1. Presenter: Ailane Mohamed Toufik Authors : Jie Yang †, Simon Sidhom †, Gayathri Chandrasekaran ∗, Tam Vu ∗, Hongbo Liu †, Nicolae Cecan ∗, Yingying Chen †, Marco Gruteser ∗, Richard P. Martin ∗ † Dept. of ECE, Stevens Institute of Technology ∗ WINLAB, Rutgers University ACM MobiCom 2011

2.  Motivation  Existing solutions  System design  Evaluation  Conclusion  Personal critics and paper weaknesses

3. 2.75 people die Every day U.S. Department of Transportation – National Highway Traffic Safety Administration Only in 2009

4. hands-free devices Minds off driving. Cognitive load distract driver! Real-world accidents indicated that hands-free and handheld users are as likely to be involved in accidents

5.  Law  Several States ban handheld phone use  China / Algeria and many countries as well  Technology  Hard blocking: radio jammer, blocking phone calls, texting, chat …  Soft interaction  Routing incoming calls to voicemail,  Delaying incoming text notifications  Automatic reply to callers Automatic Reply: “I’m driving right now; will get back with you!”

6. HandoverSignal Strength Car’s speedometer GPS The Previous solutions make use of: Problem: Only detects if the phone is in a moving vehicle or not 38% of automobile trips include passengers

7. Detect we are in a moving Car Check if it’s the Driver phone Route incoming calls…etc We will use GPSUse existing solution Our paper concern

8.  Acoustic approach based on two assumptions:  Use the seat location to determine the drivers phone  Phone is allowed to access the stereo system Bluetooth Symmetric positioning of speakers Phone connecting with head unit

9. Audio head unit

10. Propagation Delay ∆T ij = 0 : Phone is Equidistant from i and j ∆T ij > 0 : Phone is Closer to speaker i ∆T ij < 0 : Phone is Closer to speaker j ∆t ij ∆t’ ij Speaker iSpeaker j ∆T ij = ∆t’ ij - ∆t ij

11.  Unobtrusiveness  Robustness to noise and multipath  Computation Feasibility on Smartphone

12.  the frequency range of human hearing is generally considered to be 20 Hz to 20kHz, high frequency sounds must be much louder to be noticeable

13.  the current cell phone microphones are more sensitive to this high-frequency range. We experimented with an iPhone 3G and an Android Developer Phone 2 (ADP2).

14.  16-18kHz  ADP2 phone  18-20kHz  iPhone 3G

15.  Noise from: Engine, Tire/road and Wind < 1KHz  Conversations range from 300Hz to 3400Hz  Music can range from 50Hz, 15.000Hz which covers all naturally occurring sound  => To overcome robustness challenge we choose a signal above 15Khz

16.  Length:  Too short: a beep is not picked up by the microphone  Too long a beep: will add delay to the system and will be more susceptible to multi-path distortions.  We found empirically that a beep length of 400 samples (i.e., 10 ms) represents a good tradeoff.

18. ∆d24 is the distance difference from two right speakers. ∆d ij > TH lr ∆d ij = c * ∆ T ij (∆d ij + ∆d ij ) /2 > TH fb ∆T ij = S ij * f S ij is the number of samples that the beeps were apart f is the sampling frequency (typically 44.1kHz). c is the speed of the sound TH lr is a threshold that could be chosen as zero / -5cm ∆d13 represent the distance difference from two left side speakers

19.  Phones  Cars ADP2 Bluetooth radio 16-bit 44.1kHz sampling rate 192 RAM 528MHz MSM7200 processor Iphone 3G Bluetooth radio 16-bit 44.1kHz sampling rate 256 RAM 600 MHz Cortex A8processor Honda Civic Si Coupe Bluetooth radio Two channel audio system two front and two rear speakers Interior dimension  Car I: 175 x 183 cm  Car II: 185x 203cm Acura sedan

20.  Testing positions  Different number of occupants  Different noise conditions  Highway Driving  60MPH + music playing + w/o window opened  Phones at front seats only  Stationary  Varying background noise: idling engine + conversation

22. 22 4 channel, all seats2 channel, front seats

23. Cup-holder v.s. co-driver left

24.  Enabled a first generation system of detecting driver phone use through a smartphone app  Practical today in all cars with built-in Bluetooth  Leveraging car speakers – without additional hardware  Validated the generality of our approach with two kinds of phones and in two different cars  Classification accuracy of over 90%, and around 95% with some calibrations  Limitations  Phone is muffled by bag or winter coat  Driver places the phone on an empty passenger seat

25.  Why not fade music to lower the complexity.  Why not testing all positions in all scenarios.  Really, I should use this app ?  Develop a dedicated system with self-calibration ability.  ?

26. DRIVE SAFELY TALK & TEXT LATER