ENSEIRB - Advanced Project
- 1. SRWSN: Semantic Routing on Wireless Sensor Networks HUMIDITYSRWSNTEMPERATURESOUNDPRESSURELIGHT1Semantic Routing on Wireless Sensor Networks
- 2. IntroductionConventional routing requires the destination address to be known.
- 3. In a scenario where the network is expected to learn its environment, the algorithm has to be flexible.
- 4. Wireless Sensor Networks.2Semantic Routing on Wireless Sensor Networks
- 5. ContextWireless sensor networks cannot afford to have heavy routing tables.
- 6. The goal is to fulfill a request on a network without knowing its topology.3SRWSNTEMPERATURE?I DO!Semantic Routing on Wireless Sensor Networks
- 7. Outline4Semantic Routing on Wireless Sensor Networks
- 8. Outline5Semantic Routing on Wireless Sensor Networks
- 9. Problem definition Semantic Routing
- 10. All the metadata associated with a given message can be applied to a semantic reasoning engine in order to forward this data to the most appropriate receivers, over the most appropriate transmission media.
- 11. A common ontology has to be defined.6Semantic Routing on Wireless Sensor Networks
- 12. Problem definition Wireless Sensor Networks
- 14. Communications are 10 times more energy consuming than an internal process.
- 15. New applications
- 16. Surveillance
- 17. Nodes have to act more collaboratively. Using semantics to improve the process inside the network.7Semantic Routing on Wireless Sensor Networks
- 18. Outline8Semantic Routing on Wireless Sensor Networks
- 19. ScenarioSupervising a large property while the owners are absent:9Semantic Routing on Wireless Sensor Networks
- 20. Specifications Alphabet10Semantic Routing on Wireless Sensor Networks
- 21. 11Semantic Routing on Wireless Sensor NetworksMsg TypeQuery TypeQuery IdSrcIDSrcMACDstMACOptionsSpecifications Packet Type
- 22. MsgType : DISCOVERY, NORMAL STATE, ALERT
- 23. QueryType : REQUEST or ANSWER
- 24. QueryId: TEMP_VAL, TEMP_ERR, etc...
- 25. SrcID : id of the query source
- 26. SrcMAC : source MAC address (of the neighbor sender)
- 27. DstMAC : destination MAC address (of the neighbor receiver)
- 28. Options :
- 30. VALUE
- 31. ERROR
- 32. ALERT_GEN
- 33. ALERT_TIMESTAMP
- 36. Distinctly identify a node according to its direction and distance from the sink.
- 37. ID format:Direction MAC Last Byte Hops To Sink12A B C DSemantic Routing on Wireless Sensor Networks
- 38. Semantic routing tools adaptationBloomTable
- 39. Based on « peer content »
- 40. Contains all the Bloom filters of the neighborhood
- 41. Quickly know its neighbor abilities
- 42. Payload reduced
- 43. No three way handshake to discover the abilities
- 44. Store information in an efficient way13Semantic Routing on Wireless Sensor Networks
- 45. Bloom FiltersRepresent a set of n elements in a compact form
- 46. Detect the presence of an element0011SS010S11g(si)P11P110h(si)P21P2Sn-100Sn014Semantic Routing on Wireless Sensor Networks
- 47. Semantic routing tools AdaptationLearning Table
- 48. Based on the « query history »
- 49. Mimicking human actions in a social network
- 50. Process
- 51. Observe query responses sent to peers
- 52. Memorize the information (only by listening to traffic) according to its reliability
- 53. Use the information to select the most relevant peers15Semantic Routing on Wireless Sensor Networks
- 54. Learning MechanismNodeLearningPeerSelectionRELEVANT PEERQUERYIDQueryTypesLearningTableNeighborhood TableANSWER16Semantic Routing on Wireless Sensor Networks
- 55. Query Types Defines the similarities between the queries according to different scenarios:……………17Semantic Routing on Wireless Sensor Networks
- 56. Algorithm MaturityTime18Semantic Routing on Wireless Sensor Networks
- 57. DeploymentBloomTableBloomGlobalBloomPerso19Semantic Routing on Wireless Sensor Networks
- 58. AlgorithmHandleMessage() function20Semantic Routing on Wireless Sensor Networks
- 59. Algorithm21SendResponse1545ForwardResponseForwardRequest1256SendRequest1234Semantic Routing on Wireless Sensor Networks
- 60. Algorithm (intern)22NodeQUERYIDPEERRelevant PeerBloomTableNoReliable PeerLearningTableNoRandomSelectionRandom PeerSemantic Routing on Wireless Sensor Networks
- 61. Alert managementAlert message contains :
- 62. Value
- 63. Error
- 64. Alerts are directly (hop by hop) sent to the sink
- 65. If the neighborhood does not record any alert then the alert is ignored
- 66. Receiving a false alert message
- 67. Update its « reference values »23Semantic Routing on Wireless Sensor Networks
- 68. Outline24Semantic Routing on Wireless Sensor Networks
- 70. Omnet++
- 71. Why?
- 72. ZigBeeunderTest
- 73. Abstraction of material problems (such as memory management)
- 74. Easy to add components
- 75. Other simulators
- 76. NS-2
- 77. TOSSIM
- 78. …25Semantic Routing on Wireless Sensor Networks
- 79. WorkImplementing the network layer
- 80. Implementing the tools
- 81. Wiki :
- 82. Algorithm
- 84. SVN
- 85. How to participate in our projecthttp://rsimogaetan.online.fr/pfawiki/index.php5?title=PFAWiki26Semantic Routing on Wireless Sensor Networks
- 86. Outline27Semantic Routing on Wireless Sensor Networks
- 88. Bloom filter
- 89. Error probability : 10%
- 90. Alphabet size : 20 words
- 91. Hash functions : 2
- 92. Vector length = 96 bits = 12 Bytes = Stored size
- 93. Without using Bloom filters
- 94. Stored size ≈ 1200 bits Compression Ratio ≈ 8%28Semantic Routing on Wireless Sensor Networks
- 96. Nodes : 25
- 97. Average number of neighbors : 3
- 98. Scope of a node : 100 m Time to deploy : 150 s Message size: 120 bits29Semantic Routing on Wireless Sensor Networks
- 99. Performances30Semantic Routing on Wireless Sensor Networks
- 101. Random node generates a request
- 102. TTL = 4
- 103. Network learns from communications 31Semantic Routing on Wireless Sensor Networks
- 105. A learning network .32Semantic Routing on Wireless Sensor NetworksGain of 50%ReceivedrequestsSimulation time
- 106. Outline33Semantic Routing on Wireless Sensor Networks
- 107. ConclusionImplementation of an adaptative semantic routing algorithm.
- 108. SRWSN meets the Wireless Sensor energy and storage constraints.
- 109. SRWSN is quickly deployed although nodes are randomly set.
- 110. SRWSN learns from its environment and can adapt to new query types.34Semantic Routing on Wireless Sensor Networks
- 111. Further work Learning events
- 112. Recording the succession of alerts for one event
- 113. For instance : Fire = Light Temperature
- 114. Using ID’s form to direct traffic
- 115. Add location-based queries
- 116. Especially for queries from the sink to one node
- 117. Use another data-oriented algorithm in our scenario to show SRWSN comparison
- 118. Alert management improvement
- 119. Using events
- 120. Predict a temperature alert in case of a previous light one 35
- 121. Thank youHUMIDITYQuestionsTEMPERATURESOUNDPRESSURELIGHT36Semantic Routing on Wireless Sensor Networks