ECN B4 : Dao Thanh Chung ( bull@ht ) Tutor : Takatoshi Kanazawa ( takatosi@ht )

1. fNode : Reducing Network Packet Transmission Overhead in Indoor Heterogeneous Wireless Sensor Networks Graduation Thesis Final Presentation Tokuda Lab ECN B4 : Dao Thanh Chung (bull@ht) Tutor : Takatoshi Kanazawa (takatosi@ht)

2. Background • Traditional WSN deployment • Because of cost and complexity of node deployment, network deployment using a single sensor node hardware platform • Narrow monitoring • Sensing ability: either temperature, humidity or fire • Heterogeneous wireless sensor network • Consist of sensor nodes with different capabilities • Radio frequency, Sensing range, Hardware platform • Wide monitoring • Temperature, humidity, fire, sound, etc.

3. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead Room Room Room Room Hall Redundancy Data WC Plant Sink Center conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

4. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead Room Room Room Room Hall １ Redundancy Data WC Plant Sink Center conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

5. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead Room Room Room Room Hall １ Redundancy Data 2 WC Plant Sink Center conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

6. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead Room Room Room Room Hall １ Redundancy Data 2 WC Plant Sink Center 3 conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

7. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead Room Room Room Room Hall １ Redundancy Data 2 WC Plant Sink Center 3 4 conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

8. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead Room Room Room Room Hall １ Redundancy Data 2 WC Plant Sink Center 3 5 4 conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

9. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead Room Room Room Room Hall １ Redundancy Data 2 WC Plant Sink Center 6 3 5 4 conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

10. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead １ Room Room Room Room Hall １ Redundancy Data 2 WC Plant Sink Center 6 3 5 4 conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

11. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead １ 2 Room Room Room Room Hall １ Redundancy Data 2 WC Plant Sink Center 6 3 5 4 conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

12. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead １ 2 3 Room Room Room Room Hall １ Redundancy Data 2 WC Plant Sink Center 6 3 5 4 conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

13. Problem Definition • Heterogeneous sensor nodes are required to be deployed under the same environment • Sensor nodes with different hardware platforms cannot communicate with each other →Redundant Packet Transmission Overhead １ 2 3 Room Room Room Room Hall １ 4 Redundancy Data 2 WC Plant Sink Center 6 3 5 4 conference room conference room Room WC Temperature sensor • Black node in the hall needs many hops (6 hops) Fire sensor

14. packet transmission overhead • Experiments with 8 sensor nodes in classroom environment • Send packets from the first node to 8thnode • Increase power consumption and delay time when the number of hops rise Number of hops Number of hops Consumed Power Delay time

15. Proposed Solution • Propose fNode • Act as an intermediate node for packet forwarding • Replace redundant forwarding nodes • Convert and forward packet’s format compatible with various platforms • Propose fMap Algorithm • Minimize the network packet transmission overhead • Require an ideal deployment scenario of fNode • fMap algorithm estimates fNode’s positions

16. Room Room Room Room Hall １ Redundancy Data WC Plant Sink Center ２ ６ ４ conference room conference room ３ Room WC ５

17. fNode Deployment Scenario • Black node’s packets in the hall are transmitted by a shorter path via fNode (3 hops) １ Room Room Room Room 3 2 Hall fNode WC Plant Sink Center conference room conference room Room WC

18. Design of fNode • Receive packets • Convert the packet’s format • Transmit converted packets A Radio Module A Radio Module A packets B packets Incoming packets Outgoing packets B Radio Module B Radio Module Micro Controller B packets A packets

19. fMap Algorithm Forwarding node Temperature sensor Fire sensor Original Topology

20. fMap Algorithm Remove all forwarding nodes

21. fMap Algorithm Topology after one fNode added (Find fNode position by reviewing all deployable positions)

22. fMap Algorithm Topology after 2ndfNode added All nodes are connected → stop loop

23. Implementation • In order to evaluate packet transmission overhead • Design fNodetestbed • fNodetestbed • A laptop • Sensor nodes are connected • We provide a GUI to implement fMap • Java language Converting packets fMap GUI Iris node Sun Spot node

24. Evaluation Environment • Classroom environment (20m x 30m with minor radio blocking obstacles such as desks and chairs) • 2 Macbook Pro as fNodetestbeds • SunSpot and Iris nodes as sensor devices • 7SunSpot nodes • 6 Iris motes • Deploy sensor nodes with two topologies: linear and hybrid Iris mote Classroom SunSpot node

25. Testbed Topology • Linear topology • Non-overlapping sensing regions • Hybrid topology • Overlapping sensing regions • Experimental assumption • Same MTU • Cover range: 5m • Minor radio blocking obstacles • No wave noise

26. Evaluation Method • Deploy the sensor nodes and a sink node with the linear and hybrid topologies • 7 SunSpot and 6 Iris nodes • Deploy at arbitrary positions • Measure the sum of hops, the power usage and the packet transmission latency of each node that is necessary to transmit its packets to the sink node • Average of 20 times measurement • Use fMap to calculate positions of fNodesand deploy fNodes with existing sensor nodes, and perform the same measurement as above

27. Experimental Results • Total number of transmitted packets • Reduced by approximately 30% • Linear Topology • 24 hops (Original) vs. 18 hops (Using fNode): 6 hops decreased • Hybrid Topology • 30 hops (Original) vs. 20 hops (Using fNode): 10 hops decreased

28. Power consumption Result fNode’sPower comsumption = (SunSpot + Iris)/2 • Linear Topology • 688 mA (Original) vs. 448 mA (Using fNode): decreased by 33% • Hybrid Topology • 875 mA (Original) vs. 531 mA (Using fNode): decreased by 39% Transmission Power – reduced 33-39%

29. Latency Result • Linear Topology • 381 ms (Original) vs. 167 ms (Using fNode): decreased by 50% • Hybrid Topology • 386 ms(Original) vs. 251 mA (Using fNode): decreased by 35% • Linear (50%) vs. Hybrid (35%) • Delay time • SunSpot : 23ms • Iris mote: 4ms • Linear relays via Iris mote more than Hybrid does Delay time – reduced 35-50%

30. Related Work • A Framework for Flexible Packet Processing in Heterogeneous Sensor Networks (M. Leogrande, C. Pastrone… at FGCN 07) • Base on XML language • Flexible packet processing • Increase in flexibility, adaptability and extensibility • However, it only focuses on processing messages • Packet transmission overhead is still unsolved • Adaptive Online Energy Saving for Heterogeneous sensor networks (Qiu, J. Hu, E. Sha,at 19th IASTED ) • Base on time interval • Obtain the best mode assignment for each node • Adjust online • However, availability of WSN is decreased

31. Conclusion And Future Work • Propose fNode • Forward packets of different communication architecture • Packet transmission overhead, power usage and latency • Reduced approximately 30% • Application areas • Building management system • Greenhouse management system • Evaluate the benefit of fNodeis only the first step • Our future work • Implement a realistic fNode • Deploy under a larger scale WSN