self study

A Road Condition Sharing System Using Vehicle- to-Vehicle Communication in Various

Communication Environment Kenta Ito

Graduate School of Software and Information Science, Iwate Prefectural University

Iwate, Japan [email protected]

Yoshikazu Arai Faculty of Software and Information Science,

Iwate Prefectural University Iwate, Japan

[email protected]

Go Hirakawa Faculty of Software and Information Science,

Iwate Prefectural University Iwate, Japan

[email protected]

Yoshitaka Shibata Faculty of Software and Information Science,

Iwate Prefectural University Iwate, Japan

[email protected]

Abstract—. In this paper, we introduce a road condition sharing system using Vehicle-to-Vehicle communication in various communication environment. Japan is prone to natural disasters. In addition, traffic hazards and traffic accidents occur due to snow. A wide range and quick condition understanding and monitoring are needed. In addition, it is important to understand the condition of the destination and the road condition to the destination. But, especially after a disaster has occurred, areas that mass media can provide disaster condition information are limited. Recently, various higher technologies have been developed and noticed. By using those technologies, we develop our system. Using our system, we can realize road condition understanding, monitoring and recording as an alert information using multiple sensor data, information sharing between each vehicles and information is provided as web application. As a quantitative evaluation, we measure vehicle- to-vehicle communication quality.

Keywords; Road monitoring system, Sensor data gathering, DTN, ITS, Vehecle-to-Vehicle communication

I. INTRODUCTION Japan is prone to natural disasters including earthquakes,

landslide disasters and typhoons. , Recently, large scale disasters have occurred such as Eastern Japan Great Earthquake on March 11, 2011, landslide disaster by heavy rain in Hiroshima on August, 2014 and earthquake in Nagano on November, 2014. In winter, it is prone to hazards and disasters due to snow. For example, a snow slide accident in Tamagawa hot spring, Akita on February 2012 and isolation accident in Tokushima on December 2014. In addition, traffic hazards and traffic accidents occur due to road surface

freezing, snowstorms and whiteout on a daily basis. These are serious social problems.

A wide range and quick condition understanding and monitoring are needed after a disaster has occurred. In addition, it is important for commuting, disaster response and disaster relief to understand the condition of the destination and the road condition to the destination. But, especially after a disaster has occurred, due to hazard of communication infrastructure and network congestion, areas that mass media can provide disaster condition information are limited. For example, Figure 1 shows Eastern Japan Great Earthquake mass media coverage map[1][2].

This project visualize information which is provided by

mass media, disaster condition and areas that needed support after a disaster has occurred. It says that there is a deference between areas that a lot of conditions are reported and areas that have shelters. Therefore, a wide range understanding of disaster condition is difficult.

Figure 1 Mass media coverage map

2015 9th International Conference on Complex, Intelligent, and Software Intensive Systems

978-1-4799-8870-9/15 $31.00 © 2015 IEEE

DOI 10.1109/CISIS.2015.33

242

2015 Ninth International Conference on Complex, Intelligent, and Software Intensive Systems

978-1-4799-8870-9/15 $31.00 © 2015 IEEE

DOI 10.1109/CISIS.2015.33

242

Recently, various higher technologies have been developed and noticed such as sensor data gathering technology, wireless network technology, Delay Tolerant Networking(DTN)[3][4], Intelligent Transport System(ITS)[5] and Big Data[6].

In our study, by using those technologies, we developed a road alert information sharing system with multiple vehicles using vehicle-to-vehicle communication in various communication network environment, which is named SODiCS(Spatial and temporal Omnidirectional sensor data Distribution and Collection System). Using this system, we can realize road condition understanding, monitoring and recording using multiple sensor data as alert information, information sharing between each vehicles and information is provided as web application. We especially focus information sharing between each vehicles. We aim at information sharing between each vehicles that run at a speed of 40kph. In addition, we construct proto type of vehicle-to- vehicle communication, experiment and evaluate this proto type to confirm usability of this system. As a quantitative evaluation, we measure vehicle-to-vehicle communication quality.

II. RELATED WORK

A. Delay Tolerant Networking(DTN) DTN is an overlay network configuration which implements new

protocols above the Transport Layer in the OSI reference model, called Bundle Layer and Convergence Layer.

DTN is an approach which provides interoperable communication where continuous end-to-end connectivity cannot be assumed. Although the current TCP protocol needs the end-to- end connectivity to establish network communication, DTN has been architected for the environments that end-to-end communication cannot be available such as heterogeneous, interplanetary, military, and disaster network.

DTN was originally architected for interplanetary communication. In case of long distance communication such as an interplanetary communication, there are no continuous network connections. So long distance communication needed a new network protocol which provides interoperable communication.

To realize interoperable communication under a challenging network environment, DTN creates a “store-carry-forward” protocol for its routing. Each node stores the transmission data if there is no available node nearby, and the data is replicated when a node comes close enough to other nodes. The data is replicated to the other node in the same way until the data reaches the destination node.

There are many forms of routing methods in DTN. For example, Epidemic Routing, Spray and Wait, Max Prop and so on. In addition, there are a lot of study for DTN. For example, people move between areas and provide DTN function as Bytewalla[7] and a routing method using Ant Colony Optimization[8]. We use these method for our system. We also study and implement a routing method that extends the existing method.

B. Dedicated Short Range Communications(DSRC) DSRC[9][10][11] is a communication method using 5.8GHz

band to aim at interactive communication for narrow areas.

DSRC is specialized wireless communication to vehicles. By directivity of antenna and high precision carrier sense, the communication area is controlled intentional and high capacity information is transmitted and received at high speed. In our study, we compare our system and DSRC about communication range, communication speed and quantity of transmitted data.

C. CoMoSE platform CoMoSE platform[12] gathers various information that

various things have. For example, people, vehicle, surrounding environment and ITS(Figure 2). These information are stored to an in-vehicle server. CoMoSE platform provides service for people, vehicle, surrounding environment and ITS using gathered data.

Figure 3 shows a scheme of CoMoSE platform. Using this

scheme, CoMoSE platform can provide service using various information. The usage examples of this platform are as follows: � The in-vehicle server gathers information of in-

vehicle sensors suitable. � A driver asks in-vehicle server sensor information

which service is needed using smart device. � The in-vehicle server searches information from in-

vehicle sensors, sensors put outside a vehicle, other in-vehicle server, ITS and cloud storage and provides these sensor information.

Figure 2 Concept of CoMoSE platform

Figure 3 Block diagram of CoMoSE platform

243243

� The driver can confirm provided sensor information using smart device.

Figure 4 shows CoMoSE platform’s sensor node

architecture and Figure 5 shows CoMoSE platform’s sensor server architecture. In our study, we use this platform function which stores in-vehicle sensors information to database.

D. Quasi Electrostatic Field(QEF) QEF[13] is a special electric field which doesn’t include

magnetic field components that construct electromagnetic field. QEF doesn’t have transmitting quality like a radio wave. QEF is a physical phenomenon which is distributed surrounding substance like an electrostatic charge. Using QEF, contactless communication is realizable very low energy by comparison a radio wave and contactless sensor which is very sharp and wireless is developable to understand change of QEF surrounding people and vehicle.

In our study, we use this technology as a road surface condition sensor.

E. Road condition monitoring system As related works of road condition monitoring system,

road monitoring system using seismic motion sensor, sound sensor and image sensor after disaster has occurred[14], road condition monitoring using three-axis acceleration sensor and

GPS sensor[15] and road condition monitoring and alert application using in-vehicle smartphone as sensor[16]. In these studies, a few sensors are used. We think that using a lot of sensors and selection data for purpose are available.

F. Vehicle-to-Vehicle Communication As related works of vehicle-to-vehicle communication,

many of them measured using various communication protocol, various technology on simulator[17][18][19][20][21].

In our study, we set a high value on a communication experimentation in real environment.

III. SODICS CONFIGURATION

A. System configuration Our system consists of multiple In-vehicle Server and

Information Server(Figure 6). In-vehicle Server consists Sensor Server and Sensor Node, gathers various sensor data and stores them to Sensor Server storage. If it is possible to get stability network connection, In-vehicle Server transmits gathered sensor data to Information Server. If it is impossible, In-vehicle Server keeps storing sensor data and transmits sensor data to Information Server when network connection is available. Also, In-vehicle Server provides sensor data to users directly and shares information between each vehicles. Information Server provides information to users as a web application. As operation environments of this system, we suppose to get road condition information such as traffic jam, road surface freezing and damage situation in cases of

Figure 4 Sensor node architecture

Figure 5 Sensor server architecture

Figure 6 System configuration

Figure 7 Example sensor data usage

244244

commuting, going to a tourist destination, disaster response and disaster relief.

Figure 7 shows sensor data examples. Currently, our system can get sensor data such as latitude, longitude, temperature, humidity, acceleration, angular velocity, direction, road condition and image from Sensor Node. Our system chooses required sensor data from all sensor data and provides converted and analyzed sensor data to users.

In addition, as a data distribution function, Information Server provides Live mode, Archive mode and Forecast mode. Live mode is used as live streaming and monitoring camera. Archive mode is used to browse the past sensor data. Forecast mode is used visualization of forecast road condition by analyzed sensor data.

B. In-vehicle Server’s function Figure 8 shows In-vehicle Server’s function. In-vehicle

Server gathers sensor data. For example, temperature, humidity, image, direction, location, acceleration, angular velocity and road surface condition. In-vehicle Server stores, analyzes and converts these sensor data. In-vehicle Server shares raw sensor data with Information Server using vehicle- to-roadside communication. In-vehicle Server also shares processed sensor data with other In-vehicle Server using vehicle-to-vehicle communication.

C. Information Server function

Figure 9 shows Information Server’s function. Information Server receives transmitted raw sensor data. Information Server stores, analyzes and converts sensor data. Information Server provides processed sensor data as a web application.

D. Deference of shared data by communication method Figure 10 shows that in our system, in case of vehicle-to-

vehicle communication, each vehicles share processed sensor data as alert information using wireless network by In-vehicle Server’s access point. For example, road surface freezing, fallen snow and skid. In case of vehicle-to-roadside communication, vehicles transmit all gathered raw sensor data to server using wireless networks. In case of roadside- to-vehicle communication, server provides processed sensor data as a web application.

IV. SYSTEM ARCHITECTURE Figure 11 shows an architecture of vehicle-to-vehicle

communication system. Each modules are as follows. � DB Connection: Connection to database � DB Outputting: Outputting sensor data to text file � V2V Connection Checker: Checking connection to

In-vehicle Receiver Server’s access point � Data Sender: Transmitting sensor data by socket � Data Receiver: Receiving sensor data by socket � Sensor Data Inputting: Getting sensor data from

received text file and inputting to database

V. DATA TRANSMISSION FLOW We have considered some of data transmission flow.

Vehicle storing road condition information is called In- vehicle Sender Server and vehicle which wants to understand road condition information from current location to destination is called In-vehicle Receiver Server.

On the one hand, Receiving Sensor Server creates server socket and waits for socket from Sending Sensor Server. Sending Sensor Server checks connection to Receiving Sensor Server’s access point. If connection is available, Sending Sensor Server acquires image file name from

Figure 8 In-vehicle Server’s function

Figure 9 Information Server’s function

Figure 10 Deference of shared data

245245

directory which images are stored, acquired image file name is stored to array and the array is sorted to ascending order. When these processes are finished, Sending Sensor Server creates socket and requires connection to Receiving Sensor Server. Receiving Sensor Server accepts socket connection request and creates input and output stream. If socket connection request is accepted, Sending Sensor Server creates input and output stream. After these processes, image files are converted to byte data and transmitted by socket. If all image files are finished transmitting, stream and socket are closed and these processes end. Sending Sensor Server repeated from socket connection request to stream and socket

are closed if image file exists. Receiving Sensor Server receives transmitted data, converts byte data to image file and stores optical directory. When these processes are finished, stream and socket are closed. Receiving Sensor Server repeats from waiting for socket connection to stream and socket are closed while socket is been receiving(Figure 12).

On the other hand, In-vehicle Receiver Server creates

server socket and waits for socket from In-vehicle Sender Server. In-vehicle Sender Server connects database storing sensor data using JDBC driver, outputs alert information to text file and finishes connecting database. Side by side with this processing, In-vehicle Sender Server checks connection to In-vehicle Receiver Server’s access point using isRearchable() method. This method checks reachable to In- vehicle Receiver Server’s IP address. If connection is available, In-vehicle Sender Server creates socket and requires connection to In-vehicle Receiver Server. In-vehicle Receiver Server accepts socket connection request and creates input and output stream. If socket connection request is accepted, text file is transmitted to In-vehicle Receiver Server by socket. In-vehicle Receiver Server receives transmitted file(Figure 13).

VI. PROTOTYPE SYSTEM Figure 14 shows prototype system of SODiCS. Table 1

shows a construction of In-vehicle Sever and Information Server. In-vehicle Server is composed from Sensor Server and Sensor Node. In addition, PHP, Apache and MySQL are used to provide application and Java is used for vehicle-to- vehicle communication. CoMoSE platform is used as Sensor Node. Information Server uses PHP, Apache and MySQL to provide application and Open CV will be used for image analysis and image processing.

Figure 11 System architecture

Figure 12 Data transmission flow 1

Figure 13 Data transmission flow 2

246246

Table 1 Construction of proto type device Sending

Sensor Server Receiving Sensor Server

Information Server

OS Ubuntu 12.04 LTS CPU Intel Core i3

3217U Intel Core i7 3687U

Intel Xeon X5482

Memory 8GB Storage 128GB SSD 512GB SSD 2TB HDD

VII. COMMUNICATION EXPERIMENTATION We experiment about vehicle-to-vehicle communication in

order to confirm usability of our system and evaluate our system.

A. Experimentation outline As a scenario of our experimentation, we suppose to

Vehicle A and Vehicle B(Figure 15). Vehicle A wants to understand road condition. Vehicle B has road condition that vehicle A wants. Vehicle B transmits data to vehicle A. In this experimentation, In-vehicle Server is set on the vehicle’s roof. Figure 16 shows a prototype system of vehicle-to- vehicle communication in this experimentation. We experimented in parking area at our university. We experimented a day which has no rain, fallen snow and parked vehicle. A. In this experimentation, In-vehicle Server is set on the vehicle’s roof.

B. Experimentation contents and results As an experimentation 1, we measured transmission time.

Vehicle A and Vehicle B were stopped, and distance, size of buffer and the number of transmitting image files were changed(Table 2). Table 3 show a result of experimentation 1.

As an experimentation 2, we measured whether or not all image files can be transmitted, transmission time which all image files can be transmitted and the number of image files that all image files cannot be transmitted. Vehicle A was stopped and Vehicle B was driven. Running speed and the number of transmitting image files(Table 4). Table 5 shows a result of experimentation 2.

As an experimentation 3, we measured the range that transmission speed starts to decrease, Vehicle B cannot transmit data and Vehicle B cannot connect to Vehicle A’s access point, based on experimentation 2. Vehicle A and Vehicle B were stopped. Distance were changed(Table 6). Table 7 shows a result of experimentation 3.

As an experimentation 4, we measured transmission range and bandwidth using ping and iperf, based on experimentation 1 and 3. Vehicle A and Vehicle B were stopped(Table 8). Table 9 and Figure 17 show a result of experimentation 4.

C. Discussion In experimentation 1, transmission speed was kept at the

20m point, but transmission speed was decreased at the 30m point. In experimentation 3 based on experimentation 1, image files could be transmitted while transmission speed at the 30m point, but image files could not be transmitted after the30m point and In-vehicle Sender Server could not connect to In-vehicle Receiver Server’s access point after the 60m point. In experimentation 4 based on experimentation 1 and 3, we could measure transmission range and bandwidth at the 70m point using ping and iperf. Therefore, we found that there is a difference of transmission range between our system and ping, iperf. Also, there is a possibility that transmission range and bandwidth can be measured after the

Figure 14 Prototype system

Figure 15 Experimentation outline

Figure 16 Prototype syste, of V2V communication

247247

70m point. In experimentation 2, image files were transmitted at lower speed, but it is still far from our aim.

Table 2 Palameter of experientation 1 Image size(pixel) 640*360 File size(KB) 30~40 Image acquisition interval(fps) 1 Distance(m) 0, 5, 10, 20, 30 Buffer size(KB) 64, 32, 16 The number of image files 2373, 475, 238

Table 3 A result of experimentation 1 64KB 32KB 16KB 0m, 2373 66.2 54.4 55.5 0m, 475 11.7 11.4 12.1 0m, 238 5.7 5.9 6.7 5m, 2373 63.0 63.1 61.6 5m, 475 13.4 12.6 13.2 5m, 238 6.1 6.1 6.8 10m, 2373 110.4 102.4 105.6 10m, 475 21.8 21.8 22.9 10m, 238 11.6 11.1 18.2 20m, 2373 101.4 89.9 85.0 20m, 475 19.1 18.7 15.1 20m, 238 10.3 10.8 8.1 30m, 2373 568.6 1034.5 1401.7 30m, 475 57.2 127.9 106.0 30m, 238 28.7 33.6 43.3 Transmission time(sec)

Table 4 Parameter of experimentation 2 Image size(pixel) 640*360 File size(KB) 30~40 Image acquisition interval(fps) 1 Speed(kph) 10, 20, 30, 40 Buffer size(KB) 64 The number of image files 2373, 475, 238

Table 5 A result of experimentation 2 20kph

2373 20kph 238

40kph 238

10kph 475

10kph 238

10kph 2373

30kph 238

1 294 11.1 ---

238 9.0 27.5

110 4.2 ---

475 18.0 36.7

238 9.0 31.8

515 19.5 ---

206 7.8 ---

2 255 9.7 ---

238 9.0 19.8

79 3.0 ---

115 4.3 ---

3 269 10.1 ---

238 9.0 21.2

160 6.0 ---

230 8.7 ---

Upper : the number of image files that can be transmitted Middle : a size of transmitted image files(MB) Bottom : transmission time when all image files can be transmitted(sec)

Table 6 Paremeter of experimentation 3 Image size(pixel) 640*360 File size(KB) 30~40 Image acquisition interval(fps) 1 Distance(m) 0, 7.5, 17.5, 22.5, 25, 27.5,

32.5, 60, 62.5 Buffer size(KB) 64 The number of image files 100

Table 7 A result of experimentation 3 0m 7.5m 17.5m 22.5m 25m 27.5m 1 1.77 2.50 8.84 5.05 26.28 8.97 2 1.80 2.44 9.21 4.93 25.18 9.23 3 1.79 2.64 7.83 4.35 14.95 9.50 4 1.76 2.56 33.03 5 1.79 2.26 9.80 Transmission time(sec) 32.5m, 60m : data cannot be transmitted 62.5m : Sensor Server cannot connect to access point

Table 8 Parameter of experimentation 4 The number of trial ping 50 Trial interval of iperf(sec) 5 Trial time of iperf(sec) 60 Distance(m) 0, 5, 10, 15, 20, 25, 30, 35, 40, 50,

60, 70 Table 9 A result of ping

Packet loss

Time RTT min

RTT avg

RTT max

0m 0 49080 1.33 4.54 5.51 5m 0 49073 0.80 9.52 133.42 10m 0 49080 0.94 7.57 129.20 15m 0 49074 3.52 4.60 5.82 20m 0 49078 3.35 4.80 6.68 25m 0 49068 3.35 4.94 8.17 30m 0 49054 3.51 4.90 6.75 35m 0 49077 1.80 5.46 9.90 40m 0 49071 3.43 50.50 11.26 50m 0 49074 1.30 11.54 194.03 60m 0 49071 0.86 13.17 179.24 70m 0 49076 3.49 4.66 9.60 Packet loss : % Time, RTT min, RTT avg, RTT max : msec

VIII. CONCLUSION AND FUTURE WORKS In this paper, we propose a road condition monitoring

system using various sensor data in challenged communication environment. We gather various sensor data, choose sensor data as purpose and use chosen sensor data for road condition understanding. We experiment about vehicle- to-vehicle communication.

In the future, we work these contents: Continuous experimentation for our goal, Improvement system based on measurement result, Considering multiple vehicles, How to find vehicle which have required information, Research, examine and implement about DTN function, Comparison and combination other communication standard such as DSRC and IEEE802.11p

REFERENCES. [1]. https://docs.google.com/document/d/1JvpLeq_hYKNENQ-

hjAGqx1dmbyXdFnLKlfbL8lq-WxM/edit [2]. http://media.mapping.jp/ [3]. http://www.dtnrg.org/wiki/Home [4]. https://sites.google.com/site/dtnresgroup/home

Figure 17 A result of ping

248248

[5]. http://www.its-jp.org/about/ [6]. http://www.soumu.go.jp/johotsusintokei/whitepaper/ja/h24/html/nc1

21410.html [7]. Hervé Ntareme and Sebastian Domancich “Security and performance

aspects of Bytewalla: A Delay Tolerant Network on smartphones”, First International Workshop on Wireless Communication and Networking Technologies for Rural Enrichment

[8]. Chengjun Wang, Baokang Zhao, Wei Peng, Chunqing Wu and Zhenghu Gong, “Routing Algorithm based on Ant Colony Optimization for DTN Congestion Control”, 15th International Conference on Network-Based Information Systems (NBiS), 2012

[9]. http://ja.wikipedia.org/wiki/DSRC [10]. http://www.dsrc-portal.jp/summary/index.html [11]. http://www.soumu.go.jp/soutsu/tokai/musen/its/dsrc.html [12]. Go Hirakawa, Phyu Phy Kywe, Kenta Ito, and Yoshitaka Shibata,

“Application of Automotive Sensor Information Server Platform to Ubiquitous Sensing System”, 17th International Conference on Network-Based Information Systems (NBiS2014), pp. 461-464, 2014

[13]. http://www.takiguchilab.iis.u-tokyo.ac.jp/ [14]. Xu-guang Sun, Xiao-ling Sun, Qiu-ge Yang, and Shu-nan Ma,

“Application of Wireless Sensor Networks in Post-Disaster Road Monitoring System”, 4th International Conference on Intelligent Networks and Intelligent Systems, pp 105-108, 2011

[15]. Kongyang Chen, Mingming Lu, Xiaopeng Fan, Migmig Wei, and Jinwu Wu, “Road Condition Monitoring Using On-board Three-axis Accelerometer and GPS Sensor”, 6th International ICST Conference on Communications and Networking in China (CHINACOM), pp 1032-1037, 2011

[16]. Avik Ghose, Provat Biswas, Chirabrata Bhaumik, Monika Sharma, Arpan Pal, and Abhinav Jha, “Road Condition Monitoring and Alert Application Using In-Vehicle Smartphone as Internet-Connected Sensor”, PerCom Demos 2012, Lugano, pp 489-491

[17]. Ademar Takeo Akabane, Leandro Aparecido Villas, and Edmundo Roberto Mauro Madeira, “GTO: A Broadcast Protocol for Highway Environments over Diverse Traffic Conditions”, 13th International Symposium on Network Computing and Applications, pp 37-40, 2014

[18]. Muhammad Ahsan Qureshi and Rafidah Md. Noor, “Towards Improving Vehicular Communication in Modern Vehicular Environment”, 11th International Conference on Frontiers of Information Technology, pp 177-182, 2013

[19]. Claudio Correa, J´o Ueyama, Rodolfo I. Meneguette, and Leandro A. Villas, “VANets: An Exploratory Evaluation in Vehicular Ad hoc Network For Urban Environment”, 13th International Symposium on Network Computing and Applications, pp 45-49, 2014

[20]. Danda B. Rawat, Bhed B. Bista, Gongjun Yan, and Stephan Olariu, “Vehicle-to-Vehicle Connectivity and Communication Framework for Vehicular Ad-Hoc Networks”, 8th International Conference on Complex, Intelligent and Software Intensive Systems, pp 44-49, 2014

[21]. Seiichiro Ishikawa, Taiki Honda, Makoto Ikeda, and Leonard Barolli, “Performance Analysis of Vehicular DTN Routing under Urban Environment”, 8th International Conference on Complex, Intelligent and Software Intensive Systems, pp 50-53, 2014

249249