page 47: University of Denver Department of Mathematics Department of Computer Science

47

t ( x1,y1)

e

f (f)

f′ (f′)

s ( 0,-y0)

c (x, 0)

θ

θ′

f ≥ f′ > 0

Snell’s LawSnell’s Law

Cost function

f sinθ = α f ' sinθ '

Optimal point of incidence

Page 48: University of Denver Department of Mathematics Department of Computer Science

48

0/1/0/1/ Special case WRP Special case WRP

v wR

Construct a critical graph G Run Dijkstra on G

Weight 0

v w R

Weight

Page 49: University of Denver Department of Mathematics Department of Computer Science

49

Convex Polygon C

Exact path when s in C and t is arbitrary Construct “Exact Weighted” Graph Add edges that contribute to exact path Run Dijkstra shortest path Algorithm

Page 50: University of Denver Department of Mathematics Department of Computer Science

50

Critical points

θC

Critical edges

s

t

Page 51: University of Denver Department of Mathematics Department of Computer Science

51

Snell points

θ1

x

s

t

Page 52: University of Denver Department of Mathematics Department of Computer Science

52

Border points

t

s

Page 53: University of Denver Department of Mathematics Department of Computer Science

53

Topics:1. Minimum Disk covering Problem (MDC)

2. Minimum Forwarding Set Problem (MFS)

3. Two-Hop realizability (THP)

4. Exact solution to Weighted Region Problem (WRP)

5. Raster and vector based solutions to WRP Conclusion Questions?

Page 54: University of Denver Department of Mathematics Department of Computer Science

54

5. WRP - General case

-optimal path

• -optimal path from s to t is specified by users• path within a factor of (1+ ) from the optimal

Page 55: University of Denver Department of Mathematics Department of Computer Science

55

Transform weighted planar graph to uniform rectangular grid Make a graph with nodes and edges

- nodes : raster cells

- edges : the possible paths between the nodes Find the optimal path by running Dijkstra’s algorithm

Raster-based algorithms

8 connected 16 connected 32 connected

Page 56: University of Denver Department of Mathematics Department of Computer Science

56

Raster-based algorithms…

Advantages- Simple to

implement- Well suited for

grid input data- Easy to add other

cost criteria

Drawbacks- Errors in distance

estimate, since we measure grid distance instead of Euclidean distance

- Error factor :

4-connectivity:√2

8-connectivity:(√2+1)/5

Page 57: University of Denver Department of Mathematics Department of Computer Science

57

Distortions - bends in raster paths

Page 58: University of Denver Department of Mathematics Department of Computer Science

58

Approximate by a straight lineApproximate by a straight line

Reduce deviation errors

Page 59: University of Denver Department of Mathematics Department of Computer Science

59

Compare vector vs. Raster

Raster 1178.68 50 secs

Straight 1130.56 65 secs

Vector(=.1)

1128.27 < 1 sec

Page 60: University of Denver Department of Mathematics Department of Computer Science

60

Topics:1. Minimum Disk covering Problem (MDC)

2. Minimum Forwarding Set Problem (MFS)

3. Two-Hop realizability (THP)

4. Exact solution to Weighted Region Problem (WRP)

5. Raster and vector based solutions to WRP Conclusion Questions?

Page 61: University of Denver Department of Mathematics Department of Computer Science

61

ConclusionConclusion

Improved approximation to MDC Bell elimination algorithm Two-hop realizability Exact solutions to special cases of WRP Straight optimal raster paths

Page 62: University of Denver Department of Mathematics Department of Computer Science

62

Topics:1. Minimum Disk covering Problem (MDC)

2. Minimum Forwarding Set Problem (MFS)

3. Two-Hop realizability (THP)

4. Exact solution to Weighted Region Problem (WRP)

5. Raster and vector based solutions to WRP Conclusion Questions?