1. Source-Location Privacy Protection in Wireless Sensor Network Presented by: Yufei Xu Xin Wu Da Teng

2. Outline Introduction Related Work Model and Assumptions Implementation Theoretical Analysis Conclusion

3. Introduction Wireless Sensor Network 1. A set of low-cost radio devices 2. Supplied with limited amount of energy 3. Through multi-hops to deliver data to the base station

4. Introduction (Cont.) Source-location Privacy Problem in Wireless Sensor Network  Open architecture of the underlying wireless-communication technology  An adversary can easily detect the source by back tracing the routing path Evaluating Source-Location Privacy Protection Techniques  Safety Time The number of messages sent by source before it is identified  Energy Consumption Level Total number of messages sent within the entire sensor network

5. Related Work Basically, techniques for preserving source-location privacy are built upon routing protocols Two popular routing protocols employed in sensor network:  Flooding Routing  The source forwards a message to all its neighbours  Subsequent sensor nodes also forward the message to their neighbours  Each sensor node only forward the same message once  Single-Path Routing Only one path is established between the source and the base station e.g. the shortest path

6. Related Work (Cont.)  Both of them can’t protect the source-location privacy in sensor network As an approach, fake source messaging is proposed in [1]  Introduce fake source which generates false messages to mislead the adversary  Fake messages have same length and also encrypted so that an adversary can’t differentiate with the actual messages.  As indicated in [2], this approach has a set of limitations  Not efficient in energy conservation  Location of fake source is important  Frequency of message generation is also important

7. Related Work (Cont.) As a contribution, the phantom routing approach is proposed in [2]:  Phantom routing contains two phases:  Directed random walk phase to deliver a message to a phantom source (either sector-based or hop-based random walk)  Deliver the message from phantom source to base station by using either flooding or single-path routing  As indicated in [3], phantom routing still has limitations  It may lead the pre-termination of the random walk phase due to the inappropriate selection of random walk direction  Thus there may be performance dropdown in certain area of the sensing field

8. Related Work (Cont.) As an improvement, a self-adjusting directed random walk approach is presented in [3]  It divides the neighbour set into four directions (E, W, N, S)  The real source randomly selects one direction as the random walk direction  By encoding the direction vector into messages, it allows self- adjusting of random walk even when random walk is blocked on one direction  When two directions are blocked, a predetermined ratio is used to determine whether to continue the random walk or not However, we find that the above approach can be further improved

9. Model and assumptions A simulative environment is created for estimating performance.[3] – a square area of 6000x6000(m 2 ). – 10000 sensor nodes are located randomly in it. – the transmission range of each sensor node is chosen in a way such that a sensor, in average, has 8.5 neighbors.

10. Model and assumptions (Cont.) – the sink is set at the center of this area. – there is only one monitored asset. – its location remains unchanged before it is caught by the adversary. – 4 landmarks are set at 4 corners of this square, which will generate a message flood to help every sensor get location information of itself.

11. Model and assumptions (Cont.) On the other hand, an adversary may adopt two kinds of tracing strategy. – Patient Adversary : is referred as the adversary waits at a location until he receives a new message. – Cautious Adversary : which means he waits at a location for a specific period of time; if no message arrives within this period, he will return to its previous location.

12. Implementation Direction and position – four distinct directions: NE,NW,SW,SE -- more precise. – square area is divided into four parts. – each node belongs to one part. – another useful info is the distance to sink – hops number. – neighbors are grouped into four sets.

13. Implementation (Cont.) Goal for random walk – for the phantom routing protocol, the randomness of choosing phantom source is very important – unpredictable path. – it can be noticed that the farther from the phantom source to the real source, the better the location-privacy can be protected.

14. Implementation (Cont.) Workflow of improved method – each node maintains 4 lists for its neighbors. – before random walk, the real source checks whether it is near the center by comparing a threshold Dsink and its hops from sink. if in circle -> any direction if not -> direction to opposite part send msg to a neighbor in that direction

15. Implementation (Cont.) – when a node receives msg, it checks its hops if hops > hwalk -> take shortest-path to sink if not -> deliver to a neighbor on the way – if a node find no neighbor on the way, it change direction and forward msg. – if a node is in a corner, it just stops forwarding and begin to send msg to sink using shortest-path routing. -- no need to continue random walk coz msg has already traveled long enough.

16. Theoretical Analysis Longer effective distance Phantom Routing: 180 degree Our improved approach: 90 degree

17. Theoretical Analysis (Cont.) Phantom Routing: oscillation-way  near Our improved approach: relative position  far & lowest likelihood smaller probabilities to hit the boundary

18. Theoretical Analysis (Cont.) Less energy consumption Phantom Routing: directional information Our improved approach: no directional information

19. Conclusion Our approved approach achieves a longer safety time without consuming more energy than its original version. It is better than self- adjusting phantom routing in protecting source-location privacy.

20. Reference [1] Ozturk, C., Zhang, Y. and Trappe, W., ”Source-location privacy in energy-constrained sensor network routing”, Proceedings of the 2nd ACM workshop on Security of ad hoc and sensor networks SASN '04, pp. 88-93, Oct. 2004 [2] Kamat, P., Zhang, Y., Trappe, W. and Ozturk, C., “Enhancing Source- Location Privacy in Sensor Network Routing”, Proceedings of the 25th IEEE International Conference on Distributed Computing Systems, pp. 599-608, June 2005 [3] Zhang, L.,”A self-adjusting directed random walk approach for enhancing source-location privacy in sensor network routing”, Proceedings of the 2006 international conference on Wireless communications and mobile computing, pp. 33-38, 2006.

21. Question Any Questions ?