1 / 33

ArrayTrack: A Fine-Grained Indoor Location System

ArrayTrack: A Fine-Grained Indoor Location System. Jie Xiong, Kyle Jamieson University College London April 3rd, 2013 USENIX NSDI ‘13. Outdoors: GPS Accurate for navigation (meters) Signals fade in indoor environments Precise and rapid indoor location enables:

sadiew
Download Presentation

ArrayTrack: A Fine-Grained Indoor Location System

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. ArrayTrack: A Fine-Grained Indoor LocationSystem Jie Xiong, KyleJamieson University CollegeLondon April 3rd,2013 USENIX NSDI‘13

  2. Outdoors:GPS • Accurate for navigation (meters) • Signals fade in indoorenvironments • Precise and rapid indoor locationenables: • Augmented reality on the smartphone, wearable orglasses • Fine-grained location in supermarkets, libraries ormuseums • Controlling network access based on desk orroom • Known technologies: not accurate enough (WiFi), require dedicated infrastructure (ultrasound) or require cameras and good light conditions (vision) Precise location systems areimportant 2

  3. Timeline of indoor locationsystems 1 m(PinLoc) Sen etal 3 cm (Bat) Ward etal 2-3 m(RADAR) Bahl etal 39-51cm(Horus) Youssef etal 2 m(EZ) Chintalapudi etal 23 cm (ArrayTrack) 2012 1992 2004 2008 2013 2010 1997 2005 2000 time 97 cm(PinPoint) Joshi etal 1.2-4 m(Zee) Rai etal 30 cm(image-based) Hile andBoriello 5-10 cm(Cricket) Priyantha etal 5.4 m(TIX) Gwon etal >1 m(Badge) Want etal 3

  4. Increasing number of antennas on an access point(AP) • 802.11n MIMO links: improve capacity andcoverage • Draft 802.11ac (2014): 8 MIMO spatial streams (8antennas) Two observations aboutWiFi 4antennas 6antennas 14antennas 16antennas MotorolaAP8132 Cisco Aironet3600 Cisco Aironet1250 RUCKUS ZoneFlex7982 XirrusXR7630 4

  5. Increasing number of antennas on an access point(AP) • 802.11n MIMO links: improve capacity andcoverage • Draft 802.11ac (2014): eight MIMO spatial streams • WiFi is ubiquitous and denselydeployed • WiFi is now available on airplanes, subways andbuses • APs density is ever-increasing in the urbanenvironment Two observations aboutWiFi 5

  6. AP overhears a client’stransmission • AP leverages multiple antennas to generate physical angles of arrival (AoA) of a client'ssignals: • – AoA spectrum: power versusbearing at oneAP OurApproach AP1 Client • With multiple APs, central server synthesizes AoA spectra to obtain a location estimate for theclient AP2 6

  7. Basic theory ofoperation x1 Q AP x1 2πd/λ λ I d Client Measuredbaseband signal atAP 7

  8. Basic theory ofoperation AP λ/2 Q x1 x1 x2 2πd/λ θ d θ I ½λ sinθ x2 Client In a solely line-of-sight environment, phase measurements give client’s bearing to APθ x x1 ⎞ ⎛ arcsin 2 ⎜ ⎝ ⎟ ⎠  8

  9. Problem #1: Strong multipath reflectionsindoors • Problem #2: Direct path attenuated or completelyblocked • – Direct path signal may not be thestrongest The challenge: multipathreflections AoAspectrum Wall Array Client AP 0.2 0.4 0.6 0.8 Furniture 1

  10. Key observation: direct path bearing is more stable than reflection path bearings when client movesslightly ArrayTrack’s multipath suppressionalgorithm array AP Client

  11. Given: AoA spectra from two nearbylocations Find the peak bearings in each AoAspectrum Discard any peak not paired with a peak in the other AoAspectrum ArrayTrack’s multipath suppressionalgorithm ✔ Two peak bearings within five degrees are consideredpaired

  12. Content of packet and modulation type do not matter • Works with any part of apacket • – ArrayTrack utilizes the most robust preamblepart Step 1: detection andrecording 800ns 3.2µs 3.2µs G 10 short training symbols two long trainningsymbols Preamble 12

  13. N=1 N=5 190 • Very small partof a packetneeded 190 Step 1: detection andrecording 6012060 120 30 15030 180 0 150 0.5 0.5 – For a 40 MHz samplingrate, one sample is 25ns 0 180 • In the absence of noise, onesample works • Employ multiple samplesfor • 30 • averaging to removenoise • 0 N=10 190 N=100 190 300 240 60 120 60 120 150 30 150 0.5 0.5 1800 180 Preamblepart

  14. Diversity synthesis: existing 802.11 radios record the 1st half of the preamble from antenna 1 and the 2nd half from antenna2 • ArrayTrack’s diversity synthesis algorithm • Record 10 samples from the first preamble half and another 10 samples from the second preamble half with differentantennas • Double the number of antennas we can utilize forArrayTrack • antenna 1 Step 1: detection andrecording antenna2 Port1 Radio Port2 Preamblepart

  15. MUSIC algorithm [Schmidt, 1986] for AoA spectrumestimation • – Does not work well for indoor environment because of coherentsignals: • Transmitter • Receiver (Client) • (AP) • Spatial smoothing (SS) [Shan et al, 1985] handles coherentsignals Step 2: AoA spectrumgeneration NO spatial smoothing(SS) SS with 2 sub−arraygroups 90 90 1 120 60 120 60 1 0.5 0.5 30 150 30 150 x1 x2 x3 x4 x5 x6 x7 x8 180 180 0 0

  16. Step 3: AoA spectrasynthesis • N APs generate N AoAspectra P(x1)=0.45 AP1 • For a random position X, the likelihood of being atX is a multiplicationof • probabilities from multipleAPs 0.2 P(x) = P(x1) *P(x2) 0.4 0.6 0.8 X 1 P(x2)=0.6 1 0.8 0.6 0.4 0.2 AP2 16

  17. Step 4: search for highest probabilityposition

  18. AP: two WARPs, each with four radio boards (eight antennas) • Custom FPGA design using Xilinx System Generator for packet synchronization, diversity synthesis, RF oscillator synchronization • 4-16 antennas placed in a linear arrangement, spaced at λ/2 (6.13cm) • Clients: Soekris boxes equipped with 802.11radios • Backend location server: implemented in Matlab (1,000+LoC) Implementation

  19. Backend server has knowledge of each AP’slocation Floorplan: client and APpositions AP N 6 1 4 Client 3 2 5

  20. How accurate is MUSIC +SS? • ArrayTrack’s multipath suppressionimprovement • Effect of number of antennas on eachAP • Effect of client-AP differences inheight Evaluation

  21. Heatmap example of increasing number ofAPs Effects of number ofAPs twoAPs oneAP threeAPs fourAPs fiveAPs sixAPs

  22. In general, with increasing number of APs, more accurate location information can beobtained MUSIC + SS achieves 26 cm accuracy 1 6 APs (MUSIC +SS) 5 APs (MUSIC +SS) 4 APs (MUSIC +SS) 3 APs (MUSIC +SS) 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 CDF 0 1 5 10 20 50100 Location error(cm) 50010002500 22

  23. How accurate is MUSIC +SS? • ArrayTrack’s multipath suppressionimprovement • Effect of number of antennas on eachAP • Effect of client-AP differences inheight Evaluation

  24. 1 • Median: 23 cm (ArrayTrack with 6APs) Multipath suppression improvesaccuracy 0.5 ArrayTrack (6 APs) MUSIC + SS (6APs) 0 0 500 1000 150020002500 1 ArrayTrack (5 APs) MUSIC + SS (5APs) 150020002500 • With multipath suppression, thelong tail isremoved 0.5 0 0 500 1000 1 CDF 0.5 ArrayTrack (4 APs) MUSIC + SS (4APs) 0 0 500 1000 150020002500 • The fewer APs, the more important is multipathsuppression 1 0.5 ArrayTrack (3 APs) MUSIC + SS (3APs) 0 0 500 1000150020002500 Location error(cm)

  25. On average, 6 APs present the bestresult • It’s not true for a particularposition Optimal subset ofAPs 1 0.9 0.8 0.7 2.5cm 23cm 0.6 CDF 0.5 0.4 0.3 ArrayTrack (6 APs) Optimal subsets ofAPs MUSIC + SS (6 APs) MUSIC + SS (3APs) 0.2 0.1 0 1 5 102050100500 Location error (cm) 2500 25

  26. Effect of number of APs onaccuracy • Multipath suppressionimprovement • Effect of number of antennas on eachAP • Effect of client-AP differences inheight/orientation Evaluation 26

  27. ArrayTrack 4−antennaAPs ArrayTrack 6−antennaAPs 1 Number of antennas atAP 0.9 ArrayTrack 8−antennaAPs 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 CDF 0 10 20 50 100 Location error(cm) 1 500 27

  28. Effect of number of APs onaccuracy • Multipath suppressionimprovement • Effect of number of antennas on eachAP • Effect of client-AP differences inheight Evaluation

  29. High accuracy despite AP-client heightdifference Different antennaheights 1 ceiling Original 0.8 0.6 3m 1.5m CDF 0.4 0.2 0 1 10 20 Location error(cm) 100

  30. Other characteristics ofArrayTrack Smalllatency (1-3 packets needed) Robustagainst lowSNR Robustagainst collision

  31. ArrayTrack: a robust, fast and responsive localization system with a median accuracy level of 23 cm (6 APs) and one meter (3APs) • Novel multipath suppression and diversity synthesisalgorithms • Uses only the WiFi infrastructurenearby • Robust against low SNR and packetcollisions • Fast and responsive: requires only 1-3packets • Three dimensional tracking with two-dimensional array for futurework Conclusions Thankyou!

  32. Wire connects WARPs to share the same sampling clock and RF oscillator • USRP2 calibrates WARPs to remove WARP internal phaseoffsets • Remove phase offsets due to hardwareimperfections • Cables labeled with the same lengths are not exactly the same • SMA splitters are not fullybalanced Implementationchallenges

  33. Circularly-polarized antennas mitigate the performancedrop AP-client antennaorientations Different antennaorientations 1 Different antennaheights Original 0.8 0.6 CDF 0.4 0.2 0 10 20 Location error(cm) 1 100 33

More Related