1. Tracking Irregularly Moving Objects based on Alert-enabling Sensor Model in Sensor Networks 1 Chao-Chun Chen & 2 Yu-Chi Chung Dept. of Information Management Dept. of Comp. Sci. and Info. Engr 1 Shih-Chien University, Kaohsiung Campus, Taiwan 2 National Cheng-Kung University, Taiwan Proceedings of the 2005 11th International Conference on Parallel and Distributed Systems (ICPADS'05)

2. Outline Introduction Sensor Network Architecture Supporting Object Tracking Applications Alert-enabling Sensor Model Alert-based Object Tracking (AbOT) Scheme Simulation Conclusion

3. Introduction Tracking moving objects is one of widely used techniques habitat monitoring The tasks of the sensor applications Data sensing data transmission active/sleep decision

4. Introduction Object tracking applications are related to the sensor tasks (caused limitations) power constraint velocity/sensing range uncertainty tracking resolution data freshness scalability

5. Introduction (motive) the energy cost is wasted in either the operation of tracking moving objects Wake up more sensors to monitor the object the operation of finding back lost moving objects Only few sensors are in active mode to track the moving object Two extreme cases are not optimal since they waste high energy cost on different operations

6. Introduction (solution) The Alert-based Object Tracking (AbOT) Scheme Dynamically determine the number of awaken sensors According to the ability of capturing the location of the moving object

7. Sensor Network Architecture Supporting Object Tracking Applications 1.the measurement history of object locations 2.full knowledge of the network topology 3.the status of each sensor 1.Continuously storing the incoming sensor readings 2.minimizing the energy consumption 3.offering the query interface for end users

8. Sensor Network Architecture Supporting Object Tracking Applications For ease of management Sensors are suggested to deploy in the grid topology

9. Alert-enabling Sensor Model Each sensor plays the role of a sentry sends a trigger message to alert the central server while observing the urgent phenomenon The central server then makes the final decision by receiving reports

10. Alert-enabling Sensor Model The sensors are used to alert the central server for the change of the object movement alert-enabling sensors Marginal Zone Central Zone Alert Circle The alert circle greatly affects the tracking performance

11. Alert-based Object Tracking (AbOT) Scheme In object tracking applications, energy waste is a function of two elements the number of active sensors at each sampling tracking accuracy that determines the additional energy on finding the lost object back An ideal tracking method use a few active sensors to track objects with high accuracy

12. Alert-based Object Tracking (AbOT) Scheme Alert-based Object Tracking (AbOT) Scheme based on two key ideas sensor-level processing using alert-enabling sensors to capture the motion of the moving objects network-level processing using sensor/server cooperating decision to capture the trajectory of a moving object  Turning on the “best” set of sensors to track the moving object

13. Alert-based Object Tracking (AbOT) Scheme Each sensor is always in one of three states doze mode the object is not nearby, the sensor stays in the doze mode to save energy M-zone the possibility that the object exits the sensing range of the sensor is high  sensor should monitor the object with high sensing frequency

14. Alert-based Object Tracking (AbOT) Scheme C-zone the possibility that the object exits the sensing range of the sensor is small the sensor should monitor the object with low sensing frequency

15. Alert-based Object Tracking (AbOT) Scheme The moving object moves across the sensing range The OT server would determine appropriate sensors to switch the process of tracking the moving object the hand-off approach Nearest Sensor Policy Range Covering Policy Conservative Policy

16. Location Monitoring Mechanism The algorithm of the location monitoring mechanism includes two parts Sensor Part sensor OTserver To monitor object C-zone M-zone M-event monitors the object with high sampling rate C-event sensor decreases the sampling rate to monitor the object

17. Location Monitoring Mechanism Server Part deal with the arriving message storing the sensor readings executing the responding tasks of the alerts the message processing basically runs as a daemon in the OT server the object position and its associated timestamp are stored into the moving object database

18. Location Monitoring Mechanism If an alert is piggybacked in returning messages OT server knows the moving object has moved across the alert circle (two event) M-event wake up more sensors around the object to monitor Three heuristic designs Nearest Sensor Policy OTserver Wake up the nearest sensor

19. Location Monitoring Mechanism Range Covering Policy Conservative Policy

20. Location Monitoring Mechanism C-event the OT server forces other active sensors to the doze mode, except the sensor that the object resides M-zone C-zone OTserver Wake up the sensor

21. Lost-and –found Mechanism sensor C-zone M-zone a fast and irregular moving object (e.g., a taxi driver) OT server could wake up the wrong set of sensors to monitor the OT server wakes up all sensors by sending the Lost-and-Found message

22. Simulation 900 sensor nodes in 300 × 300 meter2 between two vertical (or horizontal) neighbors is ten meters the sensing range is 15 meters C-zone is set to 7.5 meter object moves from C-zone to M-zone Range Covering Policy is used to determined which sensors are waked up

23. Simulation Energy consumption the total energy consumed in the object tracking period by the network Frequency of waking up sensor the sum of active nodes at each time unit during the simulated period Accuracy probability the ratio of the number of successful detections to the total number of detections

24. Simulation Effect of the Sensor Network evaluate the performance of object tracking scheme in various sizes of sensor networks

25. Simulation Effect of the Sensor Nodes

26. Simulation Effect of the Moving Object

27. Conclusion we propose the AbOT scheme to efficient process the object tracking function for sensor applications To study the optimization problem of tracking multiple moving objects at the same time