research article sensafe: a smartphone-based traffic...

Research ArticleSenSafe A Smartphone-Based Traffic Safety Framework bySensing Vehicle and Pedestrian Behaviors

Zhenyu Liu Mengfei Wu Konglin Zhu and Lin Zhang

Key Laboratory of Universal Wireless Communications Ministry of Education Beijing University of Posts and TelecommunicationsBeijing China

Correspondence should be addressed to Lin Zhang zhanglinbupteducn

Received 10 June 2016 Revised 7 September 2016 Accepted 28 September 2016

Academic Editor Francisco Martınez

Copyright copy 2016 Zhenyu Liu et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Traffic accident involving vehicles is one of the most serious problems in the transportation system nowadays How to detectdangerous steering and then alarm drivers in real time is a problem What is more walking while using smartphones makespedestrian more susceptible to various risks Although dedicated short range communication (DSRC) provides the way for safetycommunications most of vehicles have not been deployed with DSRC components Even worse DSRC is not supported by thesmartphones for vehicle-to-pedestrian (V2P) communication In this paper a smartphone-based framework named SenSafe isdeveloped to improve the traffic safety SenSafe is a framework which only utilizes the smartphone to sense the surroundingevents and provides alerts to drivers Smartphone-based driving behaviors detectionmechanism is developed inside the frameworkto discover various steering behaviors Besides the Wi-Fi association and authentication overhead is reduced to broadcast thecompressed sensing data using the Wi-Fi beacon to inform the drivers of the surroundings Furthermore a collision estimationalgorithm is designed to issue appropriate warnings Finally an Android-based implementation of SenSafe framework has beenachieved to demonstrate the application reliability in real environments

1 Introduction

Traffic safety becomes one of the serious problems in thetransportation systems As the number of the vehicles isgrowing the levels of traffic accidents have increased signif-icantly Not only vehicles but also pedestrians and cyclistsare facing the safety threats from traffic accidents Accordingto [1] traffic accidents resulted in more than 500 thousanddeaths and 14 million injures worldwide by the end of Mayin 2016 One of the main reasons for traffic accidents is thatthe drivers cannot notice behaviors of surrounding vehicleson time Using smartphone while walking is increasinglyapparent in our society and when people walk with theirattention on smartphones and distracted from the scenerypotential accidents are in front of them To reduce the trafficaccidents it is necessary to assist the drivers to have a betterunderstanding of the surrounding environments includingthe coming vehicles and adjacent pedestrians and cyclists

Dedicated short range communication (DSRC) [2] hasbeen accepted as the most promising approach to enhance

the transportation safety It consists of a set of vehiclesequipped with the on-board unit (OBU) and a set of roadside units (RSUs) along the roads and aims to providereliable and low latency vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications However due toadditional effort for DSRC deployment most of vehicles havenot been equipped with OBU for DSRC Besides DSRC isnot supported by the general smartphones for vehicle-to-pedestrian (V2P) communication

To enable the safety of transportations many strategiesare proposed There are some works using the vehicle-mounted device for safety assistance driving system Camerasand radars are used tomonitor the driving behaviors and alertthe dangerous distances between vehicles to provide technicalsupport for the auto manufacturers Although the cost ofcameras and radars based safety technology is decreasingthese safety technologies have not been deployed in economyvehicles It still needs time before the majority of vehicles aredeployed with these safety technologies Furthermore due to

Hindawi Publishing CorporationMobile Information SystemsVolume 2016 Article ID 7967249 13 pageshttpdxdoiorg10115520167967249

2 Mobile Information Systems

lack of communication vehicles need to make decisions bythemselves

With the widespread use of the smartphone most ofdrivers and pedestrians have smartphones which providesthe potential to use them for the enhancement of trafficsafety Nowadays smartphones are progressively equippedwith functional sensors (eg accelerometer gyroscope GPScamera etc) and these sensors can be applied to recognizeand monitor various vehicle behaviors Additionally theycontain communication resources and enable the quickdeployment of new applications [3]

Sensor technology available in smartphones enables themonitoring of vehicle mobility behaviors including lanechange turns acceleration and brake If such informationcan be sensed and transmitted to other vehicles or pedes-trians they can be potentially used in collision preventionearly crash detection congestion avoidance and so forthSmartphones have abundant communication resources suchas Bluetooth Wi-Fi and 3GLTE However using themto construct appropriate networks for efficient informationsharing in transportation systems is difficult In particularthe transmission range of the Bluetooth cannot satisfy therequirements of the communication Wi-Fi has the proces-sion of authentication and association before data transmis-sion and is more suitable for 1-to-N communication 3GLTEneeds the assistance of the base station

In this paper we propose SenSafe a smartphone-basedframework for traffic safety by sensing communicationand alerting which overcomes the limitations of requestingadditional professional equipment inherent in the existingapproaches

SenSafe is a novel framework which only utilizes thesmartphone to improve the transportation safety It detectsand reports vehiclersquos events based on the sensing data col-lected from steering behaviors in urban area The changes inthe angle of vehicle heading and the corresponding displace-ment during a steering maneuver are calculated for drivingbehavior classification Driving behaviors are classified intodifferent types including turn lane change accelerationand brake Beacon-based communication is provided toexchange the driving information Surrounding environmentreminding and collision forewarning are provided based onthe data from the surroundings

We highlight our main contributions as follows

(i) We propose a framework SenSafe which only usesthe smartphone to sense the driving behavior ex-change the driving information and afford thereminding and alerting

(ii) We provide the detection and differentiation of var-ious driving behaviors by only utilizing the smart-phonersquos built-in sensors in urban areaWe analyze thesensing data collected from urban area and find thatvehicles cannot always make turns and lane changessmoothly owing to the influence of surrounding vehi-cles and pedestriansThe temporary interruptions aretaken into consideration to improve the accuracy ofthe driving behaviors detection

(iii) We furnish the beacon-based communication whichmodifies the service set identifier (SSID) field ofWi-Fibeacon frame to carry the driving behavior positionand mobility information A data compression anddecompressionmethod is provided tomake use of theSSID field Event-driven communication is utilizedto provide the surrounding drivers with immediatereminding Ordinary promptings of the surroundingvehiclesrsquo behaviors and safety alerts of the comingcollisions are provided to the drivers

(iv) AnAndroid-based implementation of SenSafe frame-work has been achieved to demonstrate the evaluationresults and application reliability in real environ-ments

The rest of this paper is organized as follows In Section 2a brief overview of the related works is provided In Section 3the framework overview of SenSafe is introducedThe detailsfor three modules of SenSafe are explained in Sections 4ndash6 The implementation performance evaluation and exper-iment results are shown in Section 7 Finally this paper isconcluded in Section 8

2 Related Works

There are some works to detect the vehicle behavior Someworks use the fixed vehicle-mounted devices to monitorthe vehicle behavior Mobileye uses cameras and radars toprovide the technical support for the auto manufacturers [4]Although the cost of vehicle safety technology is decreasingthese safety technologies have not been deployed in economyvehicles It still needs time before the majority of vehiclesare deployed with these safety technologies With the widespread of smartphones smartphone solutions can be usedin vehicles and there are some works focusing on usingsmartphones to assist drivers

Drivea [5] iOnRoad [6] and Augmented Driving [7]are apps which have capability of detecting lane departuresand warn drivers when the distance to the front vehicle istoo close CarSafe [8] alerts distracted drivers using dualcameras on smartphones one for detecting driver state andthe other for tracking road conditions Although camera-based systems have the functionality in assisting the driverthey have limitations in terms of computational overhead andrequirement of image quality and work worse at night andwith bad lighting conditions

Camera-free sensors in the smartphone have been uti-lized in traffic regulator detection [9] localization [10]transportation mode classification [11] vehicle speed esti-mation [12] and the driving behavior detection [10 13ndash15]In [13] smartphone sensing of vehicle dynamics is utilizedto determine driver phone use Inertial measurement unit(IMU) sensors including accelerometers and gyroscopes ofthe smartphone are used to capture differences in centripetalacceleration due to vehicle dynamics In [14] three algo-rithms which detect driving events using motion sensorsembedded on a smartphone are proposedThe first detectionalgorithm is based on data collected from GPS receiver Thesecond and third detection algorithms utilize accelerometer

Mobile Information Systems 3

GPSMagnetometerGyroscopeAccelerometer

BehaviorLatitudeLongitudeSpeedAngleTime

GPS

Lane changesBrakes andaccelerationsTurns

LatitudeLongitudeSpeedAngleTime

BeaconBeacon

Detections

Figure 1 Application scenarios

sensor and use pattern matching technique to detect drivingevents In [15] a vehicle steering detection middleware calledV-Sense is developed on commodity smartphones by onlyutilizing nonvision sensors on the smartphone Algorithmsare designed for detecting and distinguishing various vehiclemaneuvers including lane changes turns and driving oncurvy roads However the paper only considers the smoothdriving behaviorsOnce there is a sudden brake because of thesurrounding vehicles and pedestrians it cannot work well In[10] embedded sensors in smartphones are utilized to capturethe patterns of lane change behaviors and this paper on thedriving environment of highway

There are some works for the communication Somecompanies and researchers focus on using the DSRC forV2V communication [16 17] However all of these worksneed special devices for communications which have notbeen deployed in most of vehicles Some works use thecommunication resources of smartphone to avoid using extradevices The master access point disseminates informationto the rest of the units through Wi-Fi of smartphones [18]and it is suitable for the solid groups But for the dynamicsgroups Wi-Fi has the processions of authentication andassociation and is usually for the 1-to-N communicationsFor the Wi-Fi Direct the device discovery phase may takeover ten seconds in some cases [19] The authors modify theSSID of Wi-Fi beacons to store the location and speed ofthe smartphone for alert [20] However they do not considerthe driving behaviors of the vehicles Besides as the accesspoints (APs) can only broadcast beacons and the clients canonly receive the beacons how to achieve the bidirectionalcommunications is not considered

Different from the previous works we propose an inte-grated smartphone-based framework SenSafe for trafficsafety considering sensing vehicle behaviorsThis frameworkis only based on smartphone and thus can be easily deployeddue to the widespread use of the smartphone Vehiclebehavior sensing is one of the key points in this frameworkTo improve the robustness and the accuracy of the vehiclebehavior sensing nonvision sensors are utilized to reduce

the influence of the image quality and unsmooth drivingbehaviors are considered as there may be some brakes inthe procession of turns Considering that only AP broad-casts beacons and cannot receive beacons from the otherAPs an event-driven communication method is proposedto achieve the bidirectional communications Finally thereminder events are divided into ordinary prompting of thesurrounding vehiclesrsquo behaviors and safety alert of the comingcollisions for the drivers

3 Framework Overview

This section describes the high level overview of the frame-work SenSafe that we propose for traffic safety SenSafe isfocused on the vehicles and pedestrian safety One of themain reasons for the traffic accident is that the drivers cannotnotice the behaviors of surrounding vehicles and pedestrianson time SenSafe considers using the smartphone which iswidespread to improve the traffic safety without any cost ofinstalling new device

As is shown in Figure 1 SenSafe uses the sensors fromthe smartphone to collect the data about the vehicle mobilityAfter that these data can be used to obtain the driving behav-ior of the vehicles Using the information transferred fromthe surroundings drivers can have a better understanding ofthe environments Besides the smartphone of the pedestrianobtains themoving information from theGPS and broadcastsit to the vehicles to enhance the vulnerable pedestrian safety

System architecture is shown in Figure 2 The frameworkis made of three components driving behaviors detectionmodule beacon-based communication module and colli-sion forewarning module The driving behaviors detectionmodule considers the output of motion sensors and GPSAcceleration braking turns and lane changes are detectedusing the proposed algorithm Beacon-based communicationmodule compresses the sensed data into the SSID of thebeacon for communication Collision forewarning moduleuses the data from the surrounding to remind the driver ofthe driving behaviors and finds out the vehicles which have


Eventdetection

Data compression beacon broadcast

Driving behaviors

AccelerometerMagnetometer

GyroscopeGPS

Collision forewarning

Mobilityinformation

Figure 2 System Architecture

threat to the current vehicle and pedestrians who might bethreatened

4 Driving Behaviors Detection

During a single trip the smartphone collects input datafrom motion sensors and GPS and the driving behaviordetection system decides the type of the driving behaviorsThis section details the design and functionalities of drivingbehavior detection First we describe how the sensors ofsmartphones are utilized to determine whether the vehicleis making acceleration brake turn or changing lane Thenwe show how SenSafe classifies such different vehicle drivingbehaviors based on the detection results The Android-based smartphone is used as our target platform to collectraw data from four on-board sensors 3-axis accelerometermagnetometer gyroscope and GPS receiver

41 Coordinate Alignment Given the direction of gravity onthe smartphone body coordinate system the smartphoneattitude can be fixed within a conical surface in the earthcoordinate system As a result we align the smartphonecoordinate with the earth coordinate for removing 3 degreesof freedom to 1 degree by projection as shown in Figure 3Magnetometer is utilized to get the angle 120575 between119884119890 and119884

1015840

axis in the earth coordinate system where1198841015840 is the projectionof 119884119904 of the smartphone body coordinate system on the119883119890-119884119890 plane of the earth coordinate system 119883119890 (pointingto the earth east) 119884119890 (pointing to the earth north) and 119885119890(parallel with the gravity) are the three reference axes in theearth coordinate system Combining the result with the anglederived from the magnetometer reading and the rotationmatrix the component of sensing data corresponding tothe vehicle moving direction can be calculated The detailedformulation of the rotation matrix is explained in [21]

42 Detection Algorithm

421 Acceleration and Brake Detection With the help ofaccelerometer the acceleration and the brake can be distin-guished (Figures 4 and 5) Moving average filter is used toremove noise from the raw acceleration Due to the unpre-dictable road conditions and diverse driving preferences theacceleration in real implementation includes noise In orderto filter out the noise we use a state machine to detect the

Cone

Ze

Xe

Ye

Ys

120575

Y998400

Figure 3 Align the smartphone coordinate system with the earthcoordinate system

Acce

lera

ted

spee

d (m

2s

)

TPending

ca ca

0

05

1

15

2

25

3

35

4

2010 15 250 5Samples

Figure 4 Accelerometer readings when the vehicle accelerates

event There are basically five states Waiting-for-Acc Acc-Pending Brake-Pending Acceleration and Brake The statetransition procession is shown in Figure 6 The followinglist describes the specification of the parameters used inacceleration and brake detection algorithm

Acc acceleration from the moving average filter

Acc Threshold the threshold of the acceleration toenter the Acc-Pending state

Mobile Information Systems 5Ac

cele

rate

d sp

eed

(m2s

)

cb cb

minus25

minus2

minus15

minus1

minus05

0

5 10 15 20 250Samples

TPending

Figure 5 Accelerometer readings when the vehicle brakes

Waiting-for-Acc

Acc-Pending

Acceleration Brake

Brake-Pending

No Yes

Yes

No

Yes

No

Yes

No

Brake_Time_ThresholdT_Brake_Pending gt

Acc lt Brake_Threshold

Acc_Time_ThresholdT_Acc_Pending gt

Acc gt Acc_Threshold

Figure 6 State diagram of the acceleration and brake detection

Brake Threshold the threshold of the acceleration toenter the Brake-Pending stateAcc Time Threshold the threshold of time in Acc-Pending state to enter the Acceleration stateBrake Time Threshold the threshold of time inBrake-Pending state to enter the Brake stateT Acc Pending the time in Acc-Pending state up tonowT Brake Pending the time in Brake-Pending state upto now

Waiting-for-Acc is the initial state for the accelerationand brake detection As the real acceleration appears as theundulatory curve due to noises Acc-Pending and Brake-Pending are two transitional states to reduce the influence of

the noise For the acceleration system enters into the Acc-Pending state after the Acc is greater than Acc Threshold andexits the Acc-Pending state when the Acc is less than or equalto Acc Threshold If T Acc Pending 119879Pending in Acc-Pendingstate is greater than Acc Time Threshold the state becomesAcceleration which means an acceleration is ongoing Forthe brake system enters into the Brake-Pending state after theAcc is less than Brake Threshold and exits the Brake-Pendingstate when the Acc is greater than or equal to Acc ThresholdIf T Brake Pending in Brake-Pending state is greater thanBrake Time Threshold the state becomes Brake

422 Turn and Lane Change Detection The 119885-axis readingsof gyroscope on the smartphone are utilized to represent thevehicle angular speed When the drivers do some behaviorsto change the direction of vehicles (eg changing singleor multiple lanes and making turns) the angular speedshave the obvious improving (Figure 7) For the left turna counterclockwise rotation around the 119885-axis occurs andgenerates positive readings whereas during a right turna clockwise rotation occurs and thus generates negativereadings For a left lane change positive readings are followedby negative readings whereas for a right lane change negativereadings are followed by positive readings

By detecting bumps in the 119885-axis gyroscope readingswe can determine whether the vehicle has made a leftrightturn or has made single-lane change The other steeringmaneuvers that is turn back and multiple-lane change havea similar shape but with a different size in terms of width andheight of the bumps The parameter description of the turnand lane change detection is shown in the following list

Ang SE Threshold the threshold of angular speed toenter and leave a possible bumpAng Top Threshold the threshold of angular speedwhich the maximum value of the valid bump shouldbe greater thanT Bump duration that angular speed is greater thanAng SE Threshold in a bumpT Bump Threshold the minimum threshold of theT Bump for a valid bumpG threshold the threshold of GPS speed to estimate ifthe vehicle is motionlessDelay threshold the threshold to judge the consecu-tive bumpT stop the duration whose GPS speed is less than theG thresholdT wait duration of Waiting-for-Bump state

Moving average filter is adopted to remove noise from theraw gyroscope readingsThedelay parameter of the filter is setto 15 samples which correspond to 075 seconds in the timedomain Experimental observation shows that it is short butgood enough to extract the waveform of the bumps

To reduce false positives and differentiate the bumps fromjitter a bump should satisfy the following three constraintsfor its validity (1) all the readings during a bump should


Turn right

minus07

minus06

minus05

minus04

minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples

Raw dataFiltered data

(a) Turn right (one bump)

Turn left

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples


(b) Turn left (one bump)


Two bumps turn right

minus05

minus04

minus03

minus02

minus01

0

01

02

Ang

ular

spee

d (r

ads

)

50 100 150 200 250 300 3500Samples

(c) Turn right (two bumps)

Two bumps turn left


50 100 150 200 250 3000Sample

minus1

minus05

0

05

1

15A

ngul

ar sp

eed

(rad

s)

(d) Turn left (two bumps)

Change to the right lane


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

(e) Change to the right lane

Change to the left lane


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

(f) Change to the left lane

Figure 7 Continued


Turn back20 40 60 80 100 120 140 160 1800

Samples


minus01

0

01

02

03

04

05

06

07

08

Ang

ular

spee

d (r

ads

)

(g) Turn back

Figure 7 Gyroscope readings when the vehicle makes turns or lane changes

as asat

TBump

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples


(a) One bump turn left

Change to a left lane

asas as

as

TBump

TDelay


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

at

at

(b) Two bumps lane change

Figure 8 Symbol description in gyroscope reading

be larger than Ang SE Threshold 119886119904 (2) the largest value ofa bump should be no less than Ang Top Threshold 119886119905 and(3) the duration of a bump T Bump 119879Bump should be no lessthan T Bump Threshold 119905119887 (Figure 8(a)) For the two validcontinuous bumps the time between the two bumps shouldbe less than Delay threshold 119879Delay except the time when thevehicle stops (Figure 8(b))

There are seven states in the bump detection algo-rithm No-Bump One-Bump Waiting-for-Bump More-Bump Bump-End Turn and Lane-Change The state transi-tion is influenced by angular speed and GPS speed Angularspeed is the 119885-axis gyroscope reading The state transitionprocession is shown in Figure 9

In No-Bump state angular speed is continuously moni-toredWhen the absolute value of themeasured angular speed

reachesAng SE Threshold it is treated as the start of a possiblebump and the algorithm enters One-Bump state

The One-Bump state terminates when the absolute valueof the measured angular speed is below Ang SE ThresholdIf the time of duration in One-Bump state is larger thanT Bump and the largest angular speed is larger thanAng Top Threshold hence satisfying the three constraints thefirst detected bump is assigned to be valid In such a casethe algorithm enters Waiting-for-Bump state Otherwise itreturns to No-Bump

InWaiting-for-Bump state it monitors the angular speeduntil its absolute value reaches Ang SE Threshold and theduration is defined as T wait The GPS speed is also mon-itored to see if the turn is interrupted halfway T stop isused to define the duration whose GPS speed is less than


No-Bump

One-Bump

Waiting-for-Bump

More-Bump

Lane-Change

Turn

NoYes

Is this a valid bumpNo

Yes

Is this a valid bump

Yes

Yes

Bump-End

Bumps have the same signs

No

Yes

No

No

T_wait-T_stop lt Delay_thresholdAngular speed gt Ang_SE_Threshold amp

Angular speed gt Ang_SE_Threshold

Figure 9 State diagram of the turn and lane change detection

the G threshold which means the turn is interrupted Ifthe difference between the T wait and T stop is less thanDelay threshold the system enters the More-Bump statewhich means the oncoming bump is considered as consec-utive bumps If the new bump is valid the system entersWaiting-for-Bump state for new bumps If not the systementers Bump-End state to determine the driving behavior

In Bump-End state the signs of these consecutive bumpscan be used for driving behavior judgment If these bumpshave the same signs the driving behavior is determined to beturn If they have the opposite signs the driving behavior isdetermined to be lane change Besides if there is only onevalid bump (ie the second bump is invalid) the drivingbehavior is determined to be turn

Based on bump detection driving behaviors are classifiedinto turns and lane changes To get the detailed classificationof the behaviors (eg left turn right turn and U-turn) thedifference in the heading angle between the start and the endof a driving behavior is used

We calculate the change in vehiclersquos heading angle fromthe readings of the gyroscope Δ119905 is the time interval ofsampling 119894 is the bump index of the valid bumps 119860119899119894is the angular speed of the 119899119894th sampling and is used toapproximately represent the average angular speed during thesampling periodThus we get the change in vehiclersquos headingangle from the integration of the heading angle change of eachsampling period

120579 =119868

sum119894=1

119873119894

sum119899119894=1

119860119899119894Δ119905 (1)

Table 1 The bounds of the angle change of heading for drivingbehavior

Lower bound Upper boundTurn left 70∘ 110∘

Turn right minus70∘ minus110∘

Turn back minus200∘160∘ minus160∘200∘

Left lane change minus20∘ 20∘

Right lane change minus20∘ 20∘

For the turning left 120579 is around 90∘ For turning right it isaround +90∘ For turning back 120579 is around plusmn180∘ For thelane change it is around 0∘

As the drivers cannot make a perfect driving behavior toget the perfect coincident degree the ranges of the drivingbehaviors are extended as shown in Table 1

Based on the horizontal displacement we furthermoreextract the multiple-lane change from the lane change as thehorizontal displacement of multiple-lane change is greaterthan the single one If the horizontal displacement is less thanHD Lower Threshold (eg the width of the lane) it meansthat these consecutive bumps are caused by some interferenceinstead of lane change If the horizontal displacement isgreater than HD Upper Threshold the behavior is treated asthe multiple-lane change Otherwise the behavior is treatedas single-lane change The horizontal displacement119867 can becalculated as follows Thereinto Δ119905 is the time interval of


sampling 119860 119894 is the angular speed of the 119894th sampling and 119866119899is the GPS speed of the 119899th sampling

120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)

5 Beacon-Based Communication

We use theWi-Fi of smartphones for communication As thenormal Wi-Fi has the association and authentication proces-sion which can cause the latency we choose the beacon frameof theWi-Fi AP to carry the sensing informationThe beaconframe is used to declare the existingAP and can be transferredby the APwithout association and authentication processionThe Beacon Stuffing embeds the intended messages withinthe SSID field of the Wi-Fi beacon header and is available inthe Wi-Fi AP mode [22]

The problem of using beacon to transfer sensing informa-tion is that the beacon can only be transferred from the AP tothe client which causes the one-way transmission Howeverit is not enough that only a part of the devices transfers thesensing information as every device needs to broadcast itsmobility information to the others To solve this problemwe propose the event-driven communication algorithm oncethe vehicle detects an event (eg acceleration brake turnand lane change) the smartphone inside will switch to theAP mode to broadcast the beacons for a while after that thesmartphone will switch to the mode to scan the beacons

To provide the reference for collision forewarning mod-ule these elements are selected the latitude and longitudefrom the GPS reader the speed from the GPS reader (ms)the travel direction from the magnetometer sensor (degree0sim360) the time from the GPS reader and the drivingbehaviors These elements are combined together to replaceSSID field of beacon A special string ldquoSenrdquo is used in frontof these element for differentiating SenSafersquos SSID from theothers

To satisfy the requirements of accuracy the 01 metersapproximately correspond to the 0000001 degrees in thelatitude and longitude which means the latitude and thelongitude need 19 characters in the SSID field (eg 116364815and 39967001 in BUPT) Besides the time from the GPSreader also needs too many characters If we want to makethe precision of the time to bemillisecond (eg 202020 234)it will take 9 characters even without considering separatorsHowever the length of beacon messagesrsquo SSID field is 32characters which is not enough for all elements As a resultwe compress these elements in the AP side and decompressthese elements in the client side

For the latitude one degree is about 111 kilometers Asthe maximum transmission range of the Wi-Fi beacon is lessthan 1 kilometer most of the significant digits of latitudeand longitude are same for the sender and receiver Thereis no need to transfer all the significant digits of latitudeand longitude Thus the significant digits of latitude and

116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872

202020 234 116364815 39967001Decompressed data

Time Latitude Longitude

Figure 10 The example of compression and decompression

202059 121

59121

202100 136

Remote data

Local data

Compressed data

Decompressed data 202059 121

Time

202159 121202059 121

Difference is too large Boundary situation

Figure 11 The example of time boundary situation

longitude are chosen to cover the range around the 1000meters (ie last 5 significant digits) It is similar for the timeas hour and minute are always same for the senders andreceivers Therefore we use 4 significant digits to representthe time from 001 seconds to 10 seconds An example isshown in Figure 10 Further due to the boundary situation ofthe time when the local device decompresses the time fromthe remote devices it will compare the difference betweenthe decompressed time and the local time If the absolutevalue of difference is greater than the threshold (30 seconds)the minute of the decompressed data will be modified to bethe adjacent number (+1minus1) to get the minimum reasonabledifference An example is shown in Figure 11The remote timeis 202059 121 and the local time is 202100 136 which willmake the normal decompressed time 202159 121 As it isunreasonable the adjacent minute (2020) is chosen to be theprefix of the remote time to get the minimum the absolutevalue of time difference For the latitude and longitude thepurpose of the decompression is obtaining the minimumdifference between the remote location and the local locationSimilar to the time if the absolute value of difference is greaterthan the threshold (200meters) the first decimal places of thelatitude and longitude aremodified to be the adjacent number(+1minus1) to get the minimum reasonable difference

The format of the SSID field is demonstrated in Figure 12We use the number to represent the special driving behaviorsin the driving behavior segment (0 pedestrian 1 accelera-tion 2 brake 3 turn left 4 turn right 5 turn back 6 leftlane change and 7 right lane change) For the pedestrianstheir smartphones sense their moving information from theGPS readers and broadcast the same elements to remind thedrivers of their existenceThe driving behaviors element is setto ldquo0rdquo which means this message is from the pedestrian


ldquoSenrdquo Driving behavior Latitude Longitude Speed Direction Time Reservation0 3 5 19 2210 15 3226

Figure 12 The format of the SSID field

Front

Left Right

Behind

Moving direction

45∘

45∘

Figure 13 The area division for safety reminding

6 Collision Forewarning

The vehicle will receive two kinds of messages one kind fromthe pedestrians and the other from the vehicles

After the vehicle gets the driving information of thesurrounding vehicles it will estimate if these vehicles willcrash into it to ensure the safety of the vehicle and thedriver Besides it will also remind the drivers of the drivingbehaviors from the surrounding vehicles to make the drivershave a better understanding of the driving environment

As is shown in Figure 13 considering the vehicle movingdirection the surrounding area of the vehicle is divided intofour quarters front behind left and right After receiving anew message the vehicle will calculate which area it belongsto Furthermore the driver can get the information where thethreat comes from

When the vehicle receives the messages from the sur-rounding vehicles it translates latitudes and longitudes toGauss-Krueger plane rectangular coordinates system Thenit calculates the driving vector of each vehicle and findsvehicles that have intersection with it and gets the locationwhere they may have the intersection as shown in Figure 14If the vectors have the intersection in the near future itwill calculate the distance of the current vehicle and thethreat vehicle to intersection and calculate the time to theintersection (safety reaction time) furthermore If one ofthese two times is less than the safety reaction time thresholdand the absolute difference of them (safety intersectiontime) is less than safety intersection time threshold these twovehicles will be estimated to have the threat of crashing intoeach other When the vehicle receives the messages from thesurrounding pedestrian it will calculate the distance of thecurrent vehicle and pedestrian to intersection and the time

Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)

Figure 14 Intersection collision estimation

of vehicle to the intersection If the distance of pedestrian isless than safety distance threshold and the time of vehicle tothe intersection is less than safety reaction time threshold thevehicle will be estimated to have the threat to the pedestrian

We reference the safety reaction time threshold (3 sec-onds) recommended in [23] and safety intersection timethreshold (2 seconds) recommended in [24] to avoid acollision when GPS is accurate Assuming the inaccuracies ofGPS location are up to 10meters and the speeds of the vehiclesare around 10ms as a result the error of the safety reactiontime is up to 1 second and the error of the safety distance isup to 10 meters Thus we set safety reaction time threshold(4 seconds) to 1 second higher than the value which assumesthe GPS is accurate and safety intersection time threshold (4seconds) to 2 seconds higher than the value which assumesthe GPS is accurate

7 Performance Evaluation

To evaluate the performance of SenSafe we implementedSenSafe on a Samsung Galaxy Note II and a Huawei Chang-wan 4X

71 Parameter Setting Theparameters used in SenSafe are setas the values in the following list

Acc Threshold 08ms2

Brake Threshold minus1ms2

Acc Time Threshold 06 secondsBrake Time Threshold 06 secondsAng SE Threshold 003 radsAng Top Threshold 005 radsT Bump Threshold 15 seconds


Figure 15 Real road testing routes

Delay threshold 2 secondsG threshold 03mssafety reaction time threshold 4 secondssafety intersection time threshold 4 secondssafety distance threshold 12metersHD Lower Threshold 15metersHD Upper Threshold 4meters

72 Accuracy of the Driving Behavior Detection First weevaluated the accuracy of SenSafe in determining driv-ing behaviors around the Beijing University of Posts andTelecommunicationsThe roadsweused for testing are shownin Figure 15 The car we used for the test was DongfengPeugeot 307 During these experiments the smartphoneswere mounted on the windshield

We tested the driving behaviors including turn left turnright acceleration brake single-lane change multiple-lanechange and turn back The difference between the single-lane change and multiple-lane change is that multiple-lanechange needs more horizontal displacement The test resultsare shown in Figure 16 The real number of times for eachdriving behavior is manually recorded From the figure wecan see that almost all of the turn left turn right acceleration

Driving behaviors detection

SenSafeRealError rate

010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k

Figure 16 Driving behaviors detection results

and brake have been detected 88 singe-lane change 91multiple-lane change and 84 turn back can be successfullydetected

We furthermore summarized the horizontal displace-ment and angle change of heading for single-lane changemultiple-lane change and turning back in Tables 2 3 and 4


Table 2 Horizontal displacement and angle change of heading forsingle-lane change

Displacement 1 2 3 4 5 AverageSenSafe (m) 191 161 306 21 245 2226Real (m) 192 159 3 19 227 2136Heading angle change 06 086 163 066 424 1598

Table 3 Horizontal displacement and angle change of heading formultiple-lane change


Table 4 Horizontal displacement and angle change of heading forturning back

Displacement 1 2 3 4 5 AverageSenSafe (m) 788 747 809 701 891 787Real (m) 907 769 951 1021 779 8854Heading anglechange 17945 1724 17948 18398 18386 179834

The real horizontal displacement is calculated using the speedfrom the OBD of the vehicles From Tables 2 3 and 4 wecan see that SenSafe can provide the accurate horizontaldisplacement and angle change of heading

73 Performance of Communication We test the performanceof the communication considering the relationship betweenthe distance and the probability of successfully receiving atleast one packet per second

We used one smartphone to broadcast the beacons andused the other to scan the beacons The results are shown inTable 5 From the table we can see that when the distanceis less than 30 meters the client can almost receive at leastone packet per second After the distance is beyond 50m theprobability has an obvious drop

74 Performance of Collision Forewarning We tested theperformance of collision forewarning considering the alertfor the brake in front acceleration in behind and intersectioncollision forewarning

For the brake in front we used the front vehicle to makethe brake to see if the behind vehicle can get the alert Thedistance between the two vehicles was around 30 meters Wetested this scenario ten times and the behind vehicles hadgotten the alert ten times

For acceleration in behind we used the behind vehicleto make the acceleration to see if the front vehicle can getthe alert The distance between the two vehicles was around30 meters We tested this scenario ten times and the frontvehicles had gotten the alert ten times

Table 5 Relationship between the distance and the probability ofsuccessfully receiving at least one packet per second

Distance(m)

Total time(second) Seconds no packet Probability

10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54

For intersection collision forewarning we tested it in thecampus of Beijing University of Posts and Telecommunica-tions Considering that the probability of successfully receiv-ing at least one packet per second is less than 80 when thedistance is greater than the 40 meters we add 1 more secondin safety reaction time threshold and safety intersection timethreshold when the distances between vehicles are greaterthan the 40 meters We used two vehicles to meet at theintersection to see if the driverwill be reminded of the comingvehicles The speed of the vehicle was around 6ms One ofthe vehicles accelerated for 1 second with acceleration around1ms2 to trigger the beacon broadcasting We tested thisscenario ten times and the listening vehicle had gotten thealert nine times We used one vehicle and a pedestrian tomeet at the intersection to see if the driver will be alertedof the coming pedestrian The speed of the pedestrian wasaround 2ms and the speed of the vehicle was around 6msWe tested this scenario ten times and the vehicle had gottenthe alert nine times

8 Conclusion

In this paper we develop a smartphone-based traffic safetyframework named SenSafe to sense the surrounding eventsand provide alerts to drivers Firstly a driving behaviorsdetectionmechanism is providedwhich can runon commod-ity smartphones Secondly theWi-Fi association and authen-tication overhead is reduced to broadcast the compressedsensing data using the Wi-Fi beacon to inform the driversof the surroundingsThirdly a collision estimation algorithmis proposed to provide the appropriate warnings Finally anAndroid-based implementation of SenSafe has been achievedto evaluate the performance of SenSafe in real environmentsand experimental results show that SenSafe can work well inreal environments

Competing Interests

The authors declare that they have no competing interests

Acknowledgments

This work was supported by the National Key RampD Programof China (2016YFB0100902)


References

[1] Real time traffic accidents statistics httpwwwicebikeorgreal-time-traffic-accident-statistics

[2] Intelligent Transportation Systems-Dedicated Short RangeCommunications httpwwwitsdotgovDSRC

[3] N D Lane E Miluzzo H Lu D Peebles T Choudhury andA T Campbell ldquoA survey of mobile phone sensingrdquo IEEECommunications Magazine vol 48 no 9 pp 140ndash150 2010

[4] Applications httpwwwmobileyecomtechnologyapplica-tions

[5] Drivea-driving assistant app httpsplaygooglecomstoreappsdetailsid=comdriveassistexperimentalamphl=en

[6] Turn Your Smartphone Into a Personal Driving Assistanthttpwwwionroadcom

[7] AugmentedDriving httpsitunesapplecomusappaugment-ed-drivingid366841514mt=8

[8] C-W You N D Lane F Chen et al ldquoCarSafe app alertingdrowsy and distracted drivers using dual cameras on smart-phonesrdquo in Proceedings of the 11th Annual International Confer-ence onMobile Systems Applications and Services (MobiSys rsquo13)pp 13ndash26 ACM Taipei Taiwan June 2013

[9] S Hu L Su H Liu HWang and T F Abdelzaher ldquoSmartroadsmartphone-based crowd sensing for traffic regulator detectionand identificationrdquo ACM Transactions on Sensor Networks vol11 no 4 article 55 2015

[10] ZWu J Li J Yu Y Zhu G Xue andM Li ldquoL3 sensing drivingconditions for vehicle lane-level localization on highwaysrdquo inProceedings of the 35th Annual IEEE International Conferenceon Computer Communications (IEEE INFOCOM rsquo16) pp 1ndash9IEEE San Francisco Calif USA April 2016

[11] D Shin D Aliaga B Tuncer et al ldquoUrban sensing usingsmartphones for transportation mode classificationrdquo Comput-ers Environment and Urban Systems vol 53 pp 76ndash86 2015

[12] J Yu H Zhu H Han et al ldquoSenSpeed sensing driving con-ditions to estimate vehicle speed in urban environmentsrdquo IEEETransactions on Mobile Computing vol 15 no 1 pp 202ndash2162016

[13] YWang J Yang H Liu Y ChenM Gruteser and R PMartinldquoSensing vehicle dynamics for determining driver phone userdquoin Proceedings of the 11th Annual International Conference onMobile Systems Applications and Services (MobiSys rsquo13) pp 41ndash54 ACM Taipei Taiwan June 2013

[14] C Saiprasert T Pholprasit and SThajchayapong ldquoDetection ofdriving events using sensory data on smartphonerdquo InternationalJournal of Intelligent Transportation Systems Research 2015

[15] D Chen K-T Cho S Han Z Jin and K G Shin ldquoInvisiblesensing of vehicle steering with smartphonesrdquo in Proceedingsof the 13th Annual International Conference on Mobile SystemsApplications and Services (MobiSys rsquo15) pp 1ndash13 Florence ItalyMay 2015

[16] CohdaWireless httpwwwcohdawirelesscom[17] L Zhenyu P Lin Z Konglin and Z Lin ldquoDesign and

evaluation of V2X communication system for vehicle andpedestrian safetyrdquoThe Journal of China Universities of Posts andTelecommunications vol 22 no 6 pp 18ndash26 2015

[18] U Hernandez-Jayo I De-La-Iglesia and J Perez ldquoV-alertdescription and validation of a vulnerable road user alert systemin the framework of a smart cityrdquo Sensors vol 15 no 8 pp18480ndash18505 2015

[19] W Sun C Yang S Jin and S Choi ldquoListen channel random-ization for faster Wi-Fi direct device discoveryrdquo in Proceedingsof the 35th Annual IEEE International Conference on ComputerCommunications (IEEE INFOCOM rsquo16) IEEE San FranciscoCalif USA April 2016

[20] K Dhondge S Song B-Y Choi and H Park ldquoWiFiHonksmartphone-based beacon stuffedWiFi Car2X-communicationsystem for vulnerable road user safetyrdquo inProceedings of the 79thIEEE Vehicular Technology Conference (VTC rsquo14) pp 1ndash5 IEEESeoul South Korea May 2014

[21] P Zhou M Li and G Shen ldquoUse it free instantly knowingyour phone attituderdquo in Proceedings of the 20th ACM AnnualInternational Conference on Mobile Computing and Network-ing (MobiCom rsquo14) pp 605ndash616 ACM Maui Hawaii USASeptember 2014

[22] R Chandra J Padhye L Ravindranath and A WolmanldquoBeacon-stuffing Wi-Fi without associationsrdquo in Proceedingsof the 8th IEEE Workshop on Mobile Computing Systems andApplications (HotMobile rsquo07) pp 53ndash57 Tucson AZ USAFebruary 2007

[23] M M Minderhoud and P H L Bovy ldquoExtended time-to-collision measures for road traffic safety assessmentrdquo AccidentAnalysis amp Prevention vol 33 no 1 pp 89ndash97 2001

[24] K Vogel ldquoA comparison of headway and time to collision assafety indicatorsrdquo Accident Analysis amp Prevention vol 35 no 3pp 427ndash433 2003

Submit your manuscripts athttpwwwhindawicom

Computer Games Technology

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Distributed Sensor Networks


Advances in

FuzzySystems

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014


ReconfigurableComputing

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014


Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

Artificial Intelligence

HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014

Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi Publishing Corporation

httpwwwhindawicom Volume 2014

Advances in

Multimedia


Biomedical Imaging


ArtificialNeural Systems

Advances in


RoboticsJournal of



Computational Intelligence and Neuroscience

Industrial EngineeringJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Human-ComputerInteraction

Advances in

Computer EngineeringAdvances in



lack of communication vehicles need to make decisions bythemselves

With the widespread use of the smartphone most ofdrivers and pedestrians have smartphones which providesthe potential to use them for the enhancement of trafficsafety Nowadays smartphones are progressively equippedwith functional sensors (eg accelerometer gyroscope GPScamera etc) and these sensors can be applied to recognizeand monitor various vehicle behaviors Additionally theycontain communication resources and enable the quickdeployment of new applications [3]

Sensor technology available in smartphones enables themonitoring of vehicle mobility behaviors including lanechange turns acceleration and brake If such informationcan be sensed and transmitted to other vehicles or pedes-trians they can be potentially used in collision preventionearly crash detection congestion avoidance and so forthSmartphones have abundant communication resources suchas Bluetooth Wi-Fi and 3GLTE However using themto construct appropriate networks for efficient informationsharing in transportation systems is difficult In particularthe transmission range of the Bluetooth cannot satisfy therequirements of the communication Wi-Fi has the proces-sion of authentication and association before data transmis-sion and is more suitable for 1-to-N communication 3GLTEneeds the assistance of the base station

In this paper we propose SenSafe a smartphone-basedframework for traffic safety by sensing communicationand alerting which overcomes the limitations of requestingadditional professional equipment inherent in the existingapproaches

SenSafe is a novel framework which only utilizes thesmartphone to improve the transportation safety It detectsand reports vehiclersquos events based on the sensing data col-lected from steering behaviors in urban area The changes inthe angle of vehicle heading and the corresponding displace-ment during a steering maneuver are calculated for drivingbehavior classification Driving behaviors are classified intodifferent types including turn lane change accelerationand brake Beacon-based communication is provided toexchange the driving information Surrounding environmentreminding and collision forewarning are provided based onthe data from the surroundings

We highlight our main contributions as follows

(i) We propose a framework SenSafe which only usesthe smartphone to sense the driving behavior ex-change the driving information and afford thereminding and alerting

(ii) We provide the detection and differentiation of var-ious driving behaviors by only utilizing the smart-phonersquos built-in sensors in urban areaWe analyze thesensing data collected from urban area and find thatvehicles cannot always make turns and lane changessmoothly owing to the influence of surrounding vehi-cles and pedestriansThe temporary interruptions aretaken into consideration to improve the accuracy ofthe driving behaviors detection

(iii) We furnish the beacon-based communication whichmodifies the service set identifier (SSID) field ofWi-Fibeacon frame to carry the driving behavior positionand mobility information A data compression anddecompressionmethod is provided tomake use of theSSID field Event-driven communication is utilizedto provide the surrounding drivers with immediatereminding Ordinary promptings of the surroundingvehiclesrsquo behaviors and safety alerts of the comingcollisions are provided to the drivers

(iv) AnAndroid-based implementation of SenSafe frame-work has been achieved to demonstrate the evaluationresults and application reliability in real environ-ments

The rest of this paper is organized as follows In Section 2a brief overview of the related works is provided In Section 3the framework overview of SenSafe is introducedThe detailsfor three modules of SenSafe are explained in Sections 4ndash6 The implementation performance evaluation and exper-iment results are shown in Section 7 Finally this paper isconcluded in Section 8

2 Related Works

There are some works to detect the vehicle behavior Someworks use the fixed vehicle-mounted devices to monitorthe vehicle behavior Mobileye uses cameras and radars toprovide the technical support for the auto manufacturers [4]Although the cost of vehicle safety technology is decreasingthese safety technologies have not been deployed in economyvehicles It still needs time before the majority of vehiclesare deployed with these safety technologies With the widespread of smartphones smartphone solutions can be usedin vehicles and there are some works focusing on usingsmartphones to assist drivers

Drivea [5] iOnRoad [6] and Augmented Driving [7]are apps which have capability of detecting lane departuresand warn drivers when the distance to the front vehicle istoo close CarSafe [8] alerts distracted drivers using dualcameras on smartphones one for detecting driver state andthe other for tracking road conditions Although camera-based systems have the functionality in assisting the driverthey have limitations in terms of computational overhead andrequirement of image quality and work worse at night andwith bad lighting conditions

Camera-free sensors in the smartphone have been uti-lized in traffic regulator detection [9] localization [10]transportation mode classification [11] vehicle speed esti-mation [12] and the driving behavior detection [10 13ndash15]In [13] smartphone sensing of vehicle dynamics is utilizedto determine driver phone use Inertial measurement unit(IMU) sensors including accelerometers and gyroscopes ofthe smartphone are used to capture differences in centripetalacceleration due to vehicle dynamics In [14] three algo-rithms which detect driving events using motion sensorsembedded on a smartphone are proposedThe first detectionalgorithm is based on data collected from GPS receiver Thesecond and third detection algorithms utilize accelerometer




GPS



BeaconBeacon

Detections











Eventdetection


Driving behaviors


GyroscopeGPS


Mobilityinformation






1015840




Cone

Ze

Xe

Ye

Ys

120575

Y998400


Acce

lera

ted

spee

d (m

2s

)

TPending

ca ca

0

05

1

15

2

25

3

35

4

2010 15 250 5Samples






cele

rate

d sp

eed

(m2s

)

cb cb

minus25

minus2

minus15

minus1

minus05

0

5 10 15 20 250Samples

TPending


Waiting-for-Acc

Acc-Pending

Acceleration Brake

Brake-Pending

No Yes

Yes

No

Yes

No

Yes

No















Turn right

minus07

minus06

minus05

minus04

minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples



Turn left

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples





minus05

minus04

minus03

minus02

minus01

0

01

02

Ang

ular

spee

d (r

ads

)

50 100 150 200 250 300 3500Samples


Two bumps turn left


50 100 150 200 250 3000Sample

minus1

minus05

0

05

1

15A

ngul

ar sp

eed

(rad

s)




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples


Figure 7 Continued


Turn back20 40 60 80 100 120 140 160 1800

Samples


minus01

0

01

02

03

04

05

06

07

08

Ang

ular

spee

d (r

ads

)

(g) Turn back


as asat

TBump

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples




asas as

as

TBump

TDelay


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

at

at










No-Bump

One-Bump

Waiting-for-Bump

More-Bump

Lane-Change

Turn

NoYes


Yes


Yes

Yes

Bump-End


No

Yes

No

No








120579 =119868

sum119894=1

119873119894

sum119899119894=1

119860119899119894Δ119905 (1)












120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)







116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872




202059 121

59121

202100 136

Remote data

Local data

Compressed data


Time

202159 121202059 121








Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






GPS



BeaconBeacon

Detections











Eventdetection


Driving behaviors


GyroscopeGPS


Mobilityinformation






1015840




Cone

Ze

Xe

Ye

Ys

120575

Y998400


Acce

lera

ted

spee

d (m

2s

)

TPending

ca ca

0

05

1

15

2

25

3

35

4

2010 15 250 5Samples






cele

rate

d sp

eed

(m2s

)

cb cb

minus25

minus2

minus15

minus1

minus05

0

5 10 15 20 250Samples

TPending


Waiting-for-Acc

Acc-Pending

Acceleration Brake

Brake-Pending

No Yes

Yes

No

Yes

No

Yes

No















Turn right

minus07

minus06

minus05

minus04

minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples



Turn left

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples





minus05

minus04

minus03

minus02

minus01

0

01

02

Ang

ular

spee

d (r

ads

)

50 100 150 200 250 300 3500Samples


Two bumps turn left


50 100 150 200 250 3000Sample

minus1

minus05

0

05

1

15A

ngul

ar sp

eed

(rad

s)




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples


Figure 7 Continued


Turn back20 40 60 80 100 120 140 160 1800

Samples


minus01

0

01

02

03

04

05

06

07

08

Ang

ular

spee

d (r

ads

)

(g) Turn back


as asat

TBump

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples




asas as

as

TBump

TDelay


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

at

at










No-Bump

One-Bump

Waiting-for-Bump

More-Bump

Lane-Change

Turn

NoYes


Yes


Yes

Yes

Bump-End


No

Yes

No

No








120579 =119868

sum119894=1

119873119894

sum119899119894=1

119860119899119894Δ119905 (1)












120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)







116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872




202059 121

59121

202100 136

Remote data

Local data

Compressed data


Time

202159 121202059 121








Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




Eventdetection


Driving behaviors


GyroscopeGPS


Mobilityinformation






1015840




Cone

Ze

Xe

Ye

Ys

120575

Y998400


Acce

lera

ted

spee

d (m

2s

)

TPending

ca ca

0

05

1

15

2

25

3

35

4

2010 15 250 5Samples






cele

rate

d sp

eed

(m2s

)

cb cb

minus25

minus2

minus15

minus1

minus05

0

5 10 15 20 250Samples

TPending


Waiting-for-Acc

Acc-Pending

Acceleration Brake

Brake-Pending

No Yes

Yes

No

Yes

No

Yes

No















Turn right

minus07

minus06

minus05

minus04

minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples



Turn left

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples





minus05

minus04

minus03

minus02

minus01

0

01

02

Ang

ular

spee

d (r

ads

)

50 100 150 200 250 300 3500Samples


Two bumps turn left


50 100 150 200 250 3000Sample

minus1

minus05

0

05

1

15A

ngul

ar sp

eed

(rad

s)




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples


Figure 7 Continued


Turn back20 40 60 80 100 120 140 160 1800

Samples


minus01

0

01

02

03

04

05

06

07

08

Ang

ular

spee

d (r

ads

)

(g) Turn back


as asat

TBump

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples




asas as

as

TBump

TDelay


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

at

at










No-Bump

One-Bump

Waiting-for-Bump

More-Bump

Lane-Change

Turn

NoYes


Yes


Yes

Yes

Bump-End


No

Yes

No

No








120579 =119868

sum119894=1

119873119894

sum119899119894=1

119860119899119894Δ119905 (1)












120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)







116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872




202059 121

59121

202100 136

Remote data

Local data

Compressed data


Time

202159 121202059 121








Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




cele

rate

d sp

eed

(m2s

)

cb cb

minus25

minus2

minus15

minus1

minus05

0

5 10 15 20 250Samples

TPending


Waiting-for-Acc

Acc-Pending

Acceleration Brake

Brake-Pending

No Yes

Yes

No

Yes

No

Yes

No















Turn right

minus07

minus06

minus05

minus04

minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples



Turn left

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples





minus05

minus04

minus03

minus02

minus01

0

01

02

Ang

ular

spee

d (r

ads

)

50 100 150 200 250 300 3500Samples


Two bumps turn left


50 100 150 200 250 3000Sample

minus1

minus05

0

05

1

15A

ngul

ar sp

eed

(rad

s)




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples


Figure 7 Continued


Turn back20 40 60 80 100 120 140 160 1800

Samples


minus01

0

01

02

03

04

05

06

07

08

Ang

ular

spee

d (r

ads

)

(g) Turn back


as asat

TBump

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples




asas as

as

TBump

TDelay


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

at

at










No-Bump

One-Bump

Waiting-for-Bump

More-Bump

Lane-Change

Turn

NoYes


Yes


Yes

Yes

Bump-End


No

Yes

No

No








120579 =119868

sum119894=1

119873119894

sum119899119894=1

119860119899119894Δ119905 (1)












120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)







116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872




202059 121

59121

202100 136

Remote data

Local data

Compressed data


Time

202159 121202059 121








Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




Turn right

minus07

minus06

minus05

minus04

minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples



Turn left

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples





minus05

minus04

minus03

minus02

minus01

0

01

02

Ang

ular

spee

d (r

ads

)

50 100 150 200 250 300 3500Samples


Two bumps turn left


50 100 150 200 250 3000Sample

minus1

minus05

0

05

1

15A

ngul

ar sp

eed

(rad

s)




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples




minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples


Figure 7 Continued


Turn back20 40 60 80 100 120 140 160 1800

Samples


minus01

0

01

02

03

04

05

06

07

08

Ang

ular

spee

d (r

ads

)

(g) Turn back


as asat

TBump

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples




asas as

as

TBump

TDelay


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

at

at










No-Bump

One-Bump

Waiting-for-Bump

More-Bump

Lane-Change

Turn

NoYes


Yes


Yes

Yes

Bump-End


No

Yes

No

No








120579 =119868

sum119894=1

119873119894

sum119899119894=1

119860119899119894Δ119905 (1)












120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)







116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872




202059 121

59121

202100 136

Remote data

Local data

Compressed data


Time

202159 121202059 121








Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




Turn back20 40 60 80 100 120 140 160 1800

Samples


minus01

0

01

02

03

04

05

06

07

08

Ang

ular

spee

d (r

ads

)

(g) Turn back


as asat

TBump

minus02

minus01

0

01

02

03

04

05

06

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 140 160 1800Samples




asas as

as

TBump

TDelay


minus03

minus02

minus01

0

01

02

03

Ang

ular

spee

d (r

ads

)

20 40 60 80 100 120 1400Samples

at

at










No-Bump

One-Bump

Waiting-for-Bump

More-Bump

Lane-Change

Turn

NoYes


Yes


Yes

Yes

Bump-End


No

Yes

No

No








120579 =119868

sum119894=1

119873119894

sum119899119894=1

119860119899119894Δ119905 (1)












120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)







116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872




202059 121

59121

202100 136

Remote data

Local data

Compressed data


Time

202159 121202059 121








Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




No-Bump

One-Bump

Waiting-for-Bump

More-Bump

Lane-Change

Turn

NoYes


Yes


Yes

Yes

Bump-End


No

Yes

No

No








120579 =119868

sum119894=1

119873119894

sum119899119894=1

119860119899119894Δ119905 (1)












120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)







116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872




202059 121

59121

202100 136

Remote data

Local data

Compressed data


Time

202159 121202059 121








Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





120579119899 =119899

sum119894=1

119860 119894Δ119905

119867 =119873

sum119899=1

119866119899Δ119905 sin (120579119899)

(2)







116364815 39967001202020 234

20234 64815 67001

202021 136

Remote data

Local data

Compressed data

116365173 39966872




202059 121

59121

202100 136

Remote data

Local data

Compressed data


Time

202159 121202059 121








Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






Front

Left Right

Behind

Moving direction

45∘

45∘







Ho

Vo

Hm

Vm

(Xo Yo)

(Xm Ym)







Acc Threshold 08ms2










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in










010203040506070

Tim

es

000200400600801012014016018

Turn

left

Turn

righ

t

Acce

lera

tion

Brak

e

Sing

le-la

ne ch

ange

Mul

tiple

-lane

chan

ge

Turn

bac

k


















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in

















Distance(m)


10 1200 0 10020 1200 39 9730 1200 72 9440 1200 234 8150 1200 354 7160 1200 557 54


8 Conclusion


Competing Interests


Acknowledgments



References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




References
































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in










Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in



research article sensafe: a smartphone-based traffic...

Documents