1. Relative Accuracy based Location Estimation in Wireless Ad Hoc Sensor Networks May Wong 1 Demet Aksoy 2 1 Intel, Inc. 2 University of California, Davis ICC 2007

2. Outline Introduction Related Work QUAD ( Quadrant-Based estimation) Localization algorithm Performance evaluation Conclusions

3. Introduction In WSNs, Sensors are used in a wide range of applications. Ex. scientific research, military, healthcare, and environmental monitoring. Every user has to depend on the location provided by the sensor to analyze observations Location information is important for data analysis.

4. Introduction Sensor are not known where were they deployed. There are two ways to get sensor’s position by themselves. Equip GPS. High Cost, Big size and power consumption Localization algorithm. Location errors are inevitable in estimations. The precise location of each sensor is not necessary in most sensor network applications.

5. Introduction A B Observer Considers that pollution is from B to A. B’ A’

6. Introduction A B B’’ A’’ Observer Considers that pollution is from A to B, but real condition is from B to A.

7. Introduction Motivation Reduce hardware cost. Previous work in localization focus on individual accuracy position. Goal Minimal Specialized Hardware Eliminate error relative location. Robustness to Network Density.

8. Related Work A,(x 1,y 1 ), Hop 2 B,(x 2,y 2 ), Hop 4 C,(x 3,y 3 ), Hop 6 A(x1,y1)A(x1,y1) B(x2,y2)B(x2,y2) C(x3,y3)C(x3,y3) X (C,(X c,Y c), Hop Count) Reference Node Unlocalized Node Localized Node

9. Related Work X Reference Node Unlocalized Node Localized Node 50m 60m 100m A(x1,y1)A(x1,y1) B(x2,y2)B(x2,y2) C(x3,y3)C(x3,y3) A,(x 1,y 1 ), Hop 2 B,(x 2,y 2 ), Hop 4 C,(x 3,y 3 ), Hop 6

10. Related Work DV-Hop could not estimate a position.

11. Quad Localization algorithm Phase1 Hop distance dissemination DV Based Phase2 Position vote Determine sensor relative location Phase3 Location estimation Determine location

12. Quad Localization algorithm Hop distance dissemination A,(x 1,y 1 ), Hop 2 B,(x 2,y 2 ), Hop 4 C,(x 3,y 3 ), Hop 6 V(x1,y1)V(x1,y1) U(x2,y2)U(x2,y2) T(x3,y3)T(x3,y3) A (C,(X c,Y c), Hop Count)

13. Quad Localization algorithm Position vote A,(x 1,y 1 ), Hop 2 B,(x 2,y 2 ), Hop 4 C,(x 3,y 3 ), Hop 6 V(x1,y1)V(x1,y1) U(x2,y2)U(x2,y2) T(x3,y3)T(x3,y3) A (C,(X c,Y c), Hop Count) Near Set: A Far Set: B,C

14. Quad Localization algorithm Location estimation Ｂ Ｃ Ａ W Z Y (49,49) (50,48) (50,50) Near Set Y,(50,50),3 Far Set W,(49,49),5 Z,(50,48),5 Near Set W,(49,49),1 Far Set Y,(50,50),3 Z,(50,48),3 N W North South North South

15. Performance evaluation Simulator is implemented by C++ Radio range 5 unit 100 x 100 grid size Random number of nodes Different topology

16. Performance evaluation Smooth (1) Hop by hop gradient propagation to get estimated distance to reference node. (2) Local computation by each node using multilateration procedure. Organizing a Global Coordinate System from Local Information on an Ad Hoc Sensor Network. In Proc. 2nd Intl. Workshop on Information Processing in Sensor Networks (IPSN)

17. b ca a a Performance evaluation DV-Hop Min-Max : corrections are done by local averaging The n-Hop Multilateration Primitive for Node Localization Problems (ACM) Mobile Networks and Applications 03 b+c X Y V Set the center of the bounding box as the estimate.

18. Performance evaluation

19. X Coordinate Estimates using DV-Hop

20. Performance evaluation X-Coordinate Estimates using Min-Max

21. Performance evaluation X Coordinate Estimates using QUAD

22. Conclusions In this paper, they use coordinate and relative position to estimate location. provide a accuracy topology.

23. Thank You!