Integration of Sensor Networks with Mobile Devices

1. Computer Science Department University College London Integration of Sensor Networks with Mobile Devices Vladimir Dyo November 22, 2005

2. Outline • Introduction • Problem and Motivation • Isolating Problems • Expected Contribution • Initial Research Results • Conclusion

3. Data Dissemination in SensorNets • The Concept • Sensors, Sinks • Tasks: Dissemination, Aggregation, Streaming • Limitation: Energy, Low Reliability

4. Current Research • Infrastructure • Routing • GPSR • Clustering • LEECH • Data Dissemination and Aggregation protocols • Directed Diffusion, TAG, Synopsys Diffusion • Middleware • Milan, TinyLime, Slime, EnviroTrack

5. Problem - I • Existing solutions work with static sinks: • Very strong assumption • Not Realistic • Not Energy Efficient • No Coverage • Examples: TinyDB, Cougar

6. Problem - II: Example • Static collection: • Energy consumption • Congestion • Reliability • Not practical Suggested Approach: Mobile Collection

7. Solution – Mobile Collection • Technical Advantages • Coverage, energy, disconnected operations, flexibility • Application Potential • Environmental monitoring, industrial and pervasive applications • Not Well Researched • M. Batalin, “Coverage, Exploration and Deployment by a Mobile Robot and Communication Network” • A. Chakrabarty, “Exploiting Predictable Observer Mobility For Power Efficient Sensor Network Configuration” • “S. Bhattacharya, Energy-conserving Data Placement and Asynchronous Multicast in WSN”

8. Isolating Problems • Protocol Support • Mobile Data Collection. Routing Data towards a mobile collector • Spatial Queries and Distributed Index • select locations where (temp > 30) OR (humidity > 0.4) within 30km from (X,Y) • Middleware • Defining Primitives • Implementing Protocols

9. Expected Contribution • Protocols • Data Collection Protocol for Mobile Collectors • Distributed Index for Spatial Queries in Sensor Networks • Middleware for integration of sensor networks with mobile devices • Architecture • Primitives • Implementation

10. Evaluation • Data Collection Protocol: • Performance analysis in ns-2 simulator • Critical Metric – total communication overhead • Real traces from other projects • Implementation in TSky motes • Middleware • Implementation • Real Deployment • Climate and wildlife monitoring (in collaboration with Oxford zoologists) • Micro-climate monitoring (UCL campus)

11. Initial Results • Adaptive Distributed Indexing for Spatial Queries in Sensor Networks. • V. Dyo and C. Mascolo. In IEEE Proceedings of 8th International Workshop on Mobility in Databases and Distributed Systems (co-located with DEXA05). August 2005, Copenhagen, Denmark. IEEE Computer Society Press

12. Proactive + Reactive Mode 600 query/hour Total: 3601 Total: 2400 Proactive = 1 Proactive = 3600 Total: 1201 Reactive = 1200 Reactive = 1200

13. Future work on Index • Data Resolution • High update rates  High Power Consumption • Low response • Low latency  High Power Consumption • Trade-off between: • Data resolution/response latency • Total energy consumption

14. Conclusion • Interesting problem • Isolating issues • Initial results

15. Questions • Vladimir Dyo • v.dyo@cs.ucl.ac.uk