JOURNAL OF TELECOMMUNICATIONS, VOLUME 11, ISSUE 1, OCTOBER 2011 1

Pedestrian-to-Vehicle Communication System for Pedestrian Safety Using Both Cellular Wireless Network and WLAN Chika Sugimoto and Yasuhisa Nakamura Abstract— We propose a pedestrian-to-vehicle communication system without roadside equipments using both 3G cellular wireless network and wireless LAN for improving pedestrian safety. One of the effective measures against pedestrian accidents is to make each of pedestrians and drivers find the others and recognize the risk from out of sight and with time to spare for avoidance of accidents. We focused on intersections as a traffic scene to be covered by the system. At some blind intersections where the propagation property of NLOS roads is worse than that of straight roads, the potential for WLAN to serve as P2P communication method was verified. Information was exchanged via 3G cellular wireless networks in a wide area, and then could be directly exchanged between a pedestrian and a vehicle via WLAN 20m and 100m away from an intersection to different directions in about 20ms. The system could give an alarm to pedestrians and vehicles with collision risk. Sensory tests showed that the system seems to be effective for making drivers recognize pedestrians in NLOS areas and heightening safety. Index Terms—Cellular wireless networks, P2P communication, Safety, Wireless LAN

——————————  ——————————

1 INTRODUCTION

W

ith the widespread of the Internet and information devices, various information systems have been built for supporting social activities. In transportation field, intelligent transportation systems (ITS) have been promoted with the aim of realizing a safer and more comfortable road traffic environment [1]. This is because traffic safety measures have been required continuously in Japan for a high incidence of traffic accidents. Information on vehicles and roads can be collected and utilized for traffic safety by using road-to-vehicle communication and inter-vehicle communication. The roadto-vehicle communication and the inter-vehicle communication that use microwave and millimeter wave are just beginning to be introduced into ITS for safety. Actually, a technological study of wireless communication standard was carried out and the formulation of the communication specification of 5.8GHz band based on DSRC has been advanced. The research and development of the inter-vehicle millimeter wave communication by cooperation with the in-vehicle radar used as an autonomous accident prevention system has been undertaken [2]. These efforts are certainly effective as measures against car collisions. However, they are not enough to prevent pedestrian accidents. The ratio of pedestrian accidents in Japan is high compared with other advanced countries. 34.9% of road accident fatalities in 2009 are occupied in walking [3]. Under such circumstances pedestrian-tovehicle communication has been needed to get pedestrian ————————————————

information by making an information network of pedestrians, vehicles, and roads. The principal reasons of pedestrian-to-vehicle accidents are errors of recognition and judgment. Therefore, the effective measures against accidents are to make each of pedestrians and drivers find the others and recognize the risk correctly. A technique for recognizing pedestrians who can be detected with cameras installed in a vehicle has been put to practical use partly. However, there is not a technique for recognizing each other from out of sight without road infrastructure which is built in limited areas. The aim of this research is to develop the system for contributing to the prevention of pedestrian accidents using pedestrian-to-vehicle communication. Ultra wideband (UWB) radio communication is one of the methods expected for P2P networking due to its low power consumption and high data rates. However, UWB networks tend to interfere to other coexisting networks because the band of UWB radio overlaps the spectrum of existing narrowband and wideband systems. Therefore, there are a lot of problems to be solved toward the practical use. DSRC (Dedicated Short Range Communications) at 5.8GHz has been also studied for pedestrian-tovehicle communication [4]. DSRC is a promising method, but it is necessary to set up the roadside equipments at intersections in order to send data within the reasonable range including non-line-of-sight (NLOS). We use both cellular wireless network and wireless LAN (WLAN) in order to communicate in a wide area and with little delay time in a local area. In Japan cellular phones and car navigation systems are widely used and the number grew to more than 120

C. Sugimoto is with Division of Physics, Electrical and Computer Engineering, Graduate School of Engineering, Yokohama National University, Japan. Y. Nakamura is with Ubiquitous Services Department, NTT DoCoMo, Inc., Tokyo, Japan. © 2011 JOT www.journaloftelecommunications.co.uk

2

million and 44 million units, respectively. The penetration rate of 3G cellular phone, which is obligated to be equipped with GPS since 2007, is more than 97% in 2009. We present a prototype system developed using the network function of cellular phones and car navigation systems. The system obtains the location information from each of pedestrians and vehicles, sends the response based on the risk estimation, and makes the target pair exchange each other’s information directly in the case of high collision risk. From the analysis results of accident situations, it is certain that pedestrian accidents often occur around intersections at which pedestrians are out of sight of drivers. We focused on intersections as a traffic scene to be covered by the system. The NLOS propagation property and communication characteristics of IEEE802.11 WLAN were specifically examined at intersections. Moreover, the system functions were verified at intersections. The system proposed and the results of the verification are described in Section 3. We present conclusions in Section 4.

changes significantly by the shape of the corner and there are some areas of road where it is difficult for radio waves to penetrate sufficiently. Fig.2 is an example. In this way, the propagation property in some models can be analyzed by the simulation. However, it lacks versatility because that in real road environments changes greatly depending on the surroundings such as running and parked vehicles, buildings, trees, and road traffic signs around the road. Therefore, it was necessary to survey it in the field. concrete wall

5m

Intersection 5m center

transmitter

Fig.1 Environmental condition of simulation

2 PEDESTRIAN-TO-VEHICLE COMMUNICATION OVER WIRELESS LAN

2.1 Radio Wave Propagation Property The radio wave propagation property of WLAN depends on the environment and the situation highly because of the fading variation and the shadowing effect. For this reason, it has been studied under certain conditions of frequency band, antenna, and path in each situation [5][8]. To investigate the property in the case of direct communication between a pedestrian and a vehicle without a radio repeater, a simulation was run by the ray tracing method. As a typical road environment in a suburb, a model of intersections was selected. The roads of 5.0m in width run at right angles to one another with the concrete walls on both sides (Fig.1). The road distance from the center of an intersection to the antenna was varied within 100m. The height of a transmitting antenna and a receiving antenna was 1.0m and 1.5m, respectively. The result showed that the distribution of the propagation loss on a road in NLOS direction is affected by the distance from a transmitting antenna to a corner. It could be seen that the degree of reflection and diffraction

Y Y[m] [m ]

8

-55

6

-60

4

-65

2

-70

0

-75

-2

-80

-4

-85

-6

-90

-8 -10 -25

-95 -20

-15

-10

-5

0

5

10

X [m ]

X [m]

(a) 受信電力の距離特性 [dB m ]

10

YY [m] [m ]

Via a cellular phone network which covers a wide area the terminals can widely send and receive information through a center server. However, it has at least 400ms of communication delay. It is necessary to introduce P2P communication method which has little delay in conjunction with cellular network communication so as to exchange information between pedestrians and vehicles with enough time and distance to prevent a collision. WLAN (IEEE802.11b) is considered to be one of the best P2P communication methods between pedestrians and vehicles in terms of baud rate, communication range, power consumption, size, cost, and possibility to be installed in a cellular phone and a car navigation system at this time. Therefore, the characteristics of IEEE802.11b were evaluated.

受信電力の距離特性 [dB m ]

10

8

-55

6

-60

4

-65

2

-70

0

-75

-2

-80

-4

-85

-6

-90

-8 -10 -25

-95 -20

-15

-10

-5

0

5

10

X [m ]

X [m]

(b) Fig.2 Distance property of received power (a) curved corner, (b)square corner

2.2 Measurement of WLAN Communication Characteristics at NLOS Intersections To verify the over-the-horizon propagation property, the communication characteristics were examined at intersections of 5.0m in width in a residential district. WLANequipped PDA DELL Axim X51 and WLAN access point buffalo WHR-HP-G54 were used. The specifications of WLAN devices are shown in Table 1. First, by using WLAN analyzer AirMagnet, the signal distribution and S/N distribution in the right turn direction were measured when the access point was put on the points of 15, 20, 30, 50, 75, and 100m from the center of the intersection. As a result, more than a certain level of signal intensity was observed over several tens of meters from the

3

corner for reflection and diffraction of radio wave. Next, on condition that the data of 1220 bytes was transmitted from a client of 25m away from the center, the time for connecting and the baud rate were measured at the receiving points of 20, 50, 75, and 100m away from the center three times, respectively (Fig.3). One of the results is shown in Table 2 and Fig.4. The time for connecting was 2~4s. The baud rate varied for retransmission occurred when cars passed through the intersection in measuring. However, the result showed that it was possible to communicate from the distance of 75m away from the center of the intersection. Table 1 Specifications of Wireless LAN devices

and 20m from the center (Fig.5). The receiving side moved at 20 km/h from the points of 50, 75, 100, and 125m to the center, or got still at those points. As a result, it took 6~9s to send data when the access point was not moved and the communication-enabled distance was up to 75m away from the corner. On the other hand, it took 6~8s to send data when the access point was moved and accessed from within 100m and it was possible to communicate at the point of more than 75m away from the corner as shown in Fig.6. It means that these communication characteristics would not get worse when an access point was moved. Therefore, WLAN is believed to be able to be used as a method of direct communication between a pedestrian and a vehicle.

Fig.5 Positional relation between a pedestrian and a vehicle at an intersection

Fig.3 Measurement environment at an intersection

5

5

4

4

3

3

2

2

1

Baud rate Time for connecting

■

Fg.6 Communication-enabled distance at an intersection T im e [s]

B aud rate[M bps]

Table 2 Measurement result of Wireless LAN

1

3 PROTOTYPE OF PEDESTRIAN-TO-VEHICLE COMMUNICATION SYSTEM FOR PEDESTRIAN SAFETY

▲

0 0

20

40 ] D istance[m

0 60

80

Fig.4 Measurement result of Wireless LAN

2.3 Mobile Measurement of Communication Characteristics To examine the influence on the communication characteristics when an access point moves at high speed, the time for connecting, the transmission rate, and the communication-enabled distance were measured with the access point in a car running and parked by the method similar to 2.2. At a blind intersection of 5.0m in width, the transmitting side was located at both near and far sides of 10

3.1 Basic Configuration We designed and developed a pedestrian-to-vehicle communication experiment system for pedestrian safety which obtains the location information of each terminal and alerts drivers and pedestrians to coming risks. Fig.7 is the outline of the system configuration. The system is composed of cellular phones, car navigation systems, and a server. Notebook PCs were used as substitutes for car navigation systems. A cellular phone and a car navigation system have modules of GPS, FOMA, and WLAN. FOMA is a 3G mobile communication system. The location and time information of GPS is sent to the server with each identification (ID) number via FOMA (Fig.8, Fig.9). Fig.10

4

shows the information flow of a car navigation system. The server manages their information obtained and judges the collision risk between a pedestrian and a vehicle from the location, the speed, and the direction of movement. When a pedestrian enters within the area around a vehicle where caution is needed, the server informs the pedestrian and the vehicle of each other’s location information. The position is displayed on the digital map of the car navigation system and on the screen window of the cellular phone. At the same time, the server issues an instruction of direct communication to the terminals which have been judged as the highest risk pair (Table 3). The cellular phone and the car navigation system begin to communicate directly via WLAN by receiving the instruction and exchange each other’s location information and so on. It alerts the driver with lighting on the display. An audio warning can be also given. The server records the location, the time, and the collision risk level with ID number in order to study high-risk intersections, pedestrians, and vehicles that are likely to have a possibility of an accident.

Direct com.

Wide-area com. Standby (Timer Counting)

Standby

To server

Timeout

Data Transmission Data Reception From server

Timeout

Dialogue Update

From mobile phone

Data Reception Data Transmission

To mobile phone

Dialogue Update

reference

reference Data store

GPS data reception

Map Update

Fig.10 Information flow of car navigation system Table 3 Example of response message between server and terminal (a,b, and c represent numeric characters) Enter:ID=09011112222&N=3444.4170&E=13534.3582&D=359&V=13.1234 Highrisk:ID=09011112222&N=3444.4180&E=13534.3583&D=359& V=18.1234 Direct:ID=09011112222&N=3444.4180&E=13534.3583&IP=192.168.11.2& D=359&V=18.1234

3.2 Experiments with the Developed System

Fig.7 Outline of pedestrian-to-vehicle communication system

Fig.8 Software configuration of terminals

3.2.1 Communication Delay The function of location notification using two communication methods was verified at an intersection surrounded with buildings by using the developed system. Fig.11 shows the shape of the intersection and S/N distribution of WLAN when an access point was located on about 100m away from the center of the intersection. A vehicle with a car navigation system moved from the point of 100m to the intersection at about 5 km/h and a pedestrian with a cellular phone stood still at the points of 10m and 20m on the near side of left turn direction. As a result, it was possible to communicate the location information directly via WLAN from the point of more than 90m and to display the latest position by getting the information not only via FOMA but also via WLAN. The result indicates that the communication delay time of WLAN is less than that of FOMA. Fig.12 shows the delay time measured in the case of setting the range of a risk area to 75m and 25m respectively. The delay time of FOMA was about 400~900ms and occasionally more than 1000ms. On the other hand, that of WLAN was about 20ms regardless of the distance between a pedestrian and a vehicle.

Access point Car direction × Pedestrian

14.10m

Fig.9 Information flow of pedestrian-to-vehicle communication system

5

Fig.11 WLAN S/N distribution at an intersection

attentions of pedestrians and drivers based on location information via FOMA and inform them of more right location information via WLAN in approaching an intersection. However, both methods are needed because a good communication environment is not necessarily secured at all of the intersections continuously.

3.3 Information processing of server

(a)

(b) Fig.12 Delay time of wide-area communication and P2P communication (a) risk area of 75m (b) risk area of 25m

3.2.2 Communication-enabled Distance In the same way as 4.2.1, communication-enabled distance was measured at the different positions of a pedestrian. A vehicle with a car navigation system moved from the point of 100m to the center of the intersection at about 5 km/h and a pedestrian with a cellular phone stood at the points of 5m, 10m, and every meter of 20~25m. As a result, it was possible to communicate from the point of more than 90m away from the intersection when the pedestrian stood within 20m of the intersection. But, the communication-enabled distance took a sudden drop from over 20m as shown in Fig.13.

3.3.1 Data management The software configuration of the server is shown in Fig.14. In the database (DB) of the server, individual information, intersection information, and road information have been stored in advance. The server also manages the information send by pedestrians and vehicles with each ID. The flow of information processing is shown in Fig.15. First, the following points are estimated based on the information.  Is the terminal a vehicle, a pedestrian, or a cycle?  Is the terminal in an intersection area or not?  Is the time of GPS data available or not? The risk estimation is performed if the time of GPS data is available. Then, the server stores the data of location, time and date, direction of entering and leaving intersections, traffic situation, and collision risk level with ID in DB of history information in order to study each track and highrisk spots of traffic accidents. The estimation results are sent to the pedestrian and the vehicle. DB History Information

Individual Information

Intersection Information

Road Information

Script (PHP) HTTPd (Web Server)

com m unication-enabled distance(m )

100 90

Pedestrian/Vehicle Pedestrian/Vehicle Information Information Fig.14 Software configuration of server

80 70 60 50 40 30 20 10 0 0

5

10

15

20

25

30

pedestrian-to-intersection distance(m )

Fig.13 Communication-enabled distance by position of pedestrian

3.2.3. Consideration It takes about 9.0s for a car to run from the point of 100m to the intersection at 40km/h. And, it takes about 15.4s for a pedestrian to go 20m at 1.3m/s which is the average speed of walking. Therefore, it is considered to be enough if communication via WLAN is started when a vehicle and a pedestrian enter within 100m and 20m of an intersection respectively. As a result, it can be said that it is possible to call for

6

4 EVALUATION OF SYSTEM Data reception

4.1 Simulations

Time estimation

Intersection information

NG

Risk area estimation Repeat Collision risk estimation

Direct communication estimation

Data transmission

History information

Fig.15 Flow of information processing

3.3.2 Risk estimation algorithm The risk is estimated between vehicle and pedestrian/cycle. The risk estimation is divided into 4 phases. Firstly, whether the risk area of a vehicle overlaps with that around a pedestrian or not is judged. Each risk area is set around one as shown in Fig.16. The size of a vehicle’s risk area changes depending on the location and velocity of the vehicle. When a vehicle goes into an intersection area, the risk area of the vehicle changes the size to cover the NLOS area. Secondly, in the case of overlapping, the collision risk between the pedestrian and the vehicle is judged from the estimated time to collision and distance of closest approach based on the location, the speed, the direction of movement, and the road information. In the areas of intersections, pedestrians and vehicles sometimes turn right or left and do not move in a regular way. Therefore, the time to reach the intersection is reflected to the method of estimation in the areas, which is different from that on straight roads. Thirdly, the level of risk is estimated by judging whether the collision risk exceeds the threshold or not. Fourth, one of each pair for connecting with WLAN and communicating directly is judged from the collision risk and the estimated time to collision.

4.1.1 Evaluation of traffic situation A case of traffic accidents which actually took place was simulated. At T intersection without a traffic light a vehicle going straight hit a pedestrian walking across the street from the right side at night (Fig.17). The driver couldn’t notice the pedestrian A because it was dark and he had been paying attention to another pedestrian B who was walking ahead of the pedestrian A. Fig.18 shows the change of the collision risk against the distance between the vehicle and the pedestrian A. The graph draws a comparison with the case that an alarm was given 6 seconds before the estimated collision and the driver could find the pedestrian approaching after attention and recognition time of 3 seconds and brake after reaction time of a second. It indicates that the risk drops to a lower level and the vehicle could stop 5m back from the pedestrian. PedestrianB Vehicle PedestrianA

Fig. 17 Traffic situation risk (w ithout w arning) risk (w ith w arning) pedestrain-to-vehicle distance (w ithout w arning) pedestrian-to-vehicle distandce (w ith w arning)

110

100

100 90

80

80 70

60

60 40

50 40

D istance（ｍ）

Individual information

D egree of risk

Intersection area estimation

20

30 20

0 0

1

2

3

4 5 6 Tim e（ ｓ）

7

8

9

Fig. 18 Collision risk and distance between pedestrian and vehicle at T intersection

Fig.16 Risk areas of a vehicle and a pedestrian

4.1.2 Consideration It takes about 6 seconds from the point of 100m away to the intersection when a vehicle runs at 60km/h. Maximum vehicle deceleration on dry pavement is generally about 8m/s2. It can make a stop within 3.1 seconds including reaction time in slowing down at 8m/s2. And so, it can be said to be possible for the driver to brake, slow down the vehicle, and avoid the collision by using the developed system because the system can make the driver aware of the pedestrian more

7

Table 4 Questionnaire for driver

than 6 seconds before the estimated collision.

Question

4.2 Experiments on a road A sensory test was conducted to verify the effectiveness of information support in heightening safety by the developed system. 4.2.1 Experimentation method An experiment was performed at a blind intersection. The experiment image is shown in Fig.19. Test subjects for driver were two adult males who have adequate driving techniques. Test subjects for pedestrian were two males. The subjects were explained the function of the system and were instructed to go to the set way. The experimenters observed the subjects’ behavior and ensured the safety. The traveling directions are shown in Fig.20. When a driver went straight/ turned left/ turned right, a pedestrian went across the road from the left/right side of the vehicle. They moved at just the right moment for running into each other, and pedestrians walked at a slow/quick pace in each pattern. GPS WBT-201 (Wintec), which was set to get GPS data every 200ms, was used for location measurement. At the beginning of the experiment, the subjects made a trial run of the way without using the system. The driver could not notice the pedestrian until the vehicle approached the intersection because the pedestrian was out of sight of the driver. Next, the best timing for advice to each driver was measured in going straight by changing the time from 6s of TTC to 15s with the system. The time was used as the set value for giving advice, respectively. The position, velocity, direction of movement, and on/off of stepping the brake and the accelerator were recorded during the experiment. After the experiment, the subjects answered the questionnaire which consisted of 11 questions (Table 4) on a scale of 1 to 5 and made a comment. (The questionnaire for pedestrians was partly modified.) GPS W alkerAhead!

Car Navigation system

Cellular Phone

1

Did the system’s sound call your attention?

2

Did you look at the screen of the system？

3

Did you recognize the movement and color of pedestrian icon on the screen?

4

Could you recognize pedestrians from out of sight?

5

Where did you pay attention to when information support was given?

6

How was the timing of information provision?

7

Was it easy to recognize the image and the sound?

8

Did you depend on the system support too much?

9

Was the safety of driving improved by information support?

10

Did information support prevent you from recognizing a pedestrian?

11

Did information support bother you?

4.2.2 Result and consideration The system could make the drivers aware of the pedestrians with enough time to prepare them for the potential risk from out of the sight. The best timing for advice to each driver was preceding 12s of TTC when the vehicles went straight at an intersection. On the other hand, when the vehicles turned left or right at an intersection, it got much shorter and was the time when the driver turned on the winker or the vehicle moved quite near to the intersection. The reason is because the drivers brake and slow down the vehicle for themselves in turning at an intersection. There was no difference in the time between the drivers. It is considered to be because the subjects were close in age and were similar in driving ability. Fig.21 shows the corresponding movement locus of pedestrian1’ and vehicle1 measured by the system. The driver had stepped on the accelerator at a constant pace during driving. However, after the information support was given, the driver stopped to step on the accelerator, slowed down the vehicle gradually, and stepped on the brake some times by the vehicle reached to the intersection. As the driver could make sure the pedestrian was looking at the vehicle and was walking slowly, the driver went through the intersection at reduced speed. Fig.22 shows the collision risk. The collision risk between pedestrian and vehicle was calculated after the position of vehicle was corrected on the left side of the street. The degree of risk went up shapely at once, but it went down after information support and went up again with a gentle slope though the vehicle was getting quite close to the pedestrian.

Fig.19 Experimental situation

③ ①’”

Pedestrian1

Vehicle1

Pedestrian3

③’ ①’’

Pedestrian1’ Vehicle

①

② ①’ ②’’ ②’

Building Pedestrian2

Fig.20 Outline of experimentation

Fig.21 Corresponding movement locus of pedestrian and vehicle at the intersection (Pedestrian1’ is crossing from nearside left. Vehicle1 is going straight.)

8

driver find the other from out of sight using both 3G wireless networks and WLAN. The sensory test showed

that the system is effective for making drivers recognize pedestrians in situation where they are difficult to notice and heightening the safety. It is suitable for use in residential areas such as school zone. We will be ready for large-scale implementation in residential areas.

ACKNOWLEDGMENT Fig.22 Collision risk between pedestrian and vehicle

To the questions shown in Table 4, the drivers gave 1th, 4th, 8th, 9th, and 10th high marks and gave 3th and 7th low marks. The marks of 2th and 11th were different between two drivers. On the other hand, the pedestrians gave 1th, 4th, 7th, and 9th high marks and gave 6th and 11th middle marks. The result of the questionnaire showed the following. 1. The drivers could pay more attention to the pedestrians on the road by the information support of the developed system, in particular by sound. 2. Even though the drivers glanced at the icons on the screen, they couldn’t recognize their movement and color. They trended to pay more attention to their way and the blind areas on the road. 3. As a means of advice to drivers, the voice involving a word ‘pedestrian’ is better than the sound. 4. The best timing for advice to pedestrians is later than to drivers because pedestrians can stop soon. 5. The pedestrians needed the information on the direction to which should be paid attention in advice, like ‘Right caution’. 6. The possibility to depend on the system too much wasn’t seen during this experiment, if any. 7. There is the possibility that the system bothers drivers depending on circumstances. By using the system, the drivers could slow down and pay more attention to the pedestrians on the road by recognizing the pedestrians from out of the sight in advance. As a result, it can be said that the effectiveness of the system for improving the pedestrian safety was verified. It is considered that the optimal timing for information support differs between pedestrians and drivers and depends largely on the traveling direction.

5

CONCLUSION

A pedestrian-to-vehicle communication system which makes the information network between pedestrians and vehicles without additional infrastructure was proposed. At some blind intersections where the propagation property of NLOS roads is worse than that of straight roads, the potential for WLAN to serve as P2P communication method was verified. Consequently, it was possible to exchange information between a cellular phone and a car navigation system and make each of a pedestrian and a

This research was supported by CREST, JST.

REFERENCES [1] H. Nakamura, K. Osaku, T. Goto, R. Sagawa, T. Hashimoto, “Advanced wireless technology and VMC concept for mobile image services”, Proc. of the 24th annual conference of the robotics society of Japan, 3M21, Sep.2006. [2]

[3] [4] [5]

[6] [7] [8]

Masayuki Fujise, “ITS communication system by millimeter wave making a car safer and more comfortable-”, Journal of electronics, information and communication engineers, vol.87, no.9, pp.744-749, 2004. http://www.npa.go.jp/toukei/koutuu48/H21mistake.pdf, National Police Agency, 2004. [Sept 28, 2011] http://www.oki.com/en/press/2009/01/z08113e.html, Oki Electric Industry Co., Ltd., 2009. [Sept 28, 2011] J.R. Hampton, N.M. Merheb, W.L. Lain, D.E. Paunil, R.M. Shuford, W.T. Kasch, “Urban propagation measurements for ground based communication in the military UHF band”, IEEE Trans. Antennas Propaga., vol.54, no.2, pp.644-653, 2006. H. Masui, M. Ishii, K. Sagawa, H. Shimizu, T. Kobayashi, M. Akaike, “Microwave path-loss characteristics in urban LOS and NLOS environments”, in Proc. Vehicular Technology Conf., pp.395-398, 2001. A. Tang, J. Sun, K. Gong, “Mobile propagation loss with low base station antenna for NLOS street microcells in urban area”, in Proc. Vehicular Technology Conf., pp.333-336, 2001. V. Erceg, A.J. Rustaco Jr, R.S. Roman, “Diffraction around corners and its effects on the microcell coverage area in urban and suburban environments at 900MHz, 2GHz, and 6GHz”, IEEE Trans. Vehicular Technology, vol.43, pp.762-766, 1994.

C. Sugimoto received the B.S. degree in Engineering, and the M.S. and the Ph.D. degrees in Environment from The University of Tokyo, Japan, respectively. During 2006-2010, she was an assistant professor of Graduate school of Frontier Sciences, The University of Tokyo. She is now an associate professor of Division of Physics, Electrical and Computer Engineering, National Yokohama University. Her research interests include communications for Body Area Network, wireless communications in ITS, and Medical ICT Y. Nakamura received the B.S. degree in Engineering and the Ph.D. degree from The University of Tokyo, Japan. He is now a department director of NTT dokomo. His research interest is in mobile communications.