1 / 40

Potential-Based Entropy Adaptive Routing for Disruption Tolerant Networks

Potential-Based Entropy Adaptive Routing for Disruption Tolerant Networks. Hideya Ochiai Hiroshi Esaki The University of Tokyo / NICT DTNRG, IETF 76 Hiroshima, 2009-11-13. Outline. Introduction Potential-based entropy adaptive routing Prototype implementation Campus-wide experiment

ghita
Télécharger la présentation

Potential-Based Entropy Adaptive Routing for Disruption Tolerant Networks

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Potential-Based Entropy Adaptive Routingfor Disruption Tolerant Networks Hideya Ochiai Hiroshi Esaki The University of Tokyo / NICT DTNRG, IETF 76 Hiroshima, 2009-11-13

  2. Outline • Introduction • Potential-based entropy adaptive routing • Prototype implementation • Campus-wide experiment • Conclusion

  3. Introduction • DTN for opportunistic networking • Isolated sensor networking / vehicular ad-hoc networks • Intermittent connectivity / isolated networks • API • void sendMessage(ID dst, String msg); • void recvMessage(ID src, String msg); • All the nodes are always virtually connected. Isolated Networks 2 1 7 4 3 8 Intermittent Connectivity 6 5 Mobile Node

  4. Contributions of this work Implementation and deployment of potential-based entropy adaptive routing (PEAR) • PEAR autonomously enables message delivery in ad-hoc manner. • In general, the performance of routing algorithms are strongly dependent on mobility patterns. • PEAR dynamically adapts to wide-range of mobility patterns without being aware of mobility pattern itself.

  5. Outline • Introduction • Potential-based entropy adaptive routing • Prototype implementation • Campus-wide experiment • Conclusion Potential-field Large entropy case Small entropy case

  6. Characterize Mobility Pattern by Entropy Small Entropy Case Large Entropy Case Locally Distributed Widely Distributed Delivery by hop-by-hop forwarding Delivery by the movement of the nodes

  7. Entropy and Delivery Pattern s s 1 2 3 1 2 3 A bolder link indicates larger contact time. d d Small Entropy Case Large Entropy Case Biased contact Uniform contact It should improve delivery rate by increasing redundancy. It should choose the best path

  8. How does PEAR achieve that ?? • For choosing the next hop node: • Potential-based routing • Potential-field construction (inspired by diffusion theory) • For message delivery • Copy-based message delivery  Transfer-based message delivery

  9. Potential-Based Routing Potential Internet Gateway Data Flow Live E! Truck Sensor Data Flow To deliver sensor readings to the Internet GW : Wireless Device

  10. Potential-Field Construction in PEAR Potential-Field Construction Diffusion Equation

  11. Potential-Field Construction

  12. ConstructedPotential-Field

  13. Potential and Message RoutingSmall Entropy Case Message Delivery Source Potential Intermediate Destination Intermediate Nodes Source Node Destination Node Distribution

  14. Potential and Message RoutingLarge Entropy Case Source Node Direct Path Potential Intermediate Nodes Destination Node Source Node Destination Node Possible Intermediate Nodes Distribution

  15. Transfer-Based v.s. Copy-Based Message Delivery PEAR Copy-Based Message Delivery Transfer-Based Message Delivery

  16. Copy-Based Message Transfer in PEAR Do you have any information about msg_id= XXXX ?? H BODY H BODY s H BODY n-1 source n H BODY n+1 Algorithm d destination • NOT_HAVE • ALREADY_HAVE • DELIVERED

  17. Rough Simulation-Based Evaluation

  18. Rough Simulation-Based Evaluation (Summary) Delivery Rate PEAR 100% Spray and Wait Link State Based Mobility Pattern Village-to-village Human-to-human

  19. Outline • Introduction • Potential-based entropy adaptive routing • Prototype implementation • Campus-wide experiment • Conclusion

  20. Software Design of PEAR About 3000 lines in C. Footprint is 34k byte in object code. Application (Message Send / Receive) Potential Table Nexthop Table Advertisement Manager Message Manager Message Pool Message Copy Investigation Advertisement

  21. Armadillo220 Storage(2GByte) Wifi 802.11g Power Circuit Battery(6.0V 2100mAh)

  22. Armadillo-220

  23. Outline • Introduction • Potential-based entropy adaptive routing • Prototype implementation • Campus-wide experiment • Conclusion Delivery pattern Experiment settings Contact graph

  24. Members for the experiment • The University of Tokyo • Ochiai, Shimotada, Fujita, Kawakami, Himura, Sugita, Lert, Wan, Minshin, Motodate, Kure, Kawaguchi, Ishizuka • Nara Institute of Science and Technology • Dr. Matsuura • Keio University • Dr. Miyakawa, Yamanouchi • Cisco Systems, Inc. • Momose

  25. Scenario Overview 1. Deploy sensors in remote sites. 2. Sensors periodically transmit data to the GW 3. Mobility of nodes enables delivery of data from sensors to the GW. InternetGW (#99) Mobile Node (#4) Mobile Node (#3) Mobile Node (#5) Sensor Static (#1) Sensor Static (#2)

  26. Gateway 50m

  27. Contact Graph The number of received neighbor advertisement packets. e.g., node 8 received 45 advertisement from sensor 1. Mobility Entropy≒2.1

  28. Potential for the Gateway

  29. Message Flow (1/2) Red arrow: messages from sensor 1 (#1) GW: #99 Blue arrow: messages from sensor 2 (#2)

  30. Message Flow (2/2) The number along with an arrow indicates the time of the transfer

  31. Distribution of Delivery Latency E.g., 50 = [0,100]

  32. Collected Temperature Data from Sensor 1 Delivery Rate = 100 %

  33. Collected Temperature Data from Sensor 2 Delivery Rate = 100 %

  34. Outline • Introduction • Potential-based entropy adaptive routing • Prototype implementation • Campus-wide experiment • Conclusion

  35. Conclusion • We proposed PEAR for opportunistic networking • It directly forwards a message at small entropy cases • It replicates a message to improve delivery latency at large entropy cases • Implementation and deployment of PEAR • Prototype system with embedded computers • 10-node scale campus wide experiment • PEAR has achieved 100% delivery rate with reasonable delay on the experiment settings.

  36. Thank you... Google by "Mobility Entropy and Message Routing" Large entropy case Small entropy case PEAR maintains high delivery rate over wide-range of mobility patterns Prototype implementation

More Related