1. Routing Algorithm for Large Data Sensor Networks Raghul Gunasekaran Group Meeting Spring 2006

2. Past Trends in Sensor Networks Serve only for detection Only serve as an Alarming circuit Highly resource constrained nodes - low processing capability - less intelligence on Application Least cost nodes Small in size Basically, built for military applications

3. Design Issues Energy efficient Scalable operate in large numbers Robust working with faulty nodes Task Balancing sensing, transmission, relaying Network Dynamics support network changes Minimum Transmission Support a heterogeneous environment Data aggregation QoS – guaranteed delivery, latency

4. Present Trends in sensor networks Powerful nodes - high processing capability - intelligent systems Cost and Size not a limiting factor – tradeoff over the capability of the sensing modules Support high end sensors like cameras Need to be energy efficient with all these capabilities

5. Design Issues Energy efficient Scalable Robust Task Balancing Network Dynamics Minimum Transmission Support a heterogeneous environment Data aggregation QoS With a great emphasis on Data aggregation Data compression techniques

6. Classification of Routing Protocol for Sensor Networks Data Centric Reactive/Proactive Energy Efficient Geographic aware Hierarchical / Cluster Based PUSH/PULL Maximum Life Time Routing QoS aware routing

7. Difference between Sensor Networks and General Ad-hoc Wireless Networks In a sensor networks, you know what data is going to be transmitted - An interest message - A data message - from a temperature, light sensor - image - video stream You know the size of the data that’s going to be transmitted, unlikely in ad-hoc networks Why not make routing decisions based on the size of data ?

8. Radio in the motes In Sensor module’s Radio can be operated in more than one power levels Example: Telosb mote - Not really explored for routing algorithms ? Table referenced from Crossbow Telosb Datasheet

9. Routing decisions made on the size of the data, with knowledge of energy availability in the sensing modules Path for Large data, say an image Path for relatively smaller data, say a Temperature reading

10. What difference it makes ? Larger the data, the time taken for transmission is higher, the radio is on for a larger amount of time and so the power consumed is high. Instead of using larger number of hops for communicating large volume of data we can utilize smaller number of nodes with higher radio power level. Cumulatively, this would take less power than a collection of nodes involved for routing. This would be appropriate for large data since the transmission time is high, and for small data it wouldn’t make a big difference, since the time the radio is on very less.

11. Would it affect the network lifetime ? - Network Lifetime is the time taken for the first functional node to fail - In general nodes are considered to fail when the battery source is not sufficient to power them. Not really, because we have also taken into consideration the energy available in nodes for routing. For the next transmission routing decision would be made based on the energy available in the nodes, so a node with higher energy level would be chosen and probably would be a node which didn’t participate in the earlier routing process.

12. Needs to be proved ? Implementation would be carried out on NS2.26 NS2 has implementation of Directed Diffusion, one of the first protocols designed for wireless sensor networks and was implemented in NS2. The diffusion implementation, facilitates implementation of routing protocols for sensor networks. They have a diffusion core which has attributes declared and specific routing algorithms are implemented as filters, aiding in comparison of protocols

13. Thanks Questions ?