Poster: Who is Using the Phone within the Car? Blind Device
Localization in a Car with Unimodal Acoustic Signature
Sugandh Pargal, Bivas Mitra, Sandip Chakraborty
Indian Institute of Technology, Kharagpur, INDIA 721302
sugandhpargal@kgpian.iitkgp.ac.in,{bivas,sandipc}@cse.iitkgp.ac.in
1 INTRODUCTION
The rapid usage of smartphones while driving is a signifcant con-
cern leading to distracted driving behavior, causing severe accidents,
as listed out by the US National Highway Trafc Safety Admin-
istration
1
. Assessing the smartphone’s location inside the car is
advantageous to such applications. Existing studies try to locate the
smartphone using external equipment, such as camera, Bluetooth,
network jammers, etc. [1ś3]. However, the drivers can tamper with
such external equipment, being the adversary for such a system.
This poster aims to fnd out the relative position of a smartphone
inside a car blindly without any specialized external hardware. We
leverage the acoustic properties inside a running vehicle to locate
whether the driver or a passenger uses the smartphone.
2 BLIND DEVICE LOCALIZATION
This poster explores the acoustic signals produced by ambient me-
chanical noises to help a smartphone detect its location blindly.
These acoustics are primarily due to the engine of the car, brak-
ing action, AC vents’ sound, the sound system, wind noise, etc.,
as shown in Fig.1. These sound waves are observed at a fxed fre-
quency range of typically 20-4000Hz. As we move at diferent seats
inside the car, these acoustic efects vary. We examine this impact
using spectral as well as time-domain analysis. We utilise the au-
dio samples collected inside the car with AC vents on, at diferent
positions of the car, say Driver Hand, Front Passenger, Dashboard,
Right-Back Passenger, Left-Back Passenger. These samples were col-
lected for geared as well as automated geared cars under diferent
surrounding conditions. We collected the samples for 3 cars of vary-
ing models for duration of 2160, 1440 and 2220 seconds, respectively.
We observed from the audio data that frequency domain analysis of
the acoustic noise within the car show a consistent periodic range
at diferent positions. Thus, we considered (a) Periodic Frequency
and (b) Fundamental Frequency as the acoustic signatures which
best describe the locations inside the car.
3 PRELIMINARY EVALUATION
We conducted preliminary tests to evaluate the performance of
our method of localizing a smartphone into three locations ś (a)
1
https://www.nhtsa.gov/risky-driving/distracted-driving (Access: February 7, 2022)
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for proft or commercial advantage and that copies bear this notice and the full citation
on the frst page. Copyrights for components of this work owned by others than ACM
must be honored. Abstracting with credit is permitted. To copy otherwise, or republish,
to post on servers or to redistribute to lists, requires prior specifc permission and/or a
fee. Request permissions from permissions@acm.org.
HotMobile ’22, March 9–10, 2022, Tempe, AZ, USA
© 2022 Association for Computing Machinery.
ACM ISBN 978-1-4503-9218-1/22/03. . . $15.00
https://doi.org/10.1145/3508396.3517073
Figure 1: Source of noise within a vehicle
(a) Periodic Frequency (b) Fundamental Frequency
Figure 2: Location-wise performance
the smartphone is used by the driver while holding it in one hand
(D), (b) the smartphone is used by the passenger on the front seat
(PF), and (c) the smartphone is used by one of the passengers in
the back seat (PB). We computed micro-F1 score for our multi
class problem over the acoustic data collected for 3 cars, "Car1",
"Car2", and "Car3". Fig. 2 summarizes the results for our method
utilising the two signatures stated above (a) Periodic Frequency (P)
and (b) Fundamental Frequency (F); we see that P works well in
identifying when the phone is with the driver.
Although the system needs thorough testing under diferent en-
vironments, we believe that these initial fndings would be efective
enough to solve the challenging problem of blind device localiza-
tion within a car. In addition, a good idea would be to use another
secondary modality that can boost this performance. For example,
if the driver holds the phone while driving, then the inertial sensors
may also exhibit periodicity in the hand movement; however, we
must keep in mind the adversary model considered in this problem.
REFERENCES
[1] Chun-Yu Chen and Kang G Shin. 2022. In-vehicle Phone Localization for Preven-
tion of Distracted Driving. IEEE Transactions on Mobile Computing (2022).
[2] J. Wahlström, I. Skog, P. Händel, and A. Nehorai. 2016. IMU-based smartphone-to-
vehicle positioning. IEEE Transactions on Intelligent Vehicles 1, 2 (2016), 139ś147.
[3] Jie Yang, Simon Sidhom, Gayathri Chandrasekaran, Tam Vu, Hongbo Liu, Nicolae
Cecan, Yingying Chen, Marco Gruteser, and Richard P Martin. 2011. Detecting
driver phone use leveraging car speakers. In 17th ACM MobiCom. 97ś108.
127