The Cat and The Mouse -- The Case of Mobile Sensors and Targets
Download
Report
Transcript The Cat and The Mouse -- The Case of Mobile Sensors and Targets
Game-Theoretic Analysis of Mobile
Network Coverage
David K.Y. Yau
Outline
Introduction
Mobility models
Cats’ strategies
Mouse’s strategies
Experimental results
Conclusion
Motivation – Why Mobile?
The mouse
Evade detection
Nature of “mission”
The cat
Improved coverage with fewer sensors
Robustness against contingencies
Problem Formulation
Two player game—cats and mouse
Closed rectangular area
Cats try to shorten detection time
Mouse tries to lengthen detection time
Both move at constant speed
Both have finite sensing range
Ends when mouse is within cat’s sensing range
Mobility Model
Four-tuple <N, M, T, R>
N: network area
M: accessibility constraints
-- the “map”
T: trip selection
R: route selection
Random waypoint model is a special case
Null accessibility constraints
Uniform random trip selection
Cartesian straight line route selection
The Sensing Range
Cats’ sensing range Rc
Mouse’s sensing range Rm
Blind mouse
Rm < Rc
Caught before evasion
Seeing mouse
Rm > Rc
Active evasion possible
Cats’ Strategies
Uniform scan
Bouncing
Random waypoint model
Mouse Strategy
Blind
Hide at safe haven
Assume cats’ presence statistics is known
Seeing
Cats presence Run
Maximize the minimum distance to all cats
Cats absence
Bouncing
Random waypoint model
Static Don’t move
The Presence Matrix – ∏
Probability of a cat presence at an area
Divide network area into m × n cells
∏i,j = Probability of one or more cats
present in cell (i, j)
Best Blind Mouse Play
Find an optimal path to move to a safest cell
Detection time is maximized along the path
∏i,j is lowest at the safest cell (usually)
Dynamic programming
Greedy does not always work
Comparison with Local Greedy Strategy
0.0003
0.0300
Greedy
0.0003
0.0300
Dynamic Programming
Optimal Escape Path Formulation
Using ∏, computes
Ej[Tstay] = Expected detection time if staying at cell i
Ej[Tmove(k)] = Expected detection time if moving to cell k
Cell k is a neighboring cell of i
Make decision—stay or move
Maximize expected detection time
Optimal escape path = sequence of movement until
stay is chosen
How to compute the expected detection time?
Compute Expected Detection Time
Initialize Ej[Tdetect] as Ej[Tstay]
Insert all the cells into max-heap
i := Extract-max
Update Ej[Tdetect] for each neighbor cell k of i
Ek[Tdetect] := max(Ek[Tdetect], Ei[Tmove(k)])
Heapify
Repeat until heap becomes empty
Example Optimal Paths
0.0000
0.0374
Vm = 10 m/s
0.0000
0.0374
Vm = 15 m/s
Blind Mouse Strategies Compared
Expected
Detection Time
DP
Mouse
RWP
Strategies
Stay
Cat Strategies
Scan
Bouncing
RWP
1083.26
628.66
2823.26
415.31
442.23
271.73
511.50
305.03
226.13
Vc = 10 m/s, Vm = 10 m/s, Rc = 25 m, Rm = 0 m
Other Options for Cats
Increase sensing range
Rc (m)
Tdetect (s)
1
15136
5
3056
15
948
25
585
50
271
20
1253
40
640
80
245
160
130
2
278
10
45
50
6
100
2
Increase speed
Vc (m/s)
Tdetect (s)
10
2618
Increase quantity
Nc
Tdetect (s)
1
573
Best Seeing Mouse Play
Find and move at the optimum direction
Minimum distance to all cats is maximized
Distance between cat and mouse
d(β, t) = ║C(t) – M(β, t)║
Minimum distance moving at direction β
d*(β) = min t ≥ 0 { d(β, t) }
Optimal escape direction
β * = argmax d*(β)
Strategies When Cat Absence
Bouncing
Centric
Random waypoint model
Static Don’t move
Seeing Mouse Strategies Compared
Expected
Detection Time
Mouse
Strategies
Cat Strategies
Bouncing
RWP
Bouncing
149.53
1455.28
Centric
340.85
1092.29
Static
92.39
899.07
Stay
10.23
21.99
Vc = 10 m/s, Vm = 10 m/s, Rc = 5 m, Rm = 10 m
Result Explained
Why Bouncing is better for cats?
All area are equally likely to be visited (approx.)
Uniform presence matrix (approx.)
Safe haven eliminated
Why Centric is better for mouse?
More choices of direction
Presence Matrices
Random Waypoint Model
Bouncing
Where The Mouse Were Caught?
0.0
1543.2
Detection
Time (s)
Other Options—Sensing Range
Other Options—Speed
Other Options—Quantity
Conclusions
Detect intelligent mobile target using mobile sensor
Mobile sensors increase robustness
Strategies to evade detection without full knowledge of cat
movement
Movement model is important
Presence matrix determine the coverage performance
Bouncing movement is better than random waypoint model
Stochastic movement prevent movement prediction
Optimal escape direction helps seeing mouse
Dynamic programming algorithm helps blind mouse
Effects of sensing range, speed and number of cats are
quantified
Future Work
Radioactive, chemical plume detection
Explosion
Dispersion
Mobile target detection with presence of obstacle
Model for sensor reliability, interference, etc.
Quantification of sensing uncertainty
Question…