Chunyi Peng Guobin(Jacky) Shen, Yongguang Zhang, Yanlin Li, Kun Tan Microsoft Research Asia.


58 views
Uploaded on:
Category: Funny / Jokes
Description
A High Accuracy Acoustic Ranging System utilizing COTS Mobile Devices ... Sample estimation of the lower bound on COTS cell phones (HP iPAQ rw6828) ...
Transcripts
Slide 1

BeepBeep: A High Accuracy Acoustic Ranging System utilizing COTS Mobile Devices Chunyi Peng Guobin(Jacky) Shen, Yongguang Zhang, Yanlin Li, Kun Tan Microsoft Research Asia

Slide 2

Ranging, fundamental to limitation Many incredible frameworks Cricket (Mobicom\'00) RIPS (Sensys\'05) ENSBox (Sensys\'06) … Hardware and algorithmic development

Slide 3

Our inspiration Proximity recognition between gadgets Among convenient gadgets MobiUS (Mobisys\'07) Between versatile and non-compact gadgets Phone to PC, Xbox, Printers, Projectors …

Slide 4

The necessity A broadly relevant arrangement Work on COTS gadgets No extra equipment (e.g., ultrasound) Pure client space programming (no change to OS/driver) Not reliant on foundation Applicable in unconstrained, impromptu circumstances Minimum arrangement of sensors High precision!

Slide 5

A short time estimation Mostly, taking into account time-of-flight estimation Distance = speed x time Sound regularly picked Slower speed => less requesting on time exactness Still, a testing errand 1 ms blunder in time = 34 cm mistake in separation 1 cm going precision requires 30us planning exactness Extremely difficult on COTS/programming

Slide 6

The main driver of error – three vulnerabilities Clock synchronization instability Sending vulnerability Receiving instability programming issuing summon programming mindful of entry ... t0 = wall_clock(); write(sound_dev, signal); ... ... read(sound_dev, signal); t1 = wall_clock(); ... obscure postponements (programming, framework, driver, equipment, … ) ? obscure deferrals (equipment, interfere with, driver, planning, … ) ? sound leaves speaker sound spans mic time

Slide 7

Effects of the sending and accepting vulnerabilities Example estimation of the lower bound on COTS cell phones (HP iPAQ rw6828) Highly fluctuating, seems unusual Easily indicates 1-2 ms (=> a couple of feet mistake) CPU unmoving CPU intensely stacked

Slide 8

Our methodology – BeepBeep A basic and compelling arrangement Each gadget simply needs to radiate a sound flag and record them all the while Only require a speaker, a mic, and some method for speaking with the other gadget Achieving 1cm exactness while fulfilling every one of the prerequisites

Slide 9

Beepbeep\'s fundamental strategy Device A Device B Device A transmits a beep while both recording Device B emanates another beep while both keep recording Both gadgets distinguish TOA of the two beeps and get separate ETOAs Exchange ETOAs and figure the separation A\'s recording B\'s recording ETOA An ETOA B D AB =|ETOA A - ETOA B |/2

Slide 10

Mathematical inference 1 st Beep t A0 d B,A - d A,A = c·(t B1 - t A1 ) t A1 d A,B - d B,B = c·(t A3 - t B3 ) t B1 ETOA An ETOA B 2 nd Beep t B2 d B,A +d A,B = c·[(t A3 - t A1 )- (t B3 - t B1 )] +d A,A +d B,B = c·(ETOA A - ETOA B )+d A,A +d B,B t B3 t A3

Slide 11

Key procedures, impacts and justification (I) Self-recording Record signals from both the other party and itself Establish the beginning reference purpose of the entire running procedure Duplex sound channel Two-way detecting Avoid clock synchronization instability To catch the consummation reference purpose of the entire going procedure not endeavor to catch any framework time information

Slide 12

Key strategies, impacts and reason (II) Sample checking Avoid alluding to framework timekeepers for timing data Dedicated A/D converter, w/altered examining rate Achievable accuracy is dictated by the inspecting recurrence: 0.8cm at 44.1kHz inspecting rate Putting together: Bypass all the three instabilities by setting aside a few minutes estimation insignificant to framework tickers

Slide 13

Engineering Challenges (I) Signal Design Good flag outline helps recognition Easily discernible in computerized recording Robust against surrounding clamor Robust against acoustic mutilation Low-devotion speaker & mic in COTS cell phone Within equipment capacity Most COTS gadgets have restricted voice recurrence extend Our observational configuration ("trill" sound) 50ms long, moving recurrence from 2 to 6 kHz

Slide 14

Engineering Challenges (II) Signal Detection Algorithm Design Efficient and quick flag identification calculation Quickly find conceivable sign districts Robust against low SNR Utilize commotion floor to support SNR Combat multipath impact Multipath: huge issue indoor environment We inferred exceptional calculation to identify first "sharp pinnacle" signal relationship

Slide 15

Engineering Challenges (III) Protocol plan Coordinate two (or more) gadgets in whole going procedure Minimize going time term Device structure component Speaker/mic\'s position influence going results Vary starting with one gadget display then onto the next Need alignment to change going computation

Slide 16

System Implementation Platform: Windows Mobile 5.0 Sound: "wave" API Communication: "WinSock" (WiFi or Bluetooth) Software engineering User-mode dynamic linkable library As administration for different applications Test gadgets Dopod838 HP iPAQ 6828

Slide 17

Evaluation Case-An: Indoor, calm Case-B: Indoor, loud Case-C: Outdoor, auto park passage Case-D: Outdoor, tram station 50 runs every setting

Slide 18

Summary Identified three vulnerabilities and relieved them with three key innovations Two-way detecting Self-recording Sample tallying BeepBeep gives a basic going arrangement Achieves 1cm precision On extremely essential equipment set Purely in programming (client space)

Slide 19

Thanks! Demo session. Welcome to attempt BeepBeep! Programming downloads: http://research.microsoft.com/wn/beepbeep.aspx

Slide 20

Backup

Slide 21

Lower bound of sending and accepting vulnerabilities

Recommended
View more...