2 Zhichao Cao 2 Mingyan Liu 3 Mengshu Hou 1 and Yujun Li 1 SECON 2015 Seattle CDMAC A Contention Detectable MAC for Low DutyCycled Wireless Sensor Networks 1 ID: 1002265
Download Presentation The PPT/PDF document "Daibo Liu 1 , Xiaopei Wu" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. Daibo Liu1, Xiaopei Wu2, Zhichao Cao2, Mingyan Liu3, Mengshu Hou1 and Yujun Li1SECON, 2015, Seattle CD-MAC: A Contention Detectable MAC for Low Duty-Cycled Wireless Sensor Networks1 University of Electronic Science and Technology of China2 Tsinghua University 3 University of Michigan1
2. OutlineAbout this paperMotivationRelated works and overview of this methodDesign of CD-MACSystem implementationEvaluationSummary of this work2
3. 3About This PaperCD-MACContention detectable mechanismCollision avoidance and resolution in low-power duty-cycled networks that experience traffic burstsCopes with hidden terminals and designs for receiver-initiated duty-cycled protocolsContributionPractical mechanism to avoid contentionAdaptive to both low/burst traffic scenariosReal implementation in TinyOS-2.1.1
4. OutlineAbout this paperMotivationRelated works and our approachDesign of CD-MACSystem implementationEvaluationSummary of this work4
5. Motivation5Wireless Sensor Networks (WSNs)Energy restricted Duty cycle operating & Asynchronous workLarge number of nodesReceiver-initiated MAC (LPP)Reduce preamble consumption of LPLRI-MAC, A-MACAggregated multiple transmissions in short timeData collisionTraffic PeaksEvent detection, Network code update, Bulk downloadData collection in large networkProbeDataAck+ProbeReceiverSender
6. OutlineAbout this paperMotivationRelated works and our approachDesign of CD-MACSystem implementationEvaluationSummary of this work6
7. Related WorksCollision resolutionNeed collision detection Difficult to detect by an accurate and light-weight schemeLong resolution timeE.g., strawman (IPSN2012), staris (infocom2014).Collision avoidanceCarrier sense and backoff windowsRI-MAC, A-MAC, etc.Not good for burst trafficTDMA fashion Inflexible and inefficient for low traffic scenarios 7Can we avoid the collision just before it happens by guaranteeing the efficiency in different traffic loads?
8. Orderly Transmission for Collision AvoidanceSome challenges:Receiver solicitationSenders data report: fast and orderlyData polling according to reportsApproaches:Time diversitySlot & Virtual IDReceiver polling data8Need highly precise time synchronization in an asynchronous network!
9. Our weapon: Time diversity9The observations:Acknowledgement (ACK) can be accurately delayed to sendInstruction-level time accuracyAn ACK in time domain can be used to report a pending data and denote a specific senderLittle on-air time,Distinguishable transmitting timeChildren nodes synchronize according to the probe of the central receiverTime can be slotted after a reservation probeACKs can be sent in different slots
10. OutlineAbout this paperMotivationRelated works and our approachDesign of CD-MACSystem implementationEvaluationSummary of this work10
11. Design of CD-MACOverview11Slotted ACKsChange the number of slots and distinguish potential sendersData lossACK lossSender reservation stagePacket polling stageCollision resolution stagePolling in sequence
12. Sender ReservationCall for transmissionReceiver initiatedReservation probeTake turns to reportReply with an ACKSlotted time12CRS1CS2CAASender reservation stageReported ACKsSlot sizeReservation probeReceiver uses an ACK’s receiving time to infer the slot sequence number selected by a sender.AASenders simultaneously synchronize with receiver CCAAThe number of received ACKs denotes the number of potential senders.
13. Inference of Slot Sequence Number13Receiver infers the slot sequence number (Ks) of each received ACK by: Tbase: constant Tslot : slot size Ts : time for waiting ACKCRS1CS2CAAAAHow can senders reply ACKs in different slots to avoid collision?
14. Slot Assignment12Receiver endProposal of the number of slots (n)Auxiliary parameter (m) for senders’ slot selectionn and m are adaptive to traffic load Limited number of slotsSender endDetermine its slot position in distributed fashionUsing n, m, and the appointed selection approachLow ACK collision rate
15. Slot selection approach πInput: sender’s ID s, m, nReceiver’s selection of m, nRestriction on n and mInput: children nodes’ IDs, n and mGreedy search to find the optimalIA= 1 if A is true and 0 otherwiseComputation complexity: C is the number of children nodesSimulation of effectivenessACK collision rateAssign each n an optimal mSenders compute slot sequence with π15Slot AssignmentFor each n, CD-MAC computes an optimal m to minimize ACK collision.NS: Number of children nodesw/ m: assign optima m for nPercentage that multiple senders select a same slot
16. Selection the Number of SlotsSelection of n consists of two stages Initial n for node’s wakeupChanged n to distinguish slot-collision sendersInitial nAdapt to traffic load of previous wakeupUsing Moving average to updateChanged nAfter polling all the data of applicants Select a different n to give reservation opportunityCD-MAC: just decrease 116
17. Data PollingData polling according to ACK slotRegard ACK slot as sender’s virtual addressAcknowledgement & solicitation dataRestart reservation probe to against packet lossSleep when no ACK report17RSCCAA11DataDataSCA1DataAA 3A 3DataA 3A CA CACKSlot sequenceReservation probeOrderly Data polling
18. OutlineAbout this paperMotivationRelated works and our approachDesign of CD-MACSystem implementationEvaluationSummary of this work18
19. System Implementation19CD-MACTinyOS-2.1.1Built upon system codes of cc2420Interface for data transmitting/receivingSystem parameters:
20. OutlineAbout this paperMotivationRelated works and our approachDesign of CD-MACSystem implementationEvaluationSummary of this work20
21. EvaluationEvaluation setup Indoor testbedOne-hop network, multiple-hop network Repeat at least 5 timesOne-hop collisionRelated protocols: A-MAC (LPP), X-MAC (LPL), Strawman (IPSN12), CTPMultiple senders One receiverMultiple-hop NetworkCTP+A-MAC CTP+CD-MACCTP+X-MACPerformance: reliability, delayHidden terminal avoidanceA-MAC, X-MAC, CD-MAC21
22. One-hop Network22Mean single-hop delayMean single-hop throughputMultiple senders, one receiverSleep interval and inter-packet interval are set to 128msThe average single-hop transmission delay of CD-MAC is fur less than that of A-MAC and A-MAC+Strawman. CD-MAC provides the higher throughput than A-MAC, A-MAC combined with Strawman, and X-MAC, by 172%, 58.6%, and 42.3%, respectively.
23. Table II. CTP performance over X-MAC, A-MAC and CD-MAC with bursty traffic Multi-hop Network 23Indoor testbed: 30 TelosB nodes, 4+ hops networksPeriodic traffic (PT):Generates a data packet every 1 minutesBursty traffic (BT):One data packet each on 10 randomly-selected nodesTable I. CTP performance over X-MAC, A-MAC and CD-MAC with periodic traffic In low periodic traffic, CD-MAC with almost the same radio duty cycle, and with a slightly high PDR.In bursty traffic, CD-MAC performs better than both radio duty cycle and PDR.
24. Hidden Terminal Avoidance24Two types of topologies: With hidden terminalWithout hidden terminalA receiver, two senders CD-MAC outperforms X-MAC and A-MAC in handling hidden terminal. 795%11%78%16.8%83.2%33.7%30.4%27.2%10.7%8.2%19.8%19.3% CD-MAC is resilient against hidden terminals because it can differentiate the potential hidden nodes using different slot number.
25. OutlineAbout this paperMotivationRelated works and our approachDesign of CD-MACSystem implementationEvaluationSummary of this work25
26. SummaryCollision avoidance and resolution in low-power duty-cycled sensor networks Detect potential senders by reservationDistributed slot assignment to avoid ACK collisionSlot inference and virtual addressData polling strategiesReal system implementation in TinyOS2.1.1Evaluation in real testbed26
27. 27Thank you! Q&A