1. Localization in Wireless Sensor Networks Shafagh Alikhani ELG 7178 Fall 2008

2. Outline Wireless Sensor Networks Localization – What? Why? Classification of Localization Algorithms Examples of Localization Techniques

3. Wireless Sensor Networks a large number of self-sufficient nodes nodes have sensing capabilities can perform simple computations can communicate with each other

4. Environments of Deployment Indoor vs outdoor Stationary vs mobile 2D vs 3D

5. Localization What? – To determine the physical coordinates of a group of sensor nodes in a wireless sensor network (WSN) – Due to application context and massive scale, use of GPS is unrealistic, therefore, sensors need to self-organize a coordinate system Why? – To report data that is geographically meaningful – Services such as routing rely on location information; geographic routing protocols; context-based routing protocols, location-aware services

6. Problem Formulation Defining a coordinate system Calculating the distance between sensor nodes

7. Defining a Coordinate System Global – Aligned with some externally meaningful system (e.g., GPS) Relative – An arbitrary rigid transformation (rotation, reflection, translation) away from the global coordinate system

8. Classifications of Localization Methods Centralized vs Distributed Anchor-free vs Anchor-based Range-free vs Range-based Mobile vs Stationary

9. Centralized vs Distributed Centralized – All computation is done in a central server Distributed – Computation is distributed among the nodes

10. Anchor-Free vs Anchor-Based Anchor Nodes: – Nodes that know their coordinates a priori – By use of GPS or manual placement – For 2D three and 3D four anchor nodes are needed Anchor-free – Relative coordinates Anchor-based – Use anchor nodes to calculate global coordinates

11. Range-Free vs Range-Based Range-Free – Local Techniques – Hop-Counting Techniques Range-Based – Received Signal Strength Indicator (RSSI) Attenuation RF signal – Time of Arrival (ToA) time of flight – Time Difference of Arrival (TDoA) requires time synchronization electromagnetic (light, RF, microwave) sound (acoustic, ultrasound) – Angle of Arrival (AoA) RF signal

12. Generic Approach Using Anchor Nodes 1. Determine the distances between regular nodes and anchor nodes. (Communication) 2. Derive the position of each node from its anchor distances. (Computation) 3. Iteratively refine node positions using range information and positions of neighboring nodes. (Communication & Computation)

13. Phase 1: Calculating Distance to Anchor Nodes Three algorithms – Sum-dist – DV-Hop – Euclidean Anchors – flood network with their own position

14. Anchors – flood network with own position Nodes – add hop distances – requires range measurement Sum-dist Phase 1: C A B A: 8 8 B: 10+6 = 16 10 6 C: 7+8+6 = 21 8 7

15. Anchors – flood network with own position – flood network with avg hop distance Nodes – count number of hops to anchors – multiply with avg hop distance DV-hop Phase 1: C A B 1 1 1 1 2 2 2 3 3 4 4 A-B: 15 3 hops avg hop: 5

16. Anchors – flood network with own position Nodes – determine distance by 1. range measurement 2. geometric calculation Euclidean Phase 1: C A B

17. Needs high connectivity Error prone (selecting wrong distance) Perfect accuracy possible

18. Phase 2: Determining Position Trilateration – uses multiple distance measurements between known points – Must solve a set of linear equation Triangulation – Law of sines: (sin a)/A=(sin b)/B=(sin c)/C Min-max A B C a b c B A C

19. Phase 2: Min-max Distance to anchors determines a bounding box Center of box estimates node position A B C

20. Phase 3: Iterative refinement Node obtains initial position (phase 1 and 2) Node broadcasts its position Position is refined iteratively using: – distances to neighbours – node’s previous positions

21. Phase 3: Iterative refinement 1. Initial estimate A 2. Receive neighbour positions 4. Broadcast new position to neighbors 3. Local lateration

22. Monte Carlo Localization for Mobile Nodes Initialization: Node has no knowledge of its location. L 0 = { set of N random locations in the deployment area } Iteration Step: Compute new possible location set L t based on L t-1, the possible location set from the previous time step, and the new observations.

23. Phase 1: Initialization Initialization: Node has no knowledge of its location. L 0 = { set of N random locations in the deployment area } Node’s actual position

24. Phase 2: Prediction & Filtering Node’s actual position Prediction: Node predicts its new possible locations based on previous possible locations and given maximum velocity Filtering: Samples inconsistent with observations are filtered out Anchor node: Knows its own location and transmits it r

25. Observations Indirect Anchor If node does not hear an anchor, but one of its neighbors does, node must be within distance (r, 2r] of that anchor’s location. Direct Anchor If node hears an anchor, the node must lie on a circle with radius r of the anchor’s location S S r 2r

26. Questions 1- What are the main differences between range-free and range-based methods? Range-based methods require extra hardware therefore have a higher cost but provide more accurate distance measurements, whereas range-free methods use only connectivity information and so are less accurate. 2- What are the generic steps in calculating node position using anchor nodes? 1. Determine the distances between regular nodes and anchor nodes. 2. Derive the position of each node from its anchor distances. 3. Iteratively refine node positions using range information and positions of neighboring nodes. 3- What are the observations used for filtering the samples in the MCL algorithm. If node hears an anchor, the node must lie on a circle with radius r of the anchor’s location. If node does not hear an anchor, but one of its neighbors does, node must be within distance (r, 2r] of that anchor’s location.

27. References [1] I. Stojmenovic, Handbook of Sensor Networks: Algorithms and Architectures, Wiley Interscience, 2005. [2] K. Langendoen and N. Reijers, "Distributed Localization in Wireless Sensor Networks: A Quantitative Comparison“ Computer Networks (Elsevier), special issue on Wireless Sensor Networks, November 2003. [3] E. Stevens-Navarro, V. Vivekanandan, and V.W.S. Wong, “Dual and Mixture Monte Carlo Localization Algorithms for Mobile Wireless Sensor Networks,” in Proceedings of IEEE Wireless Communications and Networking Conference (WCNC), pp. 4024 – 4028, March 2007. [4] Y. Shang and W. Ruml, “Improved MDS-Based Localization,” in Proceedings of IEEE INFOCOM, 2004. [5] D. Niculescu and B. Nath, “DV Based Positioning in Ad hoc Networks,” Kluwer Journal of Telecommunication Systems. 2003. [6] L. Hu, and D. Evans, “Localization for Mobile Sensor Networks,” in Proceeding of Tenth Annual International Conference on Mobile Computing and Networking (MobiCom 2004), October 2004. [7] Y. Shang, W. Ruml, Y. Zhang, M. Fromherz, “Localization from Mere Connectivity,” in Proceedings of ACM MobiHoc 2003. June 2003. [8] Y. Shang, W. Ruml, Y. Zhang, M. Fromherz, “Localization from Connectivity in Sensor Networks,” IEEE Transactions on Parallel and Distributed Systems, vol. 15, no. 11, pp. 961-974, November 2004. [9] A. Savvides, W. Garber, S. Adlakha, R. Moses, and M.B. Srivastava, “On the Error Characteristics of Multihop Node Localization in Ad-Hoc Sensor Networks,“ Proceedings of the Second International Workshop on Information Processing in Sensor Networks (IPSN'03), pp. 317-332, April 2003. [10] A. Savvides, H. Park and M.B. Srivastava, "The N-Hop Multilateration Primitive for Node Localization Problems,", ACM Mobile Networks and Applications (Special Issue on Wireless Sensor Networks and Applications), pp. 443-451, 2003.