1. Autonomic Wireless Sensor Networks: Intelligent Ubiquitous Sensing G.M.P. O’Hare, M.J. O’Grady, A. Ruzzelli, R. Tynan Adaptive Information Cluster (AIC) Group, University College Dublin, Ireland.

2. Summary  Wireless Sensor networks (WSNs)  Generality  Prototypes  Applications  Requirements  Autonomic computing  Properties  Autonomic WSNs  Agent technology in WSNs  Autonomic network management  Intelligent power management  Intelligent communication and routing  Intelligent network coverage  Conclusion

3. Energy-Efficient Wireless Sensor Networks (WSNs) A large number of tiny wireless devices to sense the environment: –Sensor nodes Sensing unit (temp,pressur e, vibration, sound, etc.) Small processing unit Short range radio unit Small power unit Few more powerful devices to collect the data: –Gateways (or sinks) High processing unit Wireless comm. unitHigh power unit Wired comm. unit PDA, laptop, PC etc.

4. Some WSN applications Remote area monitoring Object location Industry machinery monitoring Disaster prevention Wireless medical systems Wind Response Of Golden Gate Bridge environmental data collection: temperature light, humidity, pressure, solar radiation.

5. Wireless sensor characteristics Sensors are battery operated for long unattended period:  Saving energy is a primary objective Sensors are of : Low cost Low processing capability  System strength based on sensor collaboration Large scale networks  Multihop communication WSN manager

6. WSN issues Large number of nodes  Scalability issues High dynamic condition (number and position of nodes might change)  Network Reactivity and Self-organization Power management  The network need to be connected as long as possible System reliability –The wireless signal might be subjected to interference Robustness –Subject to environmental variability –Complex interoperability of network devices

7. Sensor node prototypes Philips sand nodes Mica2 mote Eyes node prototype Tyndall sensor

8. General wireless sensor network architecture: Any layer try to achieve the task using the smallest amount of energy possible Antenna Sensing devices Application Data interpolation Routing MAC Physical Sensing coverage Cross layer interaction Localization

9. Autonomic WSNs: Origin of autonomic computing by IBM Relieve human of the burden of managing large scale computer systems Autonomic WSNs properties: –Self healing –Self protection –Self configuration –Self optimization –Self managing

10. Agent technology for autonomic WSNs Agent properties: –Sense-deliberate-act cycle Sensing data is used as input for the decision making process –Mobility Useful characteristic of agents that well map onto WSNs Agent can migrate from one node to another processing data as it goes –Fault tolerance Agents can still take decision if some data are missing

11. An example: Network anomaly intervention Possible solution Multiple Notification messages (High energy consuming) Proposed solution: Migrating agent (Moderate energy consuming)

12. Contribution of autonomic computing to WSNs Self configuring nodes (1) can set up a network; (2) might not be well positioned but still work; (3) can evaluate network gaps; (4) can decide communication schedule. Self protection attribute –Migrating agents check channel condition and battery level before migrating Self healing –Repair network damage due to hash work condition –Negotiating new routes; –Activating redundant nodes; –Ask for replacement of damaged nodes. Self optimization –Quality of service –Network efficiency –Delay control and data prioritization

13. Intelligence-aided sensor network Opportunistic power management Intelligent coverage Intelligent routing

14. Opportunistic power management (1) Increase network longevity by deactivating redundant nodes: node hibernation Sensing Coverage: –All points within the sensed area need to be covered by at least 1 sensor. Traditionally, a point is covered if it is within the sensing range of a given sensor. Gateway Redundant based on sensor coverage

15. Intelligent sensing coverage It deals with the quality of sensory data provided to the application which is using it; Data sampling frequency at the node and surrounding nodes should be enough to have a certain detail of the phenomena of interest; Migrating agents control: –Sensor sampling rate by tuning it; –Might request an increase of node density in an area

16. Intelligent routing By interacting with different layers the agent can check several parameters A look-up table with neighbouring nodes parameters (RSSI, battery level, location) is provided Even with incomplete data an agent can figure out the best neighbours to which to forward the data to Antenna Routing MAC Physical Route managing Agent table

17. Conclusion A methodology for the automation of the sensor network technogy has been presented; WSNs offer significant potential in numerous application domains Given the diverse nature of such domain, robustness and reliability is essential; The method outlined exploits the integration of autonomic principles into design and operation of WSNs Autonomic WSN can effectively both enhance the network reliability and extend the network lifespan.

18. Thank you! www.csi.ucd.ie/researchwww.csi.ucd.ie/research (Prism LAB web site) www.adaptiveinformation.iewww.adaptiveinformation.ie (project)