A Practical Method for Validating Data Quality by RSE ...

0 downloads 0 Views 356KB Size Report
A Practical Method for Validating Data Quality by. RSE Connectedness : A Case Study in China. Shuxian He1, Jiangchen Li2, Wei Xiong1. Intelligent Transport ...
WK,QWHUQDWLRQDO&RQIHUHQFHRQ7UDQVSRUWDWLRQ,QIRUPDWLRQDQG6DIHW\ ,&7,6 $XJXVW%DQII&DQDGD

A Practical Method for Validating Data Quality by RSE Connectedness : A Case Study in China Shuxian He1, Jiangchen Li2, Wei Xiong1

Tony Z. Qiu12

Intelligent Transport System Research Center1 Wuhan University of Technology Wuhan, Hubei, China [email protected], [email protected]

Department of Civil and Environmental Engineerin2 University of Alberta Edmonton, Canada [email protected] perspectives, which are packet transmission, communication delay and data sending frequency. A study did a comprehensive research on packet transmission to evaluate DSRC data transmission, referred to packet error rate (PER), different packet size, such factors will result in delay spread [1]. In terms of packet collision. If number of vehicles broadcast messages to the others, packet collision probability will undeniably increase significantly, such a high percentage of packet collision probability will lead to channel death, data exchange will be impacted, therefore, real available data will get lost, which finally cause unsuccessfully traffic application [2, 3]. As a communication mode, handshake process undoubtedly plays an irreplaceable role on communication delay [4] Handshake mechanism may enhance broadcast protocol, which is dedicated to the emergency message broadcast in urban road environment. Meanwhile, after connection establishment between two communication nodes, during message transmission phase, several delay species also cannot be neglected, for instance, packet delivery rate, data sending frequency[5]. Besides, random transmission delay and packet drops are inevitable in communication [6]. Such factor mentioned above will have impacts on communication delay or data qualify when DSRC is used in connected vehicle technology (CVT) application, so, for realizing traffic application, they have to be taken into consideration.. Actually, only in regularly active network should we consider data communication influence factors, what is more, if something wrong cannot be forecasted that lead to a network environment is non-continuous, what will happen when traffic applications are used that desired to be further discussed. In terms of data communication qualify under noncontinuous network environment, there are also several researches referred. Some researchers compared the impacts of traffic flow prediction between continuous and noncontinuous network, find out that compared to continuous network, non-continuous will make it difficult and inaccurate to collect traffic data for predicting traffic flow, including to traffic flow, speed, density[7]. Further, as to traffic management application, it will become inappropriate for traffic management and decision-making. In addition, signal coverage for vehicular users will be decreased [8]. For noncontinuous network, which provided worst-case condition on

Abstract—Traffic safety has been one of the most primary social problems puzzling big cities over the world. It is widely believed that connected vehicle technology will tackle this problem. To achieve that, we installed RSE (Road Side Equipment) in urban road environment to communicate with OBE (On Board Equipment) through DSRC (dedicated shortrange communication) to organize a DSRC network system. However, as a fundamental communication device, RSE may lose connections, which will have impacts on the quality of collected data and finally this might lead to failures of traffic applications. Therefore, it is necessary to validate the connection status of DSRC network environment. In this paper, based on analyzing non-continuous DSRC network system, a field test method for connected vehicle technology (CVT) application is proposed to verify whether such a network system is in good condition to ensure normal usages of traffic applications or not. Then, a case study in the field is conducted to evaluate the validities of some important parameters (i.e., latency, distance, and speed). We draw a conclusion that the network system was available for CVT application; but was desired to be improved. Keywords—Non-continuous network, Connected technology, data evaluation, Dedication short communication(DSRC)

vehicle range

I. INTRODUCTION The precondition of realizing connected vehicle system is to construct a stable and sufficient wireless network. For setting up such a reliable wireless network system, developing wireless technology has already became popular topic. Under the impetus of the development trend of wireless technology, dedicated short range communication emerges at a historic moment and has been widely used in web of things, especially connected vehicle. Since there are various kinds of application in connected vehicle, different types of data are needed to support application serving to traffic management or traffic safety, for example, hotspot location information collection and active pedestrian protection. However, a certain level data communication quality is necessary no matter which Connected Vehicle Technology (CVT) application we use. For this case, further discussion is essential about which factors will have negative effects on data communication qualify, especially in non-continuous network environment. According to existing research on impact factors of data communication qualify, it is major concentration on three

‹,(((



WK,QWHUQDWLRQDO&RQIHUHQFHRQ7UDQVSRUWDWLRQ,QIRUPDWLRQDQG6DIHW\ ,&7,6 $XJXVW%DQII&DQDGD

the interconnection gap at the same time using active protection systems (APs). Hence, worst-case guarantees were necessary for non-continuous coverage requirement. Also, in non-continuous network environment, reliability of communication link is extremely weak. The performance of DSRC will heavily discount. Even in dense and high load scenarios, where the throughput is decreases while the delay and packet loss ratio is increasing significantly [9, 10]. In this case, such new problems like data collection accuracy, signal coverage and link reliability will occur in non-continuous network environment, therefore, a more comprehensive and in-depth consideration have to be taken by us to analyze the relationship between message transmission delay time and vehicular speed. And it is because of there are data transmission delay, data qualify need to be further researched to evaluate whether data could satisfy different application needs. Obviously, data qualify needs vary from different traffic applications, hotspot location information collection require lower qualify level while traffic safety applications requires high, whose delay time needs reach as little as 200 milliseconds [6]. On one hand, some researchers evaluated data qualify impact reasons and factors in continuous network environment [4-6]. On the other hand, the others researched which processes will have seriously impacted in non-continuous network environment [7-9]. Whilst, as to different traffic applications, data qualify evaluation is relatively defective in non-continuous network environment. Therefore, data quality needs to be verified further. In this paper, we studied hotspot location information collection and traffic safety application respectively to explore efficiency of data for different traffic application under non-continuous network. Correspondingly, an evaluation method is developed for research purpose. The specific contributions of this work are given as follows: Firstly, a non-continuous DSRC network system is described, including to hotspot location information collection center, roadside equipment (RSE), onboard equipment (OBE), DSRC communication device for pedestrian. In this system, part of broadcasting messages will loss because of some unpredicted reasons. Secondly, a method aiming at traffic data collection application and traffic safety application (active pedestrian protection respectively) is developed to evaluated data qualify through verifying efficiency of message delivery latency. Thirdly, field test is designed to collect data to analyze the relationship between latency and speed or distance. The result shows that when speed is low, data qualify can satisfy the basic needs of both traffic application. In comparison to traffic collection application whose needs is not strict [6], data qualify is unavailable for traffic safety application to a certain extent.

received by the other devices, or time interval for receiving message will get longer than normal circumstance. Ultimately, performance of communication will decline. We give a description of a typical non-continuous network system.

Data Collection Server RSE

Follow too Close Warning Active Pedestrian Protection FIGURE 1 Problem definition a non-continuous DSRC network system

As figure 1 shows, in this System, a RSE has been installed alongside the road, which broadcasts map data (MAP) of the local road segment and the intersection. And a collection server included hotspot location information connecting with RSE is used to gather real-time data. We assume that vehicles are equipped with on-board equipment (OBE), and pedestrians are carried with handheld onboard equipment. Real-time speed and location information of the vehicle and pedestrian can be obtained. Through V2I communication, the real-time data, including speed and location, can be sent from vehicles and pedestrians to RSE. TMC can send strategies or additional context information to OBE, RSE, and handheld OBE. Besides, pedestrian, as a main traffic participant, their mobile phones will get the information. Assuming several vehicles carrying OBE can communicate with the other vehicles or RSE, and display information from portable DSRC device. In terms of signal transmission, on one hand, tall buildings definitely could hinder the signal broadcasting, which forms non-light-of-sight (NLOS) scenario, and result in the existence of communication delay of packet loss. On the other hand, channel congestion of communication devices will also make contribution to message transmission delay during the data transmission process. III. METHOD Fig.2 illustrates the research method adopted in this study to determine the efficiency of non-continuous network environment in supporting CVT application, which is adopted from the previous references [6]. In this section, we use the proposed adopted methodology to verify the vehicle-toinfrastructure and vehicle-to-pedestrian communication scenarios.

II. PROBLEM DEFINITION The problem considered is to define a non-continuous DSRC network system. In this system, something that cannot be predicted will certainly impact the healthy network environment. The problem is that part of message cannot be

‹,(((



WK,QWHUQDWLRQDO&RQIHUHQFHRQ7UDQVSRUWDWLRQ,QIRUPDWLRQDQG6DIHW\ ,&7,6 $XJXVW%DQII&DQDGD

Traffic data collection (CVT mobility application) Can CVT application be supported by noncontinuous DSRC network environment ?

Probelem Definition

Case Study 1 How field test could be used to evaluate noncontinuous DSRC network environment for CVT application

Case Study 2

Vehicle to infrastructure(V2I): Data logging application

Active pedestrian protection (CVT safety application) Vehicle to Pedestrian(V2P): Safety Message broadcast application

Research Method for Two Cases

B. Case Study 2 description: Active Pedestrian Protection application Being similar to V2V communication, V2P communication also enabled safety application data collected from nearby vehicle. Including vehicle location, direction and speed in determination of any potential risk of vehicle collision to achieve the aim to protect pedestrian. We also did field test that will be detailed described in section 4.

Is noncontinuous network environment suitable for traffic management application

Is noncontinuous network environment suitable for active pedestrian protection application

Start Appplication

Get time stamp and GPS coordination of pedestrian through DSRC

Research Conclusion

FIGURE 2 Evaluation Method of non-continuous network environment for CVT Applications-Research.

Calculate distance between OBEs for time stamp n and n-1

Field tests experiments have been done to answer the research questions about evaluating the data qualify whether it is suitable for CVT application or not. To evaluate the performance of non-continuous network environment, case study 1 is conducted for a CVT supported hotspot location information collection. To evaluate the performance of noncontinuous network environment for a CVT safety application, a forward collision warning application for V2P is conducted in case study 2. The data flow between roadside equipment (RSE) – on board equipment (OBE) and OBE-OBE has been logged. Two CVT applications are complementary as it does multiple usages in an integrated scenario. The developed application is designed to integrate such 2 application in an interface to reach a comprehensive purpose. A detailed description of case studies is presented in the following sections.

Distance at n< Distance at n-1

Yes Yes Distance>threshold?

A. Case Study 1 description: Hotspot location information collection In order to strategically understand and test the idea of supplement DSRC with non-continuous network environment for hotspot location information collection application, we did tests in the non-continuous network environment shown in section. To evaluate the efficiency of V2I data communication using DSRC for a hotspot location information collection application, a server (Fig .1) was constructed to store traffic data from RSE, in which the data contains GPS coordinates, time stamp, and vehicle ID number. Specifically, we do a statistic calculation to calculate received packet quantities by a server in a certain amount of time, and then average time for receiving each packet will be got. For received packets, we will compare with the number of sending from RSE to calculate the packet delivery rate and loss ratio, and for the delay time, the value will be equal to average receiving time minus average sending time.

‹,(((

No

No

Continue to drive

Send warning alert from mobile phone

End Application FIGURE 3 Active Pedestrian Protection Application Procedure

As shown in Fig.3, portable DSRC communication device for pedestrian broadcast GPS location to nearby area, this message could be received by the other vehicles inside the communication range. Combining with GPS location and time stamp itself, distance at time stamp n and n-1 between vehicle and pedestrian will be calculated. We just use this value to judge whether it declines or not. If it rises, that means vehicle is moving away from pedestrian, in that condition which will be no risk. However, if it decreases, which presents vehicle is on an approach to the pedestrian. In the approaching condition, there will be a risk for the pedestrian and it needs to be further judged. Thus, we go further for pedestrian protection application. Distance being equal to certain threshold (e.g., 20



WK,QWHUQDWLRQDO&RQIHUHQFHRQ7UDQVSRUWDWLRQ,QIRUPDWLRQDQG6DIHW\ ,&7,6 $XJXVW%DQII&DQDGD

meters) is chosen and use as the standard for sending pedestrian close warning alert. If the distance is no longer more than certain threshold (e.g., 20 meters), the warning alert will be sent timely. Time between when the distance reaches threshold and alert sent will be recorded as the delay time. IV. FIELD TEST SETUP For field test setup, a segment in Youyi Street near Wuhan University of Technology was chosen. As shown in Fig.4 below, a RSE is installed on a platform bridge crossing Youyi Street connecting Yujiatou. Except that, a server is linked to RSE through an appropriative wired network for collecting traffic data. As to case 1 study, i.e., hotspot location information collection, we drove a car with a constant speed from the start point of yellow array 1 to terminal point of yellow array 2. During this process, RSE received basic safety message (BSM) from vehicle, and the data were stored inside the server. During the tests, we changed vehicle speed and repeated the process for several times. As to case 2 study, i.e., active pedestrian protection application, one person carrying a portable DSRC communication device gone from start point of yellow array 3 on the platform bridge to terminal point of red array 2. At the same time, another person drove a car from the start point of yellow array 1. When car arrived at the start point of red array 1 (20m line), it will recorded timestamps until pedestrian warning was sent. With no exception, we changed vehicle speed and repeated the process for several times.

Figure 5 Latency Proportion

Figure 5 described proportional distributions of all recorded latency data. Obviously, what is noteworthy is that in most case (70%), the network was worthy of trusts, where latency time approached to be 0s. However, the network is desired to be optimized. Because the longest latency is almost 0.2s which takes up as much high as approximately 10% of the total. We can image, if a vehicle is running at 20 m/s, even 30 m/s (which was general on freeway), the vehicle will move 6m. In this case, it must bring some traffic safety problems. The rest of one fifth was just 0.1s, which was acceptable.

Figure 6 Relationship between Latency and Speed

Fig.6 considers the distribution of latency when the speed is in the range of 0-2.5m/s. The valid data are mainly concentrated on 1 - 1.5m/s, which is equal to pedestrian walking speed. In this speed range, latency are almost evenlydistributed. However, when the speed is approximately 0.5m/s, the data is sparse. This is because during field test process, we choose to speed up to normal range immediately without covering the value of 0.5m/s. Beside, when the speed is more than 1.5m/s, the latency data appeared to be heterogeneous, which presents the network communication was unstable. To sum up, the network environment is available for pedestrian protection application. But whether is suitable for vehicle to vehicle security application, it needs to be further discussed.

FIGURE 4 Field Test Setup Used in Non-continuous Network Environment

V. RESULT AND DISCUSSION Figure 5, 6 and 7 show the field test results, including figures of latency proportion, relationship between latency and speed or distance. In these three figure, speed, GPS and time information are got from field test (about 25000 pieces of data).

‹,(((



WK,QWHUQDWLRQDO&RQIHUHQFHRQ7UDQVSRUWDWLRQ,QIRUPDWLRQDQG6DIHW\ ,&7,6 $XJXVW%DQII&DQDGD

the network system was available for CVT application; but was desired to be improved. In the future, another fundamental communicaiton way will be validated by the proposed methodology, which is V2V communication. This will enable safety applications using data collected from nearby connected vehicles, such as vehicle speed, vehicle location, and heading direction. Through combination of different vehicle locations, the distance could be calculated. In a similar way, relative velocity and time to collision (TTC) will be got. At that time, warning alert will send to driver form mobile phone connected with OBE to enhance the traffic safety. ACKNOWLEDGYMENT This work is supported by Traffic Congestion Situation Awareness Oriented Signalized Intersections Regulation Method Based on Connected Traffic (41161381).

Figure 7 Relationship between Latency and Distance

REFERENCES

Figure7 dipicts the relationship between latency and distance. According to the figure, when the distance is less than 150m, the latency is acturally evenly distributed, and more effective distance data converge from 90m to 140m. neverthelessly, the latency is not continuous in about 80-140 this is because we have a brief stay at 80m, 100m ,120m 140m respecfically so that we get more data than the other location. It appears to be more dense ultimately. We can conclude that, at that time, the network environment is in good condition. But when the distance reachs 150m or more, the latency of received messages becomes less, which illustrates that distance has negative effects on communication quality.

[1]

Rami Sabouni M. Evaluation of DSRC For V2V communications., pp.1133, 2011. [2] E. E C, Zhang S, Liu E. Estimation of collision probability in a saturated vehicular Ad-Hoc networks. 2015 Fourth International Conference on Future Generation Communication Technology (FGCT)., pp. 1-7, Jul 2015. [3] Kloiber B, Härri J, Strang T, Sand S, García CR. Random Transmit Power Control for DSRC and its Application to Cooperative Safety. IEEE Transactions on Dependable and Secure Computing., vol.10, no.1, pp.18-31, 2016 [4] Wanting Zhu DG, Chuan Heng Foh, Weicheng Zhao, Hongke Zhang. A Collision Avoidance Mechanism for Emergency Message Broadcast in Urban VANET. Vehicular Technology Conference., pp.1-5, May 2015. [5] Pan N, Huang L. Sending and receiving circuit design for ground-based coherent high-frequency radar., 2016 IEEE International Conference on Aircraft Utility Systems., pp. 933-937, Oct 2016. [6] Kakan Chandra Dey a, Anjan Rayamajhi , Mashrur Chowdhury , Parth Bhavsar, James Martin. Vehicle-to-vehicle (V2V) and vehicle-toinfrastructure (V2I) communication in a heterogeneous wireless network –Performance evaluation. Transportation Research Part C: Emerging Technologies., vol.68, pp.169-184, 2016. [7] Jeon S, Hong B. Monte Carlo simulation-based traffic speed forecasting using historical big data. Future Generation Computer Systems., vol. 65, pp. 182-195, 2016. [8] R. M. Kadwe PK, Yogesh Bhute. Optimization Techniques For High Performance Content. International Journal Of Engineering And Computer Science., vol. 4, no. 1, pp. 9814-9818, 2015. [9] Tong Z, Lu H, Haenggi M, Poellabauer C. A Stochastic Geometry Approach to the Modeling of DSRC for Vehicular Safety Communication. IEEE Transactions on Intelligent Transportation Systems., vol.17, no. 5, pp. 1448-1458, 2016. [10] Eichler S. Performance Evaluation of the IEEE 802.11p WAVE Communication Standard. Vehicular Technology Conference., pp. 21992203, 2007

VI. CONCLUSION In this paper, based on analyzing non-continuous DSRC network system, a field test method for connected vehicle technology (CVT) application is proposed to verify whether such a network system is in good condition to ensure normal usages of traffic applications or not. Then, a case study in the field is conducted to evaluate the validities of some important parameters (i.e., latency, distance, and speed). We find it that both pedestrian speed and communication distance will have impacts on latency. Specifically, among average pedestrian speed, different latency data can be collected sufficiently, which explains that communication connection is subsistent. However, with speed goes up (1.5m/s), it becomes increasing difficult to collect effective data, network environment turns to be non-continuous. Similarly, when communication distance exceeds to certain condition (over 150m), non-continuous circumstance comes to be extremely possible. Therefore, we draw a conclusion that

‹,(((