RateAdaptive Reading in COTS RFID Systems Qiongzheng Lin Lei Yang Huanyu Jia Chunhui Duan Yunhao Liu The Hong Kong Polytechnic University Tsinghua University ID: 1048584
Download Presentation The PPT/PDF document "Revisiting Reading Rate with Mobility:" 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. Revisiting Reading Rate with Mobility:Rate-Adaptive Reading in COTS RFID SystemsQiongzheng Lin, Lei Yang, Huanyu Jia , Chunhui Duan, Yunhao LiuThe Hong Kong Polytechnic UniversityTsinghua UniversityDec.14, 2017
2. 1. Background & MotivationMore info
3. TaTagReaderRFID Systems have been widely employed for various applications!1. What is RFID?Air ProtocolIDTechEx: 18.2 billion tags have been sold so far in 2017.
4. EPCglobal Gen 2 has become the widely adopted air protocol.1.1 The-State-of-Art: UHF RFID StandardsGEN 2
5. Reading Procedure: Framed Aloha Protocol1.2 The-State-of-Art: Anti-collisionQueryReaderTags1st frame12345Tags2nd frame12345Tags3rd frame12345
6. Gen 2 Reading Protocol: Q-Adaptive1.3 The-State-of-Art: Anti-collisionIt looks like that Q-adaptive is not efficient…Actually, Q-adaptive protocol almost approaches the optimal reading rate.
7. RFID assisted sorting system1.4 Real Case Unsorted and sorted tags can be easily distinguished by leveraging the physical layer signal patterns.However, such approach requires many readings of the mobile tags…
8. 1.5 Case StudyReading Trace & Trace DistributionReading TraceTrace DistributionInsight: A large number of readings are consumed on static tags, which are nearby the reader. 10% of the tags are read over 655 times Four hour real trace, acquired by one reader obtained up to 367, 536 readings from 527 tags. Tag #271 has been continuously read over 90,000 times!
9. 1.6 InsightsFact 1: The reading demands of mobile tags are considerably more urgent than those of static tags because the states of the static tags nearly remain unchangedFact 2: The participations of the static tags seriously affect the reading rates of the mobile tags.Reducing the total number of participating tags is a good way of improving the IRR of mobile tags.Moving tagsStatic tags
10. 1.6 Basic IdeaExploring the reading rate from a new perspectiveMobilityReal-time adjustment of the IRR according to tag’s current motion state.More readings are given to mobile tags
11. 2. Design of Rate-Adaptive Reading RFID SystemsMore info
12. System architecture: Two-phase reading design2.1 Overview mail@mail.com
13. System architecture: Two-phase reading design2.1 Overview mail@mail.comThe two phases constitute a basic cycle, which occurs alternatively and periodically.A periodical motion assessment is necessary to capture the state transitions of tags.The scheduling phase is longer than the assessment phase to guarantee target tags gain sufficient time to be read.
14. 2.2 Phase IPhase I: Motion AssessmentHow to find out mobile tags?
15. 2.2 Motion AssessmentUse of GMM to model the immobility of a tag.
16. 2.3 Phase IIPhase II: Target ScheduleHow to read target tags only?
17. 2.3 Target Schedule: Selective ReadingSelective reading allows us to read a subset of tags with a bitmaskSL
18. 2.4 Target Schedule: ChallengesGoal: find masks to cover all target tags with minimal cost0 0 1 1 1 00 1 0 0 1 01 0 1 1 0 11 0 0 1 0 11 1 0 1 1 0TargetsNon-targetsUsually, one mask cannot cover all targets.More masks are expected.
19. 2.4 Target Schedule: ChallengesGoal: find masks to cover all target tags with minimal cost0 0 1 1 1 00 1 0 0 1 01 0 1 1 0 11 0 0 1 0 11 1 0 1 1 0Mask 1: (10,4, 2)Mask 2: (11,3,2)1 1 0 1 1 0Unfortunately, one non-target tag is also selected!Plan ICollateral tags
20. 2.4 Target Schedule: ChallengesGoal: find masks to cover all target tags with minimal cost0 0 1 1 1 00 1 0 0 1 01 0 1 1 0 11 0 0 1 0 11 1 0 1 1 0Mask 1: (11,2, 2)Mask 2: (01,0,2)Plan IISet-cover optimization problem
21. 2.5 Target Schedule:Algorithmmail@mail.comGreedy algorithm based schedule: Hamming distanceCollateral tags
22. 3. Implementation & Evaluation More info
23. 3.1 ImplementationGen2 v.s. LLRP
24. 3.2 ImplementationCompletely using COTS RFID devicesImpinJ R420 reader. Directional antennaReaderTagAlien 2 × 2 InlayAlien Squiggle InlaySoftwareEPCglobal LLRPJava
25. 3.3 Evaluation of Phase IWe can find an appropriate detection threshold to achieve ≥ 0.95 TPR while ≤ 0.1 FPR using Phase-MoG.Detection ROCDetection sensitivityFalse Positive Rate True Positive Rate DistanceDetection rate
26. 3.4 Evaluation of Phase ILearning curveLearning timeAccuracy‘Quick start’
27. 3.4 Evaluation of Phase IISchedule Feasibility2/405/40
28. 3.4 Evaluation of Phase IISchedule Cost & IRR GainIRR Gain
29. 3.5 Case StudyReal-time TrackingWithout Rate-adaptiveWith Rate-adaptive
30. ConclusionWe present a system (aka Tagwatch ) for the rate-adaptive reading of mobile tags through selective reading.Tagwatch can successfully identify mobile tags with a mean probability of 80% as long as a tag moves beyond 1cm. It also achieves 95% of accuracy for motion detection, whereas the false positive rate (FPR) is maintained at below 10%.Tagwatch can outperform the individual reading rates of mobile tags by a median of 3.2X and 1.9X when there are 5% and 10% mobile tags.
31. ThankYou!