计算机科学与技术（参考文献）Device-Free Gesture Tracking Using Acoustic Signals

Device-Free Gesture Tracking Using Acoustic Signals Wei Wangt Alex X.Liut Ke Sunt tNanjing University,Michigan State University MobiCom'16 Oct4h,2016 日+回1三1交非三勺风 1/20

Device-Free Gesture Tracking Using Acoustic Signals Wei Wang† Alex X. Liu†‡ Ke Sun† †Nanjing University, ‡Michigan State University MobiCom’16 Oct 4th, 2016 1/20

Motivation System Desion Experiments Conclusions 市商 Motivation It is difficult to input on smart watches 口，01三13卡至QC 2/20

Motivation System Design Experiments Conclusions Motivation It is difficult to input on smart watches 2/20

Motivation System Design Experiments Conclusions Limitation of Prior Arts Google Soli designs specialized 60GHz chips for gesture input 日，+司1三13卡三分Q0 3/20

Motivation System Design Experiments Conclusions Limitation of Prior Arts Google Soli designs specialized 60GHz chips for gesture input 3/20

Motivation System Design Experiments Conclusions Problem Statment Can we build software-based to replace specialized hardware? and meet these design goals .High accuracy (mm-level) .Low latency (30 ms) Low computational cost (works on mobile devices) 。Low energy 日，+司1三13卡三分Q0 4/20

Motivation System Design Experiments Conclusions Problem Statment Can we build software-based to replace specialized hardware? . . . and meet these design goals • High accuracy (mm-level) • Low latency (< 30 ms) • Low computational cost (works on mobile devices) • Low energy 4/20

Motivation System Design Experiments Conclusions Problem Statment Can we build software-based to replace specialized hardware? ..and meet these design goals High accuracy(mm-level) ·Low latency(<30ms) Low computational cost (works on mobile devices) 。Low energy 日，+司1三13卡三分Q0 4/20

Motivation System Design Experiments Conclusions Problem Statment Can we build software-based to replace specialized hardware? . . . and meet these design goals • High accuracy (mm-level) • Low latency (< 30 ms) • Low computational cost (works on mobile devices) • Low energy 4/20

Motivation System Design Experiments Conclusions 市 Can we do that? 日，+司1三13卡三分Q0 5/20

Motivation System Design Experiments Conclusions Can we do that? 5/20

Motivation System Design Experiments Conclusions Basic Idea Acos 2nft CIC Dynamic 年c0s2nft Static Component component CIC -sin 2nft Combined Use plain cos wave rather than impulses Measure the phase rather than Doppler shifts Decompose the received signal in vector space rather than in the time/frequency domain 日+回1三1交非三勺风 6/20

Motivation System Design Experiments Conclusions Basic Idea cos 2πft —sin 2πft CIC CIC I Q Acos2πft I Q Combined Static component Dynamic Component • Use plain cos wave rather than impulses • Measure the phase rather than Doppler shifts • Decompose the received signal in vector space rather than in the time/frequency domain 6/20

Motivation System Design Experiments Conclusions Basic Idea Dynamic Component Acos 2nft CIC Static 年c0s2nft component CIC Combined -sin 2nft Use plain cos wave rather than impulses Measure the phase rather than Doppler shifts Decompose the received signal in vector space rather than in the time/frequency domain 日+回1三1交非三勺风 6/20

Motivation System Design Experiments Conclusions Basic Idea cos 2πft —sin 2πft CIC CIC I Q Acos2πft I Q Combined Static component Dynamic Component • Use plain cos wave rather than impulses • Measure the phase rather than Doppler shifts • Decompose the received signal in vector space rather than in the time/frequency domain 6/20

Motivation System Design Experiments Conclusions Basic Idea Combined Dynamic Component Acos 2nft +8 ↑c0s2nft Static component -sin 2nft Use plain cos wave rather than impulses Measure the phase rather than Doppler shifts Decompose the received signal in vector space rather than in the time/frequency domain 日+回1三1交非三勺风 6/20

Motivation System Design Experiments Conclusions Basic Idea cos 2πft —sin 2πft CIC CIC I Q Acos2πft I Q Combined Static component Dynamic Component • Use plain cos wave rather than impulses • Measure the phase rather than Doppler shifts • Decompose the received signal in vector space rather than in the time/frequency domain 6/20

Motivation System Design Experiments Conclusions Real World Signals 464 Time (seconds) Sound frequency 17~22 kHz->wavelength 1.5~2 cm Path length change of wavelength>phase change of 2m 。Move1.25mm→path length change2.5mm→phase changeπ/4 。Phase change direction→movement direction 日，+司1三13卡三分Q0 720

Motivation System Design Experiments Conclusions Real World Signals 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 −300 −200 −100 0 100 200 300 Time (seconds) I/Q (normalized) I Q −200 −100 0 100 200 −300 −200 −100 0 100 I (normalized) Q (normalized) Starting (4.04s) Static vector Ending (4.64s) Dynamic vector • Sound frequency 17 ∼ 22 kHz → wavelength 1.5 ∼ 2 cm • Path length change of wavelength → phase change of 2π • Move 1.25 mm → path length change 2.5 mm → phase change π/4 • Phase change direction → movement direction 7/20