1. Mechanical Transport of Bits - Part II Jue Wang and Runhe Zhang EE206A In-class presentation May 5, 2004

2. Outline  A Message Ferrying Approach for Data Delivery in Sparse Mobile Ad Hoc Networks (MobiHoc 2004) W. Zhao et al.  Intelligent Fluid Infrastructure for Embedded Networks (MobiSys 2004) A. Kansal et al.

3. Sparse MANET  What is Sparse MANET?  What are the challenges for Sparse MANET?  What are the solutions? Extending transmission range Store – carry – forward Reactive vs Proactive

4. Related Work  Ad Hoc Network Routing: DSR, DSDV, AODV, GPRS, ZRP, LAR, CEDAR  Ad Hoc Network Routing for Sparse Networks (in detail next slide)  Capacity of Wireless Network Gupta and Kumar Grossglauser and Tse  Topology Control

5. Related Work:sparse – MANET  Infostation: (Goodman et al): High bit rate connection, geographically distributed, discrete coverage  DataMule: (Shah et al): Static sensor nodes, controllable mobile entities to ‘move’ data  DTN: (Fall et al): no control over the network, nodes stay there, hope for the best  Range extension (Ahmed et al): increase range to overcome the partitions  SWIM (Small and Hass): combines infostation and ad hoc networking architecture (has been presented – Whale)

6. Sparse – MANET (cont.)  Mobility assisted: Proactive, Reactive  Epidemic routing: (Vahdat et al) Flooding (pros: robustness, cons: redundant messages)  Mobile Relay Protocol: (Nain et al) take advantage of node mobility to overcome message delivery  Actively modify trajectories to transmit as soon as possible (hard to have multiple transmission simultaneously)

7. Message Ferrying (MF)  Proposed in this paper  Two different types: Node-Initiated MF (NIMF)  Ferries move around the deployed area according to known routes and communicate with other nodes they meet. With knowledge of ferry routes, nodes periodically move close to a ferry and communicate with the ferry. Ferry-Initiated MF (FIMF)  Ferries move proactively to meet the nodes. When a node wants to send packets to other nodes, it generates a service request and transmits it to a chosen ferry using long range radio. Upon reception of a service requests, the ferry will adjust its trajectory to meet up with the nodes.

8. Node-initiated Msg Ferrying

9. NIMF – node operation

10. NIMF – Ferry Operation

11. FIMF – node operation

12. FIMF – Ferry Operation

13. Performances  NS: Network Simulator  802.11 with 250m communication range  5000x5000m – make sparse  40 nodes, Random Waypoint Models  Single Ferry, 15m/s, buffer size 400 message, route: rectangle with (1250,1250), (3750,3750) as diagonal pts.  25 nodes chose to send message every 20 seconds

14. Performance: Impact of Buffer Size

15. Performance: Impact of mobility pattern

16. Performance: FIMF: impact of transmission range

17. Application  Crisis Driven Battlefield and disaster applications  Geography Driven Wide area sensing and surveillance app.  Cost Driven  Service Driven

18. Conclusion  Sparse network  Solution: Proactive vs Reactive  Proposal: two schemes for message ferry  Simulation results.

19. Intelligent Fluid Infrastructure for Embedded Networks A. Kansal et al. (MobiSys 2004)

20. What is the paper talk about?  Use of external mobility for improving network performance.  External mobility: Controllable mobility – autonomous mobile router  Network: Sensor Network

21. Type of Mobility  Random Mobility Increase capacity (Grossglauser and Tse) Application: Whale, Zebranet Problem: Unbounded Delay  Predictable Mobility Chakrabarty et al. (commuter bus model) Problem: Usually mobility pattern is not predictable  Controlled Mobility This paper: External mobility – (for ecological or habitat researches, no free mobile components, these mobiles may be limited in capacity, maneuverability, etc.) Another application: DTN

22. Advantages using controlled mobility – 1  Increased system life time. How? In paper: Reducing the packet sent (relays – fewer hops) - reducing energy consumption. More: when you reduce the hop count, you increase the spatial reuse, you also increase the throughput

23. Advantages using controlled mobility – 1 – Simulation

24. Advantages using controlled mobility – 2  Data Fidelity The less hop it is, the less probability of error it occurs. Increase quality of received data, decrease the number of retransmission.

25. Advantages using controlled mobility – 3

26.  Reduced latency No mobile router case: T ideal= T(A,A,B)+T(A,B,Base)+T(B,B,Base) Mobile router case: T mobile =D(Base,A)+T(A,A,MR)+ D(A,B)+T(B,B,MR)+D(B,Base)+ T({A,B},MR,base)

27. Advantages using controlled mobility – 4 and others  Sparse and disconnected Networks Reduced communication range Reduced energy consumption  Less hop counts, easier synchro- nization  Security  Localization

28. Processing Platform  Stargate xScale  802.11 cards  Motes  Packbot (60W)

29. Adaptive Motion Control - Constraint  Energy limitations  Terrain constraints  Disturbances, noises  Environment constraints

30. Adaptive Motion Control - Objective  Maximize the lifetime of the system  Maximize the total amount of data collected  Minimize the data transfer delay  Minimize the buffer size  Minimize the recharge time?  Formulate as Optimization Problem?

31. Influence of speed of data collection  No effect on packets/second

32. Latency Sensitive Data Collection  SCD: Stop to Collect data Stop at locations where static nodes are found waiting with data

33. Latency Sensitive Data Collection  ASC: Adaptive Speed Control Move slower in regions where data collection is moderately poor and stop in regions where data loss is severe. N1: nodes with low delivery % N2: nodes with high delivery % T: round traversal time Delta = T/2 * 1/(n1+n2/2) SL : encounter of node type N2 ST : encounter of node type N1 TE : current time timer expired Sigma : duration which a timer is reset

34. Latency Sensitive Data Collection

35. Latency Sensitive Communication in sparse networks  Propose to use SCD algorithm

36. Experimental Results – 1

37. Experimental Results – 2

38. Experimental Results – 3

39. Conclusion and Future Works  Controllable Mobility introduced  Advantages for using mobile router  2 Strategies for moving 1 mobile router  Collaboration between mobile routers  Scenarios where the sensor nodes are moving themselves – MANET

40. Thank you!