POT : An Efficient Top- k Monitoring Method for Spatially Correlated Sensor Readings

1. POT : An Efficient Top-k Monitoring Method for SpatiallyCorrelated Sensor Readings YongHyun Cho, Jihoon Son and Yon Dohn Chung Data management for sensor networks 2008

2. Outline • Introduction • Partial Ordered Tree • Simulation • Conclusion

3. Introduction • Since sensor readings are usually correlated with location, top-k nodes are clustered at some areas • This paper propose a novel tree structure to efficiently maintain clusters of the highest readings

4. Temperature of Kawah Ijen crater lake

5. POT : Partial Ordered Tree

6. Partial Ordered Trees

7. Initial construction of POTs TOP 3 (50,49,47) (48,45,39) (50,49,48,47,45,39) (48+47)/2=47.5

8. Local Top-k evaluation

9. Global Top-k evaluation

10. Updating the POTs • If the size of GR is k, the new threshold is the same to the previous one. • Otherwise, the new threshold is set as the average of the and the • In each POT, only the nodes that have reported their readings are required to update their thresholds.

11. Skewed distribution

12. AverageEnergyConsumption(1/3)

13. AverageEnergyConsumption(2/3)

14. AverageEnergyConsumption(3/3)

15. Network Lifetime(1/3)

16. Network Lifetime(2/3)

17. Network Lifetime(3/3)

18. Monitoring accuracy

19. Conclusion • In sensor networks, sensor readings are usually correlated with their locations • In this paper proposed a novel structure to manage the highest readings and process top-k query