Slides - University of Virginia

advertisement
Object Localization Using RFID
Kirti Chawla
Department of Computer Science
University of Virginia
Outline
•
•
•
•
•
•
•
The Problem of Locating Objects
Research Milestones
Background
Motivation
Proposed Approach
Experimental Evaluation
Conclusion
1/23
Problem
Locating Objects
Locate
Objects
Environments
Goal: Find positions of objects in an environment
Hypothesis: Standard RFID is sufficient and effective
Key-factors: Performance, applicability and shortcomings
2/23
Milestones
Research Deliverables
Journal Publication:
•
Kirti Chawla, and Gabriel Robins, An RFID-Based Object Localization Framework,
International Journal of Radio Frequency Identification Technology and Applications,
Inderscience Publishers, 2011, Vol. 3, Nos. 1/2, pp. 2-30
Conference Publications:
•
•
Kirti Chawla, Gabriel Robins, and Liuyi Zhang, Efficient RFID-Based Mobile Object
Localization, Proceedings of IEEE International Conference on Wireless and Mobile
Computing, Networking and Communications, 2010, Canada, pp. 683-690
Kirti Chawla, Gabriel Robins, and Liuyi Zhang, Object Localization using RFID,
Proceedings of IEEE International Symposium on Wireless Pervasive Computing, 2010,
Italy, pp. 301-306
Patent:
•
Kirti Chawla, and Gabriel Robins, Object Localization with RFID Infrastructure, US Patent
Application Number: PCT/US2011/053067, filed with WIPO/USPTO September 2011
Copyright:
•
Kirti Chawla, and Gabriel Robins, An RFID-Based Object Localization Framework, US
Copyright Case Number: 1-633487801, 2011
Startup Venture:
•
•
•
Co-founded Diorama Technologies LLC, in partnership with private investors
Raised venture capital funding and negotiated licensing terms
Other investors have shown strong interest in commercializing our ideas
Patent
WIPO/USPTO
Object Localization with RFID Infrastructure
Background
Current State of the Art
Mismatched
Solutions
Technologies
Techniques
Limiting
Constraints
3/23
Background
RFID Primer
RFID Reader
RFID Tag
Far-field Propagation
Near-field
Propagation
Readers: Variety of form-factors and frequencies
Tags: Flexible power source, frequency, and form factors
4/23
Motivation
Why locate objects using RFID ?
Dark
Environment
No Line of
Sight
Cost Effective
Solid
Obstacles
Natural Fit
Adaptive
5/23
Approach
Power-Distance Relationship
Comparison
N
Tag Power
 Wavelength 
 Tag Gain ×Reader Gain × 

Reader Power
4×
π
×Distance


Reader Power
Distance
Tag Power
Problem: Radio variability renders Friis equation
practically useless
Insight: Utilize empirical power-distance relationship
6/23
Approach
Empirical Power-Distance
Relationship
Antenna
Shared Region
Radio Wave
Insight: Similarly behaving tags are close to each other
7/23
Approach
Tag Sensitivity Characterization
Results
Key Challenges
13%
Pile of Tags
25%
54%
8%
High Sensitive
Average Sensitive
Low Sensitive
Problem: Tags have variable sensitivities / performance
Insight: Bin tags based on their sensitivity
8/23
Approach
Reliability through Multi-Tags
Results
Vertical
Parallel
RFID Reader
Platform
Side View
Horizontal
Orthogonal
Platform
RFIDTop
TagView
Problem: Optimal tag reads occur at certain orientations
Insight: Multi-tags provide orientation redundancy
9/23
Approach
Tag Localization Approach
Signal Strength
Metric: MTDP
Localization
Phase
Setup Phase
10/23
Approach
0
Tag Localization Algorithms
MID
MAX
Reader Output Power Range
Linear Search
Binary Search
Parallel Search
O(#Tags 
#Power-Levels)
O(#Tags 
Log#Power-Levels)
O(#Power-Levels)
11/23
Approach
Reader Localization Approach
Localization
Phase
Setup Phase
12/23
Approach
Localization Error
Heuristics
Reference Tags
Target Tag
Problem: Assumption that target and reference tag
locations coincide leads to localization error
Insight: Consider other nearby reference tags
in order to minimize the localization error
13/23
Evaluation
Track Design
Experimental Setup
Robot Design
14/23
Evaluation
Back
Empirical Power-Distance
Relationship
Insight: Only empirical power-distance relationship can
provide high localization performance
120.00
Theoretical (N = 6)
Power (dBm)
100.00
80.00
60.00
Theoretical (N = 3)
40.00
Empirical
20.00
Theoretical (N = 2)
0.00
Distance (Meters)
15/23
Evaluation
4
Euclidean Distance (Meters)
3.5
Localization Accuracy
Insight: Performance can be improved by denser
reference tag deployment
Actual Position
3
2.5
2
Inferred Position
1.5
1
0.5
0
Location Measurement #
16/23
Evaluation
Localization Time
Insight: Faster algorithms provide lower tag
detectability
25.00
Average Time (Seconds)
Linear Search (LH)
20.00
15.00
Parallel Search
10.00
5.00
Binary Search
Linear Search (HL)
0.00
Distance from Antenna (Meters)
17/23
Evaluation
Localization Performance
Vs #Tags
Average Localization Error (Meters)
Insight: Localization performance varies with tag density
1.6
1.4
1.2
1
0.8
0.6
0.4
Diminishing
returns
0.2
0
Number of Reference Tags
18/23
Evaluation
Comparative Evaluation
Average Time (minutes)
Approach
Setup Phase Localization
Phase
Ni et al., 2003
-
-
Alippi et al., 2006
-
-
Bekkali et al., 2007
-
-
Senta et al., 2007
-
-
Wang et al., 2007
-
Zhang et al., 2007
Test Area
(m2)
-
Localization
Error (m)
2
20
0.68
9
0.5 – 1.0
Pure RFID
2
0.2
-
-
0.1 – 0.9
-
-
-
Seo and Lee, 2008
-
-
Choi and Lee, 2009
-
-
Choi et al., 2009
-
-
-
Joho et al., 2009
27
-
-
161.23
29.78
47.24
1.67
0
161.23
5.28
1.42
1.95
1.67
0
10.32
Linear Search (LH)
Linear Search (HL)
Binary Search
Parallel Search
Measure and Report
Combined Approach
Notes
1
Hybrid
5
0.2 – 1.6
Pure RFID
14.4
0.02
Hybrid
8
0.21
0.38
0.27
0.29
0.31
0.35
0.25
0.18
Pure RFID
Pure RFID
19/23
Applications
Tier-I Applications
Locating Objects
Warehouses
Hospitals
Supply Chains
Tier-II Applications
Assisted Living
Location-Aware
Services
Energy Saving
in Buildings
Monitor life-critical events
Smart Carts
Provide ground truth
20/23
Interface
Object Location Visualization
21/23
Conclusion
Future Directions
RFID-only Object Localization Framework:
- Showed that pure RFID can be used for object localization
- Introduced a power-distance relationship metric
- Proposed tag binning to mitigate tag sensitivity variability
- Devised effective localization algorithms and heuristics
- Identified / mitigated key localization challenges
Future Research Directions:
- Scalability
- Technology Evolution
- Localization performance
- Visualization tools
- Field testing and Commercialization
22/23
Milestones
Research Deliverables
Journal Publication:
•
Kirti Chawla, and Gabriel Robins, An RFID-Based Object Localization Framework,
International Journal of Radio Frequency Identification Technology and Applications,
Inderscience Publishers, 2011, Vol. 3, Nos. 1/2, pp. 2-30.
Conference Publications:
•
•
Kirti Chawla, Gabriel Robins, and Liuyi Zhang, Efficient RFID-Based Mobile Object
Localization, Proceedings of IEEE International Conference on Wireless and Mobile
Computing, Networking and Communications, 2010, Canada, pp. 683-690.
Kirti Chawla, Gabriel Robins, and Liuyi Zhang, Object Localization using RFID,
Proceedings of IEEE International Symposium on Wireless Pervasive Computing, 2010,
Italy, pp. 301-306.
Patent:
•
Kirti Chawla, and Gabriel Robins, Object Localization with RFID Infrastructure, US Patent
Application Number: PCT/US2011/053067, filed with WIPO/USPTO September 2011.
Copyright:
•
Kirti Chawla, and Gabriel Robins, An RFID-Based Object Localization Framework, US
Copyright Case Number: 1-633487801, 2011.
Startup Venture:
•
•
•
Co-founded Diorama Technologies LLC, in partnership with private investors
Raised venture capital funding and negotiated licensing terms
Other investors have shown strong interest in commercializing our ideas
23/23
Patent
WIPO/USPTO
Object Localization with RFID Infrastructure
Backup Slides
Background
Organizing Localization Space
Localization Solutions
Localization Type
Environmental
Localization
Technique
Signal Strength
Signal Phase
Self
Arrival Time
Approach
Localization Challenges
Back
Radio Interference
Occlusions
Tag Sensitivity
Tag Spatiality
Tag Orientation
Reader Locality
Evaluation
Tag Sensitivity – Single Tag
Back
Constant Distance/Variable Power
140
120
100
80
60
40
20
0
Cumulative Read Count
140
120
100
80
60
40
20
0
Cumulative Read Count
31.6 dBm
Number of Tags
28.6 dBm
Number of Tags
Number of Tags
2.54 Meters,
25.6 dBm
140
120
100
80
60
40
20
0
Cumulative Read Count
Variable Distance/Constant Power
Cumulative Read Count
140
120
100
80
60
40
20
0
Cumulative Read Count
3.81 Meters
Number of Tags
140
120
100
80
60
40
20
0
2.54 Meters
Number of Tags
Number of Tags
1.27 Meters,
31.6 dBm
140
120
100
80
60
40
20
0
Cumulative Read Count
Evaluation
Tag Sensitivity – Multi-Tag
(Proximity)
Back
2.54 Meters,
25.6 dBm
12
10
8
6
4
2
0
Multi-Tag Position
28.6 dBm
Average Read
Count
12
10
8
6
4
2
0
Average Read
Count
Average Read
Count
Constant Distance/Variable Power
12
10
8
6
4
2
0
Multi-Tag Position
31.6 dBm
Multi-Tag Position
1.27 Meters,
31.6 dBm
Multi-Tag Position
12
10
8
6
4
2
0
2.54 Meters
Multi-Tag Position
Average Read
Count
12
10
8
6
4
2
0
Average Read
Count
Average Read
Count
Variable Distance/Constant Power
12
10
8
6
4
2
0
3.81 Meters
Multi-Tag Position
Tag Sensitivity – Multi-Tag
(Rotation-1)
Evaluation
16
14
12
10
8
6
4
2
0
31.6 dBm
Multi-Tag Position
Multi-Tag Position
Average Read
Count
Average Read
Count
Multi-Tag Position
16
14
12
10
8
6
4
2
0
Multi-Tag Position
Average Read
Count
16
14
12
10
8
6
4
2
0
Multi-Tag Position
Average Read
Count
16
14
12
10
8
6
4
2
0
16
14
12
10
8
6
4
2
0
Multi-Tag Position
16
14
12
10
8
6
4
2
0
Multi-Tag Position
16
14
12
10
8
6
4
2
0
Multi-Tag Position
Average Read
Count
Average Read
Count
Average Read
Count
28.6 dBm
16
14
12
10
8
6
4
2
0
Multi-Tag Position
Multi-Tag Position
16
14
12
10
8
6
4
2
0
Average Read
Count
16
14
12
10
8
6
4
2
0
Average Read
Count
2.54 Meters,
25.6 dBm
Average Read
Count
16
14
12
10
8
6
4
2
0
Constant Distance/Variable Power
Average Read
Count
Average Read
Count
Back
Multi-Tag Position
16
14
12
10
8
6
4
2
0
Multi-Tag Position
Tag Sensitivity – Multi-Tag
(Rotation-2)
Evaluation
12
10
8
6
4
2
0
12
10
8
6
4
2
0
3.81 Meters
Average Read
Count
Average Read
Count
Multi-Tag Position
Multi-Tag Position
Multi-Tag Position
12
10
8
6
4
2
0
Multi-Tag Position
Average Read
Count
12
10
8
6
4
2
0
Multi-Tag Position
Average Read
Count
2.54 Meters
12
10
8
6
4
2
0
Multi-Tag Position
12
10
8
6
4
2
0
Multi-Tag Position
12
10
8
6
4
2
0
Multi-Tag Position
Average Read
Count
12
10
8
6
4
2
0
12
10
8
6
4
2
0
Multi-Tag Position
Average Read
Count
Average Read
Count
Multi-Tag Position
Average Read
Count
12
10
8
6
4
2
0
Average Read
Count
1.27 Meters,
31.6 dBm
Average Read
Count
12
10
8
6
4
2
0
Variable Distance/Constant Power
Average Read
Count
Average Read
Count
Back
Multi-Tag Position
12
10
8
6
4
2
0
Multi-Tag Position
Approach
Error-Reducing Heuristics
Back
Absolute
Difference
Localization
Error
Minimum
Power
Selection
Meta
Heuristic
Root Sum
Square
Problem: There can be multiple nearby reference tags
Insight: Select nearby reference tags using different
schemes
Download