Realtime Watermarking Techniques for Sensor Networks

1
Real-time Watermarking Techniques for Sensor
Networks
Jessica Feng, Miodrag Potkonjak
Computer Science Dept., University of California,
Los Angeles
IST/SPIEs EI, Jan 2003
Santa Clara, California
jessicaf_at_cs.ucla.edu
2
Architecture of Sensor Networks
Power supply Sensor
Processor Memory Radio Actuator
jessicaf_at_cs.ucla.edu
3
Why Sensor Networks?

• Marine Microorganisms

• Contaminant Transport Monitoring

• Habit Sensing Array

• Seismic Monitoring

Bridge between the Internet and the physical world
jessicaf_at_cs.ucla.edu
Courtesy to http//www.cens.ucla.edu
4
Watermarking
Watermarking Embeds a secret message into a
cover message
jessicaf_at_cs.ucla.edu
Courtesy to http//www.dribbleglass.com/subpages/
counterfeit.htm
5
1010001010
6
Basic Concept
A
C
B
jessicaf_at_cs.ucla.edu
7
Basic Concept
A
C
B
jessicaf_at_cs.ucla.edu
8
(No Transcript)
9
Example Atomic Trilateration Process
L1 MAD-EAD MBD-EBD MCD-ECD
Measured vs. Estimated L2 (MAD-EAD)2
(MBD-EBD)2 (MCD-ECD)2 ½ L8 max ( (MAD -
EAD)/EAD , (MBD EBD)/EBD , (MCD
ECD)/ECD )   Where EAD (Dx - Ax)2 (Dy -
Ay)2 ½ Estimated distances EBD (Dx -
Bx)2 (Dy - By)2 ½ ECD (Dx - Cx)2 (Dy -
Cy)2 ½
jessicaf_at_cs.ucla.edu
10
Watermark Atomic Trilateration Process
Assigning Weight Factors
001010000100
jessicaf_at_cs.ucla.edu
11
Presentation Organization

• Fundamental Requirements

• Real-time watermarking techniques
• During data acquisition
• During data processing

• Generic watermarking procedure
• Example trajectory motion

• Experimental results and comparisons of different
  watermarking techniques and parameters

jessicaf_at_cs.ucla.edu
12
Fundamental Requirements

• Quality of solution before vs. after watermarking

• Low overhead

• Runtime

• Robustness

• Resilience against attacks

• No significant change in meaning or function of
  the original data

jessicaf_at_cs.ucla.edu
13
New Idea Real-time Watermarking Techniques

• Impose additional constraints during data
  capturing or sensor data processing as suppose to
  post-processing watermark techniques

Strength of authorship
Accuracy
jessicaf_at_cs.ucla.edu
14
Where Can Watermarking Take Place?
jessicaf_at_cs.ucla.edu
15
Watermarking Technique 1
During Data Acquisition

• Impose constraints on intrinsic conditions such
  as
• Location
• Orientation on sensors
• Time management
• Resolution
• Intentional addition of obstacles
• Use of actuators

• Where?
• Each individual sensor node
• Properly selected collection of nodes

jessicaf_at_cs.ucla.edu
16
Watermarking Technique 1
During Data Acquisition

• Impose constraints on intrinsic conditions such
  as
• Location
• Orientation on sensors
• Time management
• Resolution
• Intentional addition of obstacles
• Use of actuators

• Where?
• Each individual sensor node
• Properly selected collection of nodes

jessicaf_at_cs.ucla.edu
17
Watermarking Technique 1
During Data Acquisition

• Impose constraints on intrinsic conditions such
  as
• Location
• Orientation on sensors
• Time management
• Resolution
• Intentional addition of obstacles
• Use of actuators

• Where?
• Each individual sensor node
• Properly selected collection of nodes

jessicaf_at_cs.ucla.edu
18
Watermarking Technique 1
During Data Acquisition

• Impose constraints on intrinsic conditions such
  as
• Location
• Orientation on sensors
• Time management
• Resolution
• Intentional addition of obstacles
• Use of actuators

• Where?
• Each individual sensor node
• Properly selected collection of nodes

jessicaf_at_cs.ucla.edu
19
Watermarking Technique 2
During Data Processing

• Three degrees of freedom
• Error minimization procedures
• Maximal consistency vs. strong strength of
  signature
• Physical world model building
• Additional constraints during the model building
• Solving computationally intractable problems
• NP-completeness
• High quality solution vs. strength of the
  signature

jessicaf_at_cs.ucla.edu
20
Generic Multi-sensor Fusion Model
jessicaf_at_cs.ucla.edu
21
Generic Watermarking Procedure
jessicaf_at_cs.ucla.edu
22
Trajectory Motion

• Known values
• Velocity, acceleration, angle, time interval,
  measured distances, coordinates of 3 nodes

• Unknown variables
• Trajectory path (coordinates of node at each
  discrete time instance)

jessicaf_at_cs.ucla.edu
23
Watermark Trajectory Motion
Time instance 2

• System of non-linear equations
• dobj,a ((xa xobj,0)2 (ya yobj,0)2)1/2
• dt0?t1 (Vobj,0) ?t aobj,0/2 (?t)2
• Vobj,1 Vobj,0 (aobj,0) ?t
• xobj,1 (dt0?t1) cos(?obj,0) xobj,0
• yobj,1 (dt0?t1) sin(?obj,0) yobj,0

• Error associated with unknown variables
• ?1 xobj,0
• ?2 yobj,0
• ?3 xobj,1
• ?4 yobj,1
• ?5 xobj,2
• ?6 yobj,2

jessicaf_at_cs.ucla.edu
24
Watermark Trajectory Motion

• Rewrite system of equations with errors

• OF min ?1 ?2 ?3 ?4 ?5
  ?6
• st dobj,a ((xa xobj,0) (ya
  yobj,0))1/2
• dt0?t1 (Vobj,0) ?t aobj,0/2 (?t)
• Vobj,1 Vobj,0 (aobj,0) ?t
• xobj,1 (dt0?t1) cos(?obj,0) xobj,0
• yobj,1 (dt0?t1) sin(?obj,0)
  yobj,0
• xobj,0 Exobj,0 ?1
• yobj,0 Eyobj,1 ?2
• xobj,0 Exobj,0 ?3
• yobj,1 Eyobj,1 ?4
• xobj,0 Exobj0 ?5
• yobj,1 Eyobj,1 ?6

jessicaf_at_cs.ucla.edu
25
Watermark Trajectory Motion

• Impose additional constraints/ Alter the OF

• OF min ?1 ?2 ?3 ?4 ?5
  ?6
• st dobj,a ((xa xobj,0) (ya
  yobj,0))1/2
• dt0?t1 (Vobj,0) ?t aobj,0/2 (?t)
• Vobj,1 Vobj,0 (aobj,0) ?t
• xobj,1 (dt0?t1) cos(?obj,0) xobj,0
• yobj,1 (dt0?t1) sin(?obj,0)
  yobj,0
• xobj,0 Exobj,0 ?1
• yobj,0 Eyobj,1 ?2
• xobj,0 Exobj,0 ?3
• yobj,1 Eyobj,1 ?4
• xobj,0 Exobj0 ?5
• yobj,1 Eyobj,1 ?6

1?1 4?2 2?3
jessicaf_at_cs.ucla.edu
26
Watermark Trajectory Motion

• Impose additional constraints/ Alter the OF

• OF min
• st dobj,a ((xa xobj,0) (ya
  yobj,0))1/2
• dt0?t1 (Vobj,0) ?t aobj,0/2 (?t)
• Vobj,1 Vobj,0 (aobj,0) ?t
• xobj,1 (dt0?t1) cos(?obj,0) xobj,0
• yobj,1 (dt0?t1) sin(?obj,0)
  yobj,0
• xobj,0 Exobj,0 ?1
• yobj,0 Eyobj,1 ?2
• xobj,0 Exobj,0 ?3
• yobj,1 Eyobj,1 ?4
• xobj,0 Exobj0 ?5
• yobj,1 Eyobj,1 ?6

0?1 2?2 2?3 0?4 1?5 0?6
jessicaf_at_cs.ucla.edu
27
Watermark Trajectory Motion

• Impose additional constraints/ Alter the OF

• OF min ?1 ?2 ?3 ?4 ?5
  ?6
• st dobj,a ((xa xobj,0) (ya
  yobj,0))1/2
• dt0?t1 (Vobj,0) ?t aobj,0/2 (?t)
• Vobj,1 Vobj,0 (aobj,0) ?t
• xobj,1 (dt0?t1) cos(?obj,0) xobj,0
• yobj,1 (dt0?t1) sin(?obj,0)
  yobj,0
• xobj,0 Exobj,0 0?1
• yobj,0 Eyobj,1 2?2
• xobj,0 Exobj,0 2?3
• yobj,1 Eyobj,1 0?4
• xobj,0 Exobj0 1?5
• yobj,1 Eyobj,1 0?6

jessicaf_at_cs.ucla.edu
28
Experimental Results
Watermarking the Atomic Trilateration Process

• 3 Watermarking techniques applied
• Flipping the LSB
• Assigning weight factors
• Hamming distance

• Coordinates limit 0,1

• Coordinates generation
• Uniform distribution on interval 0,1

• Error generation
• Gaussian/Normal distribution (0,1)

jessicaf_at_cs.ucla.edu
29
Experimental Results
Various Resolution and Sigma
Correctness
Authorship Strength
jessicaf_at_cs.ucla.edu
30
Experimental Results
Correctness based on Resolution Before vs. After
Watermarking
Technique 1
Technique 3
Technique 2
jessicaf_at_cs.ucla.edu
31
Experimental Results
Correctness based on Sigma Before vs. After
Watermarking
Technique 1
Technique 3
Technique 2
jessicaf_at_cs.ucla.edu
32
Experimental Results
Correctness 2D vs. 3D
Technique 1
Technique 3
Technique 2
jessicaf_at_cs.ucla.edu
33
Experimental Results
Authorship Strength 2D vs. 3D
 


Technique 1
Technique 3
Technique 2
jessicaf_at_cs.ucla.edu
jessicaf_at_cs.ucla.edu
34
Future Directions

• Fingerprinting

• Watermarking protocols

• Static modeling

• Probabilistic modeling

• Watermarking in other domains

• General intellectual property protection (IPP)
  techniques computational security issues

jessicaf_at_cs.ucla.edu
35
Conclusion

• Sensor network The bridge between physical
  phenomenon and the Internet

• Significant need of Intellectual property
  protection (IPP) and security issues

• Real-time watermarking
• During data capturing
• During data processing

• Preliminary study shows effectiveness

jessicaf_at_cs.ucla.edu