1. FAR: Face-Aware Routing for Mobicast in Large-Scale Sensor Networks QINGFENG HUANG Palo Alto Research Center (PARC) Inc. and SANGEETA BHATTACHARYA, CHENYANG LU, and GRUIA- CATALIN ROMAN Washington University in St. Louis ACM Transactions on Sensor Networks (TOSN), November 2005 Chien-Ku Lai

2. Outline Introduction Mobicast Overview Face-Aware Routing for Mobicast Topology Discovery Simulation Results Conclusion

3. Introduction Wireless Sensor Networks Mobicast

4. Introduction - Wireless Sensor Networks Wireless sensor networks : are large-scale distributed embedded systems composed of small devices sensors wireless communication interfaces microprocessors maybe some actuators will soon be feasible to deploy dense collections of sensors to perform distributed micro-sensing of physical environments

5. Introduction - Wireless Sensor Networks (cont.) Many sensor network applications have fundamental spatiotemporal constraints an intruder tracking application examplean information scouting example

6. Introduction - Mobicast Mobicast represents a new information dissemination paradigm with spatiotemporal semantics allow applications to specify their spatiotemporal constraints by requesting a mobile delivery zone in turn enables the application to build a continuously changing group configuration

7. Introduction - Mobicast (cont.) Challenge Hole

8. Introduction This article presents : a new Face-Aware Routing protocol (FAR) for mobicast a related spatial neighborhood discovery algorithm

9. Mobicast Overview Limitation of Approaches Based on Geocast Advantages of Just-in-Time Delivery

10. About Mobicast The mobicast service supports a type of application information delivery request right-place and right-time  just-in-time

11. Limitation of Approaches Based on Geocast Duplicate area

12. Advantages of Just-in-Time Delivery

13. Face-Aware Routing for Mobicast The Planar Spatial Neighborhood Face-Aware Routing

14. About face routing The idea of face routing is inspired by previous geometric routing algorithms Compass routing FACE-2 GPSR GOAFR+ Faces

15. The Planar Spatial Neighborhood A Face D Spatial neighbor E C BP F

16. The Planar Spatial Neighborhood G Face Spatial neighbor C BP F L N HI J K M

17. The Planar Spatial Neighborhood M Face D Spatial neighbor E C F L N G O HI J K

18. Face-Aware Routing The face-aware algorithm consists of two methods for forwarding packets: greedy forwarding timed forwarding

19. About FAR Packet Format Greedy Forwarding Timed Forwarding Protocol Termination

20. Packet Format Packet header 12345678 Sender location Packet sending time-stamp Initial delivery zone coordinates Delivery zone velocity Message lifetime Message type Sender packet sequence number The last forwarder location identify each packet on the network

21. Greedy Forwarding Greedy forwarding applies to all nodes that are currently (or previously) covered by the mobicast delivery zone or have at least one spatial neighbor that is currently (or previously) covered by the mobicast delivery zone In such cases, a node forwards a new packet in an “as-soon-as possible” fashion

22. Timed Forwarding Timed forwarding applies to a node that has no spatial neighbor in the current delivery zone but either itself will soon be in the delivery zone or has at least one spatial neighbor that will be in the delivery zone

23. Example G L M E F I H P C B A J D N K P G L M E N F AA J C B D K Greedy forwarding mode Timed forwarding mode

24. Timed Forwarding (cont.) the expected 1-hop network latency the hop distance The forwarding decision of X is as follows: (1) If T a ≤ 0 forward the packet as soon as possible (2) If T a > 0 delay the forwarding for time length T a

25. Protocol Termination A packet is not simply ignored if it has expired An expired packet is dropped only in the timed forwarding mode If a node is in greedy forwarding mode it will forward the packet even if the packet has expired

26. FAR assumption The perpendicular span to be no smaller than the maximum communication range

27. Topology Discovery Face Identification Face Traversal Termination Cost Minimization The Outer Face

28. Topology Discovery

29. Face Identification How to make each discovery message traverse the correct face ? using a ring-buffer on each node for storing the incident planar edges A B C D A’s ring-buffer: A-B A-C A-D

30. Face Traversal Termination How to determine when a message has traversed the whole face ? A node determines if an incoming discovery message dm has completed a full traversal of a face by the following criterion : the outgoing edge for dm is contained in its ordered traversal list  “when the message comes back to a node already traversed” does not work

31. Face Traversal Termination Example G

32. Cost Minimization How to coordinate between nodes such that only one discovery message flows around each face ? On each face, ideally one traversing discovery message will suffice

33. Cost Minimization (cont.) Two strategies for reducing the number of discovery messages Random starting time Starting location-based tiebreaking rule east is preferred, if there is still a tie, north is preferred

34. The Outer Face A practical way to identify an “outer” face is from its size a discovery message has a max hop count

35. Simulation Results Spatial Reliability Temporal Characteristics

36. Simulation environment 600 sensors 1000 × 400 m 2 area dispersed randomly under the uniform distribution 120 × 100 m 2 rectangular delivery zone NS2 network simulator IEEE 802.11 MAC protocol

37. Spatial Reliability Average and Standard Deviation of Delivery Ratio of FAR, Greedy and Geocast

38. Temporal Characteristics - Slacktime

39. Temporal Characteristics - Slacktime (cont.)

41. Conclusion FAR Spatial neighborhood discovery protocol

42. Conclusion - FAR FAR a new face aware mobicast routing protocol which, in theory, reliably delivers messages spatially has good mobicast temporal characteristics relies on the notion of spatial neighborhoods features a novel timed face aware forwarding method

43. Conclusion - Spatial neighborhood discovery protocol This paper addressed key issues a spatial neighborhood discovery protocol must consider face identification discovery termination duplicate elimination

44. Question? Thank you.