WiPrint 3D Printing Your Wireless Coverage Justin Chan, Changxi - PowerPoint PPT Presentation

WiPrint 3D Printing Your Wireless Coverage Justin Chan, Changxi Zheng*, Xia Zhou Dartmouth College, *Columbia University 1

Problem 1 Rx Tx Performance 2

Problem 2 Security 3

Problem Statement -35 -40 -45 -50 -55 -60 -65 -70 -75 RSS (dBm) How to create custom signal maps? 4

Existing Solution Expensive (~$9000) Bulky Cumbersome h"p://x.net.au/wp.content/ uploads/2007/09/yagi.jpg< 5

Solution: Soda Cans? 6

Solution: 3D Printed Reflectors Cheap (~$50) Lightweight Flexible 7

Fabrication 8

Fabrication Concave reflector Square-wave reflector 9

Fabrication 10

Search for optimal reflector shape Fabricate 11

Roadmap Modeling Compute reflector Evaluate results 12

Roadmap Modeling Compute reflector Evaluate results 13

Ray Tracer Input: • Environmental layout • Router location • Reflector shape 14

Ray Tracer Output: Predicted signal map -35 -40 -45 -50 -55 -60 -65 -70 -75 RSS (dBm) 15

Propagation Model • Site-specific models • Solving wave equations 16

Propagation Model Di f raction Specular Transmission tion Re fm ection Θ i Θ r Θ i Θ r 17

Roadmap Modeling Compute Reflector Evaluate results 18

Shape Of A Reflector Set of points Interpolated to polynomial 19

Objective Function For High-Res Input Desired Ray tracer Predicted Reflector • Mean absolute error • Find reflector to optimize fitness function 20

Simulated Annealing • Randomized algorithm for global optimization h"p://www.novinite.com/media/images/2011.04/photo_verybig_127084.jpg< 21

Low-Res Map 22

Scoring Low-Res Maps • Red cells – The higher the signal, the better the red score • Black cells – The lower the signal, the better the black score • Maximize both red and black scores http://gymnastjudge.com/wp-content/uploads/2012/01/scoring42.png 23

Pareto Optimal Solution Pareto Optimal Solution Black score Red Red score A point where we cannot improve a one objective function without causing a decrease in another objective function 24

Pareto Simulated Annealing • Give weights to objective functions • Maximize weighted sum • Vary weights, allows exploration along the frontier Black score Red Red score 25

Roadmap Modeling Compute Reflector Evaluate results 26

With reflector Without reflector -35 -40 -45 -50 -55 -60 -65 -70 -75 27 RSS (dBm)

Baseline Measurements Measured Measured 14 14 y (m) y (m) 7 7 0 0 0 5 10 15 20 25 0 5 10 15 20 25 Estimated Estimated 14 14 y (m) y (m) 7 7 0 0 0 5 10 15 20 25 0 5 10 15 20 25 x (m) x (m) -35 -40 -45 -50 -55 -60 -65 -70 -75 RSS (dBm) Router in center Router in corner 28

Reflectors Measured Measured 14 14 y (m) y (m) 7 7 0 0 0 5 10 15 20 25 0 5 10 15 20 25 Estimated Estimated 14 14 y (m) y (m) 7 7 0 0 0 5 10 15 20 25 0 5 10 15 20 25 x (m) x (m) -35 -40 -45 -50 -55 -60 -65 -70 -75 RSS (dBm) Concave reflector Square wave reflector 29

Future Work • Other waves frequencies • Different reflector materials • More advanced 3D ray tracer • System of multiple reflectors h"p://www.horsesforsources.com/wp.content/uploads/2011/05/The.present.of.work.jpg< 30

Intuitive and novel solution to both performance and security problems Questions? 31

5GHz Router 2.4GHz 2.4GHz 14 14 -40 -40 -40 -40 y (m) y (m) 7 7 -50 -50 -50 -50 0 0 RSS (dB) RSS (dB) RSS (dB) RSS (dB) 0 5 10 15 20 25 0 5 10 15 20 25 5GHz 5GHz -60 -60 -60 -60 14 14 y (m) y (m) 7 7 -70 -70 -70 -70 0 0 0 5 10 15 20 25 -80 -80 0 5 10 15 20 25 -80 -80 x (m) x (m) No reflector Concave reflector

Simulator Error 1 1 0.8 0.8 CDF CDF 0.6 0.6 Ray tracing Ray tracing 0.4 0.4 Ray w/o diffraction Ray w/o diffraction 0.2 0.2 ITU ITU Uniform Uniform 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Error (dBm) Error (dBm) TX in center TX in corner

Simulator Error 1 1 0.8 0.8 CDF CDF 0.6 0.6 Ray tracing Ray tracing 0.4 0.4 Ray w/o diffraction Ray w/o diffraction 0.2 0.2 ITU ITU Uniform Uniform 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Error (dBm) Error (dBm) Concave Square-wave reflector reflector