A Survey Paper On Non Intrusive Texting While Driving Detection Using Smart Phone
1. DOI:10.21884/IJMTER.2018.5031.PM34L 13
A SURVEY PAPER ON NON INTRUSIVE TEXTING WHILE DRIVING
DETECTION USING SMART PHONE
S.Santhanalakshmi1
, N.Rohini2
, U.Nivethitha3
and Dr.G.Umarani Srikanth4
1, 2, 3
U.G. Student Computer science and engineering, S.A Engineering College, Chennai
4
Professor Computer science and engineering, S.A Engineering College
Abstractâ Many interesting systems on mobile phone are introduced to prevent the dangerous
activity of texting while driving (T&D). However these systems require userâs manual input to block
the texting activity or some extra devices are used to identify userâs location and also some
localization techniques are used. Here the idea is to propose a method which is able to detect T&D
without any user intervention. To implement this idea various smart phone embedded sensors such as
gyroscopes, accelerometers and GPS are used. They are used to collect the information such as touch
strokes, holding orientation and vehicle speed. Using these patterns the whether the T&D is existing
or not identified.
Keywordsâ Vehicles, Smart phones, Global Positioning System, Accelerometers.
I. INTRODUCTION
Smart phones are small computers which uses android as operating system. Data
communication between the people/user has become faster using smart phones. Calling and texting
otherwise called as messaging is the most important and basic activity that has been done using smart
phones. But texting while driving has become most common habit between the drivers. This is one of
the most hazardous things performed by the users of smart phone. This act becomes major distraction
from driving. This leads to road accidents and causes damage to the drivers and also other common
people. Usage of mobile phones while driving can affect your alertness of surroundings, which leads
to accidents. This term in short is referred to as T&D that is Texting while Driving. This is
considered as one of the top most dangerous activity done by the drivers. A research says that, the
drivers who send messages while driving are 23 times more likely to experience a crash on others
when compared to other drivers who are dialing, talking or listening. Even though, many laws are
imposed to avoid T&D, the users are involving in this activity. To stop these activities of drivers and
to identify the T&D, many mobile apps are now developed and practised between the users.
This system does not require any manual input from the driver. This acts as the major advantage of
this over other T&D system. In other detection system the major problem is they require user input
or activation or it either disables all the mobile phones in the user location (such as car) , this causes
unnecessary inconvenience to the users. The key challenge is detecting whether the mobile phone
belongs to driver or passenger. For this purpose, some systems uses camera to monitor driverâs
activities. But this method is infrastructure heavy and required high hardware facilities to implement
real time video processing. Some of the works are done for detecting T&D based localization. Here
some extra devices in combination with mobile phone work collaboratively to determine the user
location. To enlighten this idea most successfully a method which does not require any extra devices
except userâs mobile phone was needed. Here are some patters that is used to distinguish between the
driver and the passenger.
Pattern1: Editing messages after the car speed is decreased.
Pattern 2: Stop editing when the car is taking turns.
Pattern 3: Holding the mobile phone uprightly while editing the messages.
Therefore by using this system the T&D patterns are identified by the data collected the smart
phones and the other embedded sensors like accelerometers, gyroscopes and GPS. The other
2. International Journal of Modern Trends in Engineering and Research (IJMTER)
Volume 05, Issue 2, [Februaryâ 2018] ISSN (Online):2349â9745; ISSN (Print):2393-8161
@IJMTER-2018, All rights Reserved 14
important factor is that, this system does not try to read the content of the messages. Therefore
privacy is preserved here.
II. RELATED WORKS
The existing system to prevent T&D there are two categories, the first one is which simply
blocks all the messages from the user end. The best example for this is driver mode or bike mode
available in the mobile phones. But this existing system does not identify the driverâs mobile with
the passengers.
The next category is identifying whether the mobile is used drivers or passengers and it
blocks the driverâs mobile only. In this method identifying the location of the mobile phone in the
vehicle is important. For this localization techniques are used to identify the driverâs location.
However this approach has some drawbacks that identifying the driverâs location requires some
special sensors and required high deployment cost. Identification of driverâs location and driving
patters in also infrastructure heavy and intrusive.
Therefore for overcoming these drawbacks infrastructure-less approach is required. That is
identifying the driverâs usage of mobile without any extra devises except his/her own mobile phone.
III. TOUCH STROKE DETECTION
Here the touch strokes are identified by using the gyroscope data. Using the training data set
a touch stroke template is constructed. On this gyroscope data the template is utilized as wavelet
basis and they carry out wavelet transform. According to the occurrence of touch strokes the
significant peaks and the location of the peaks are selected.
IV. LITRATURE SURVEY
In [1], this uses embedded sensors on the phone .It is used in tracking different driving
conditions. It also makes use of vision based algorithms to detect critical driving behaviours
including taking unsafe turns. The results demonstrate that the safe cam is effective in detecting real-
road driving environments and alert drivers during dangerous situation. In [2], it provides a effective
driver assistance services using ever richer set of sensors. It provides assistance for the drivers in
services like traffic advisory and road condition monitoring .This paper introduce a sensing platform
called as windshield-mounted smart phones. It uses a prototype called as signal guru, it predicts the
schedule of traffic signals and also enabling Green Light Optimal Speed Advisory (GLOSA)
In this system [3], a novel method that uses Dynamic Dime Warping (DTW) and smart phone
based sensor-fusion is used. They detect and recognize the driver actions without external
processing. It performs the pattern recognition research by combining the inter-axial data from
multiple sensors into a single one that is single classifier. Drive safe [4], this application uses
computer vision and pattern recognition techniques to alert during their unsafe driving behaviour.
The inbuilt sensors present are able to detect in attentive driving behaviours and also evaluating the
quality of driving at the same time. It produces sound alarms in case of unsafe driving. It becomes
the first application for smart phones to detect the driving behaviours.
This application [5], introduces loss functions designed for applications. It mainly aims to
minimize the number of missed detections and false alarms, this estimates the risk level in each
turning and cornering event. This estimation only uses Global Navigation Satellite System (GNSS)
measurements. It also supports real-time value added services. In this paper [6], the system available
makes use of three axis accelerometer of android based smart phone. It records and analyses various
driver behaviours. It also alerts about the external road conditions that would be hazardous for the
drivers. By using the real-time analysis and auditory, we can increase the driverâs overall awareness.
In this system [7], the fundamentally addressed problem is differentiating driverâs and
passengerâs mobile phone. In this system, it leverages the stereo infrastructure, particularly speakers
and Bluetooth. Using car stereo, a customized high frequency beeps are produced. Sequential
change-point detection scheme is used to time the arrival and phoneâs distance from the carâs centre
3. International Journal of Modern Trends in Engineering and Research (IJMTER)
Volume 05, Issue 2, [Februaryâ 2018] ISSN (Online):2349â9745; ISSN (Print):2393-8161
@IJMTER-2018, All rights Reserved 15
is estimated. This paper [8], provides some well-established techniques for driverâs inattention
monitoring and recent solutions for exploiting mobile technologies such as smart phone and
wearable devices .It primary aim is safe driving and uses active systems for car-to-car
communication to support vehicle ADHOC network VANET.
This system [9], uses embedded sensors in smart phones for capturing the centripetal
acceleration. Their differences are obtained. This is a low infrastructure approach. It has different
turn size and difference in the driving speed. This method is adopted in many traffic related safety
applications. Here, in this paper [10], TEXIVE is implemented which detects the texting operations
during driving. It also notices the irregularities and rich micro-movements of the users. TEXIVE is
very accurate that the dangerous driving behaviours are identified without any extra devices rather
than smart phone.
This paper [11], gives the solution to the problem of distracted driving by giving a event-
driven solution called as Automatic Identification of Driverâs Smart phone (AIDS). It makes use of
features that are available for identifying the driverâs phone. It uses electromagnetic field spikes to
differentiate driverâs phone from the passengers in the car. In this paper [12], recognition accuracy of
vehicle steering patterns are improved. This paper presents a new method to reduce both energy and
computation complexity. Different statistical mode is identified and statistical sensor features are
reflected. Real-time accuracy is produced in steering modes. Different machine learning modes are
compared to produce improved classifier training.
This paper [13], is about to identify how powerful some of the low cost sensors life Inertial
Measurement Unit (IMU) and GPS are useful for intelligent vehicles. The relation between the smart
phone reference system and vehicle reference system are identified by accelerometer and gyroscope.
Based on longitudinal vertical acceleration on automatic method is proposed to calibrate a smart
phone using IMU and GPS. Here filter algorithm was used to decrease the impact of IMU noise. By
fusion sensor data on smart phone a system for car navigation was proposed [14]. To support GPS,
both the internal sensor and care sensors are used. To fuse the data from different sensorâs streams
strap down algorithm and Kalman filters are used. The speed information of the car is gives by car
sensors and gyroscope act as internal sensor.
This paper [15], builds a statistical model of the driver, vehicle and the smart phone and an
adaptive driving maneuver detection methods is proposed. This combination is done using
multivariate normal model. To detect risky driving behaviours a mechanism is knows as training
mechanism is adopted. It adapts profiling model to the driver and road topology. In this [16], paper a
ubiquitous camera feature is used to differentiate driverâs phone from the passengers. Here the key
factor is non-intrusive detection of the smart phone. No manual input from the driver phone is
needed. It does not rely on any other hardware devices. This method provides a very accurate
localization of the driverâs phone from the other users in the car.
In this paper [17], a system was proposed that the smart phone sensors are used to estimate
the vehicle speed. This system is mainly useful when GPS is unavailable or irregular in Urban areas.
The estimation of acceleration errors and large deviations are found here. And deviation between the
estimated and the real one are calculated. The changes in the acceleration errors are corrected when
needed. These points are called as reference points. In this paper [18], Controller Area Network
(CAN) was used in this system for detecting driver distractions. It mainly focuses on leg and head
movements of the drivers. It detects up to a high accuracy of distraction over 90%. Using this highly
reliable reduces the density of the accidents that are caused using driver distractions.
In this paper [19], a system was proposed to monitor the smart phone activities and it also blocks the
calls that are coming to the driver while driving. It also concentrates on phone position on the
vehicle. It blocks the calls from the phone that comes to the driverâs location. This method is
position-independent. It also reduces the noise that is produced from in and around the vehicleâs
location. Here [20], some components which can obtain and access the eco-driving and safe driving,
the vehicle information is obtained through CAN (Controller Area Network). It is inner vehicle
network protocol. It used Bluetooth network to transfer the information between the modules.
4. International Journal of Modern Trends in Engineering and Research (IJMTER)
Volume 05, Issue 2, [Februaryâ 2018] ISSN (Online):2349â9745; ISSN (Print):2393-8161
@IJMTER-2018, All rights Reserved 16
V. CONCLUSION
In this paper, to detect T&D we are submitting a novel method to make it simple. Here we are
using the authority method with some patterns that will guide us how smart phones are used in
moving vehicles. The associated information about some build in sensors in the smart phones was
collected and these sensors are analysed with hypothesis testing and checked for T&D patterns
match. The outcome of this approach will achieve good detection accuracy. This proposed T&D
method could be appropriate for usage based insurance. Many anti T&D mobile phone applications
are sustain with this method.
REFERENCES
[1] Landu Jiang, Xi Chen and Wenbo He, âSafeCam: Analyzing Intersection-Related Driver Behaviors using Multi-
Sensor Smartphones,â IEEE International Conference on Pervasive Computing and Communications, pp. 1-9,
March. 2016.
[2] Emmanouil Koukoumidis, Margaret Martonosi, and Li-Shiwan Peh, âLeveraging Smartphone Cameras for
Collaborative Road Advisories,â IEEE Trans. Mobile Comput., vol. 11, no. 5, pp. 707-723, May. 2012.
[3] D. A. Johnson and M. M. Trivedi, âDriving Style Recognition Using a Smartphone as a Sensor Platform,â IEEE
Transactions on Intelligent Transportation Systems, pp. 1609-1615, 2011.
[4] L. M. Bergasa, D. AlmerĂa, J. AlmazĂĄn, J. J. Yebes, and R. Arroyo, âDriveSafe: an App for Alerting Inattentive
Drivers and Scoring Driving Behaviors,â in IEEE Intelligent Vehicles Symposium (IV), Detroit, USA, June 2014,
pp. 240â245.
[5] J. Wahlström, I. Skog, and P. HĂ€ndel, âDetection of dangerous cornering in GNSS-data-driven insurance
telematics,â IEEE Trans. Intell. Transp. Syst., vol. 16, no. 6, pp. 3073â3083, Dec. 2015.
[6] M. Fazeen, B. Gozick, R. Dantu, M. Bhukhiya, and M. C. Gonzlez, âSafe Driving Using Mobile Phones,â Intelligent
Transportation Systems, IEEE Transactions on , vol. 13, no. 3, pp.1462-1468, 2012.
[7] J. Yang, S. Sidhom, G. Chandrasekaran, T. Vu, H. Liu, N. Cecan, Y. Chen, M. Gruteser, and R. Martin, âSensing
driver phone use with acoustic ranging through car speakers,â IEEE Trans. Mobile Comput., vol. 11, no. 9, pp.
1426â1440, Sep. 2012.
[8] H. Eren, S. Makinist, E. Akin, and A. Yilmaz, âEstimating driving behavior by a smartphone,â in Proceedings of
IEEE Intelligent Vehicles Symposium (IV), pp. 234-239, 2012.
[9] Y. Wang, et al., âDetermining driver phone use by exploiting smartphone integrated sensors,â IEEE Trans. Mobile
Comput., vol. 15, no. 8, pp. 1965â1981, Aug. 2016.
[10]Cheng Bo, Xuesi Jian, Taeho Jung, Junze Han, and Xiang-Yang Li, âDetecting Driverâs Smartphone Usage via
Nonintrusively Sensing Driving Dynamics,â IEEE Internet of Things Journal., vol. 4, no. 2, pp. 340-350, April.
2017.
[11]Homin Park, DaeHan Ahn, and Taejoon Park, âAutomatic Identification of Driverâs Smartphone Exploiting
Common Vehicle-Riding Actions,â IEEE Trans. Mobile Comput., vol. 17, no. 2, pp. 265-278, Feb. 2018.
[12]Zhenchao Ouyang, Jianwei Niu, and Mohsen Guizani, âImproved Vehicle Steering Pattern Recognition by Using
Selected Sensor Data,â IEEE Trans. Mobile Comput., vol. 14, no. 8, pp. 1-1, Aug. 2015.
[13]J. AlmazĂĄn, L. M. Bergasa, J. Yebes, R. Barea, and R. Arroyo, âFull auto-calibration of a smartphone on board a
vehicle using IMU and GPS embedded sensors,â in Proc. IEEE Intell. Vehicles Symp. (IV), Jun. 2013, pp. 1374â
1380.
[14]O. Walter, J. Schmalenstroeer, A. Engler, and R. Haeb-Umbach, âSmartphone-based sensor fusion for improved
vehicular navigation,â in Proc. 10th WPNC, Dresden, Germany, Mar. 2013, pp. 1â6.
[15]German Castignani, Thierry Derrmann, RaphaĂ«l Frank, and Thomas Engel âSmartphone-Based Adaptive Driving
Maneuver Detection: A Large-Scale Evaluation Study,â IEEE Trans. Intelligent Transportation System., vol. 18, no.
9, pp. 2330â2339, Sep. 2017.
[16]Y. Liang, âDetecting driver distraction,â Ph.D. dissertation, Univ. Iowa, Iowa City, IA, USA, 2009.
[17]H. Han, J. Yu, H. Zhu, Y. Chen, J. Yang, Y. Zhu, G. Xue and M. Li, âSenSpeed: Sensing Driving Conditions to
Estimate Vehicle Speed in Urban Environments,â in Proceedings of the 33rd Annual IEEE International Conference
on Computer Communications (INFOCOMâ14), Toronto, Canada, 2014, pp. 727-735.
[18]A. Sathyanarayana, S. Nageswaren, H. Ghasemzadeh, R. Jafari, and J. H. L. Hansen, âBody sensor networks for
driver distraction identifi- cation,â in Proc. Veh. Electron. Safety, 2008, pp. 120â125.
[19]Tianyi Song , Xiuzhen Cheng , Hongjuan Li , Jiguo Yu , Shengling Wang , Rongfang Bie, âDetecting Driver Phone
Calls in a Moving Vehicle Based on Voice Features,â IEEE International Conference on Computer
Communications, pp. 1-9, April. 2016.
[20]Doo Seop Yun, Jeong-Woo Lee, Shin-Kyung Lee, and Oh-Cheon Kwon, âDevelopment Of the Eco-Driving and
Safe-Driving Components using Vehicle Information,â 2012 International Conference on ICT Convergence (ICTC),
pp. 561-562, Oct. 2012.