1. A Survey on Routing Protocols for Wireless Sensor Networks Kemal Akkaya & Mohamed Younis By Yalda Edalat

2. 2 Introduction - Routing No global addressing Redundant data traffic Multiple-source single-destination network Careful resource management – Transmission power – On-board energy – Processing capacity – Storage

3. 3 Introduction - Constraints Limitations – Energy Constrains – Bandwidth All layers must be energy aware Need for energy efficient and reliable network routing Maximize the lifetime of the network

4. 4 System Architecture & Designing Network Dynamics – Mobile or Stationary nodes – Static Events (Temperature) – Dynamic Events ( Target Detection) Node Deployment – Deterministic – Placed manually – Self-organizing – Scattered randomly

5. 5 System Architecture & Designing Energy Considerations – Direct vs Multi-hop communication Direct Preferred – Sensors close to sink Multi-hop – unavoidable in randomly scattered networks Data Delivery Models – Continuous – Event-driven – Query-driven – Hybrid

6. 6 System Architecture & Designing Node Capabilities – Homogenous – Heterogeneous – Nodes dedicated to a particular task (relaying, sensing, aggregation) Data Aggregation/Fusion – Aggregation – Combination of data by eliminating redundancy – Data Fusion is Aggregation through signal processing techniques – Aggregation achieves energy savings

7. 7 Taxonomy Classification of Routing Protocols – Data Centric – Hierarchical – Location-based – Network Flow & QoS Aware

8. 8 Data-centric Protocols Sink sends queries to certain regions and waits data from sensors located in that region Attribute-based naming is necessary to specify properties of data

9. 9 Data-centric Protocols Flooding Gossiping Sensor Protocols for Information via Negotiation (SPIN) Directed Diffusion Energy-aware Routing Rumor Routing Gradient-Based Routing (GBR) Constrained Anisotropic Diffusion Routing (CADR) COUGAR ACtive QUery forwarding In sensoR nEtworks (ACQUIRE)

10. 10 Hierarchical Protocols Maintain energy consumption of sensor nodes – By multi-hop communication within a particular cluster – By data aggregation and fusion  decrease the number of the total transmitted packets

11. 11 Hierarchical Protocols LEACH – Low-Energy Adaptive Clustering Hierarchy Power-Efficient GAthering in Sensor Information Systems (PEGASIS) – Hierarchical PEGASIS Threshold sensitive Energy Efficient sensor Network protocol (TEEN) – Adaptive Threshold TEEN (APTEEN) Energy-aware routing for cluster-based sensor networks Self-organizing protocol

12. 12 Hierarchical Protocols Advantages – Useful for applications which need communication of a specific node (e.g. parking-lot networks) – Small cost of maintaining routing tables – Energy Savings – Utilization of a limited subset of nodes – Scalable Disadvantages – Organization phase not on demand – Many cuts in the network increase the probability of applying reorganization phase

13. 13 Location-based Protocols Distance between two nodes is calculated using location information Energy consumption can be estimated – Efficient energy utilization Protocols designed for Ad hoc networks with mobility in mind – Applicable to Sensor Networks as well – Only energy-aware protocols are considered

14. 14 Location-based Protocols MECN & SMECN – Minimum Energy Communication Network GAF – Geographic Adaptive Fidelity GEAR – Geographic and Energy Aware Routing

15. 15 Network Flow & QoS-aware Protocols Network Flow: Maximize traffic flow between two nodes, respecting the capacities of the links QoS-aware protocols consider end-to-end delay requirements while setting up paths – Guaranteed bandwidth

16. 16 Network Flow & QoS-aware Protocols Maximum Lifetime Energy Routing Maximum Lifetime Data Gathering Minimum Cost Forwarding Sequential Assignment Routing Energy Aware QoS Routing Protocol SPEED

17. 17 Summary

18. Directed Diffusion: A Scalable and Robust Communication Paradigm for Sensor Networks C. Intanagonwiwat, R. Govindan and D. Estrin Presented by Yalda Edalat 18

19. Directed Diffusion In a Distributed Sensor Network System, how do you query the system for events of interest? Directed Diffusion is a data-centric approach. – Two Parts Query Propagation Data Propagation 19

20. Interest and Event Naming Query/interest: 1.Type=four-legged animal 2.Interval=20ms (event data rate) 3.Duration=10 seconds (time to cache) 4.Rect=[-100, 100, 200, 400] Reply: 1.Type=four-legged animal 2.Instance = elephant 3.Location = [125, 220] 4.Intensity = 0.6 5.Confidence = 0.85 6.Timestamp = 01:20:40 Attribute-Value pairs describe a task. 20

21. Diffusion (High Level) Sinks broadcast interest to neighbors (with large interval attribute, ex: 1 event per second) Interests are cached by neighbors Gradients are set up pointing back to where interests came from at low data rate Events flow towards originator among multiple paths Network reinforces one or a small number of paths 21

22. Directed Diffusion (Data) Nodes sensing events, search its interest cache for matching interest entry Data is forwarded according to the gradients associated with the interest Receiving node: – Finds matching entry in interest cache, no match – silent drop – Checks and updates data cache (loop prevention, aggregation) – Retrieve all gradients, and resend message 22

23. Example 23

24. Directed Diffusion (Reinforcement) Source sends low-rate events to sink across multiple paths Sink then reinforces one particular neighbor to draw down high quality/data rate events – Choose neighbor based on data driven rules – Use data cache Sink resend the interest with higher data rate to that reinforced neighbor. Results in high data rate path between source and sink 24

25. Negative Reinforcement How to prevent multiple paths from being reinforced?? Idea is to negatively reinforce paths – Time out high data rate gradients – Explicitly resend interest with low data rate Choosing neighbor – Neighbor from which no new events have come in the last T seconds. 25

26. Design Considerations Design space for diffusion 26

27. Repair Initiated by intermediate nodes on a reinforced path Reinforce alternate path and negatively reinforce degraded path Intermediate nodes may need to interpolate data to prevent all downstream nodes from initiating reinforcement 27

28. Evaluation ns-2 simulations, 50 nodes in 160*160 square sqm, 802.11 MAC – Network size up to 250 nodes – Constant network density Metrics – Average Dissipated Energy – Average Delay – Event Delivery Ratio 28

29. Metrics Average dissipated energy – Ratio of total energy expended per node to number of distinct events received at sink – Measures average work budget Average delay – Average one-way latency between event transmission and reception at sink – Measures temporal accuracy of location estimates Both measured as functions of network size 29

30. Average Dissipated Energy Directed Diffusion achieves better performance by: Reducing redundant data In-network aggregation 30

31. Average Delay Directed Diffusion finds least delay paths as Omniscient Multicast Reinforcement helps 31

32. Average Dissipated Energy in the presence of node failures Counter- Intuitive Energy efficiency improves due to the negative reinforcem ent pruning paths 32

33. Average Dissipated Energy with and without negative reinforcement Energy dissipation without NR is twice that with NR 33

34. Average Dissipated Energy with and without duplicate suppression Energy dissipation without suppression is between twice and 5 times that with suppression 34

35. Conclusions Directed Diffusion provides a data-centric communication protocol for sensor sources and sinks Its gains due to aggregation and duplicate suppression may make it more viable than ad-hoc routing in sensor networks Its performance needs more extensive evaluation before strong claims can be made Achieves better energy efficiency Application awareness 35

36. Scalable and Reliable Sensor Network Routing: Performance study from Field Deployment M. Nassr, J. Jun, S. Eidenbenz, A. Hansson and A. Mielke Infocom 2007 Presented by Yalda Edalat 36

37. Introduction Two important requirement for sensor networks: Scalability: limited bandwidth, energy and computational capacity Reliability: resiliency against changes in network status Problem: Reliability and scalability trade-off Introducing DTRP Proposed sensor network routing schemes: Gossiping (probabilistic transmission). MINTRoute (the standard routing protocol software for TinyOS). 37

38. Directed Transmission Routing Protocol (DTRP) DTRP is a multi-path proactive routing protocol Goal: provide improved scalability and reliability. DTRP is parametric and probabilistic: Parametric: provides a tunable parameter for transmission probability Probabilistic: tunable parameter affects reflooding probability Beacon packets provides the hop-count distance value between the sink and other sensor nodes. Not to determine the next hop node for the destination Sent periodically 38

39. Directed Transmission Routing Protocol (DTRP) d 1 : Shortest hop count distance between originating source and destination sink. (known at “S” and stored in packet) d 2 : Hop-count distance a packet traveled before reaching a node (stored in packet) d 3 : Shortest hop count distance between relaying node and destination sink (stored in R). 39

40. Directed Transmission Routing Protocol (DTRP) The propagation model: 40

41. Directed Transmission Routing Protocol (DTRP) 41

42. Implementation Hardware: Mica2 motes + Stargates (monitoring system) Software: TinyOS 1.1.15 Environment: Small network scenario (3×5 nodes) Large network scenario (3×10 nodes) Protocols: DTRP MINTRoute (Default single path routing protocol for TinyOS) Gossiping (fix probability set to 0.7) 42

43. Topologies Two topologies: Grid topology Random topology 43

44. Experiments The number of sources: 1, 3, 5, 7, 9 This results in different network load scenarios Single sink Experiment time : 10 minutes each Combination of 4 cases (small, large & grid, random) Two specific network performance measures: Packets delivery ratio (PDR) Total network load 44

45. Performance Comparison (I) Small Network (Packet Delivery Ratio) 45

46. Performance Comparison (II) Small Network (total network load) 46

47. Performance Comparison (III) Large network (Packet Delivery Ratio) 47

48. Performance Comparison (IV) Large network (total network load) 48

49. Conclusion DTRP is a more efficient approach than fixed probability Gossiping. DTRP satisfies a balance between reliability and scalability (intelligent intermediate method between flooding and single path). DTRP delivers higher packet percentage than MINTRoute and Gossiping at the same load (for lower network load cases) but the performance of MINTRoute for higher load networks tends to be better (because of single path). 49

50. 50