suchisman gangopadhyay montgomery high school, new jersey

30
Planning Troop & Resource Deployments at Military Bases using Exponentially Weighted Voronoi Diagrams Abstract 1106 for WG-16 Presented at 2013 81.1 MORS Conference May 14, 2013 @ 1400 Hrs EDT Suchisman Gangopadhyay Montgomery High School, New Jersey Admitted as Freshman, University of Illinois, Urbana-Champaign [email protected] Suchisman Gangopadhyay 1

Upload: oke

Post on 24-Feb-2016

66 views

Category:

Documents


0 download

DESCRIPTION

Planning Troop & Resource Deployments at Military Bases using Exponentially Weighted Voronoi Diagrams Abstract 1106 for WG-16 Presented at 2013 81.1 MORS Conference May 14, 2013 @ 1400 Hrs EDT. Suchisman Gangopadhyay Montgomery High School, New Jersey - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

1

Planning Troop & Resource Deployments at Military Bases using Exponentially WeightedVoronoi DiagramsAbstract 1106 for WG-16 Presented at 2013 81.1 MORS ConferenceMay 14, 2013 @ 1400 Hrs EDT

Suchisman GangopadhyayMontgomery High School, New JerseyAdmitted as Freshman, University of Illinois, [email protected]

Page 2: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

2Outline

• Premise• Troop Deployment and Voronoi• Insurgent Activity and Weighted Voronoi• Boulding’s Loss of Strength Gradient and Exponential

Weightage• Method• Observations & Results• Next steps• References

Page 3: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

3Premise

Page 4: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

4

Deciding on how many resources to send to a military base is a

time consuming process

Due to the unpredictable nature of war, reducing the time this process takes can win battles and save lives

Page 5: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

5

Assuming uniform land structure, resources can be deployed based on how

much land each base must administer

One of the methods of determining these areas is to

use Voronoi Diagrams

Page 6: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

6

Voronoi Diagrams tessellate a plane into sections based on how close parts of the plane are to certain points, called sites

A cell is formed around each site, consisting of the points which are closest to that site. Points on cell walls are equally close to two sites, while points

on vertices are equally close to three or more sites

Page 7: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

7

The positions of the bases can be used as sites to create a

Voronoi diagram

Resources can then be distributed in the ratio of areas

of the sites

Page 8: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

8

Unfortunately, this method does not account for the

uneven presence of insurgents

A base in a safer area may receive the same amount of resources as a base that has to protect regions that are constantly

under attack

Page 9: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

9

The use of a weighted Vornonoi can account for this

If a normal Voronoi uses the simple distance formula, then a weighted

Voronoi uses a weight to modify that distance

Page 10: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

10

One frequent type of distance formula often substituted for the regular distance

formula is “Manhattan distance”

Other types of weighting have each site holding a different weightage, which is

then added, multiplied, etc. to a distance between some point and that site

Page 11: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

11

In 1962, Kenneth Boulding devised a loss of strength

gradient

He argued that the strength of a military force was related to the distance that force had to travel

Page 12: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

12

The loss of strength gradient can be expressed as:

Due to the use of exponents in this gradient, it makes sense for the

weightage used in this Voronoi to be exponential

log[ (10 )]_ _miles e

miles per dayAdjustedPower Power

Page 13: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

13Method

Page 14: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

14

Step 1: Plot the coordinates of the bases on a Cartesian plane (represented by blue stars)

Page 15: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

15

Step 2: Use a Voronoi tessellation to find the land each base administers

Page 16: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

16

Step 3: Add locations of insurgent activity to the Voronoi (represented by orange dots

Page 17: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

17Step 4: Find the concentration of enemy activity in each cell

Page 18: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

18

Step 5: Use these concentrations as weightages to create an exponentially weighted Voronoi

Page 19: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

19

Observations & Results

Page 20: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

20

Step 6: Distribute resources using the ratios of the area of the weighted cells

Site Coordinates

Site Metrics

Area of Site Insurgent DensityWeighted

Voronoi Ratio

Recommended

Allocation of

Resources

A: (9,7) 16.866 0.474 1102 a 13,638

B: (10,7) 29.969 1.334 15875 196,469

C: (6,6) 28.459 0.808 2652 32,821

D: (3,6) 38.375 1.251 15944 197,322

E: (5,8) 29.333 0.954 4828 59,751

TOTAL 144 N.A. 40401 500,001

a. Ratio is calculated as the number of points plotted in each weighted Voronoi cell by MATLAB ®.

Page 21: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

21

This method results in more resources being given to bases in the middle of enemy activity

While still accounting for how much land each base must

administer

Page 22: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

22

When making an exponentially weighted Voronoi diagram, there are a few things one must look out for

Page 23: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

23

First, exponentially weighted Voronoi diagrams differ by scale

Therefore, it is important to be consistent with units when constructing the diagram

Page 24: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

24

Next, there are some strange structures one may see when

creating an exponential Voronoi

A cell seeming like it is subsumed within another may

actually be much bigger

Page 25: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

25Next Steps

Page 26: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

26

Exponential weightage is only the beginning

By using the deployment of resources in successful operations as an example, the formula for weightage can be adjusted to

replicate those results

Page 27: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

27

Another way this method can become more accurate and versatile is to account

for natural formations and enemy structures

Instead of starting with a regular Voronoi, the land each base administers could be

split based on a Voronoi weighted to account for uneven territory

Page 28: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

28

Lastly, while simple and additively weighted Voronoi diagrams can be constructed efficiently, other types of weighted Voronoi diagrams can

take much longer to create

The quicker the algorithm for creating these weighted Voronoi diagrams is, the

quicker one can respond to new information

Page 29: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

29Q & A

• Thanks!

Page 30: Suchisman Gangopadhyay Montgomery High School, New Jersey

Suchisman Gangopadhyay

30References

I. K. Boulding, Conflict and Defense, Harper, New York, 1962, 262.

II. K. Boulding, The Meaning of the 20th Century, The Great Transition, George Allen & Unwin, London, 1965, 87

III. A.G. Boyer, N.J. Gauthier, and C.W. Johnson, The Paradoxes of Military Risk Assessment, In Proceedings of the 25th International Systems Safety Conference, Baltimore, USA, International Systems Safety Society, Unionville, VA, USA, 859-869, 0-9721385-7-9, 2007

IV. B. B. de Mesquita, The War Trap, Yale University Press, New Haven, 1981, 103-108

V. D. Lemke, Regions of War and Peace, Press Syndicate of the University of Cambridge, Cambridge, England, 2002, 71-73

VI. J.D. Vanvactor, Risk Mitigation Through a Composite Risk Management Process, The U.S. Army Risk Assessment Organization Development Journal Vol. 25 Nbr. (2007)

VII. K. Webb, The Continued Importance of Geographic Distance and Boulding's Loss of Strength Gradient, Comparative Strategy, Volume 26 Issue 4, 2007, 295-310