Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Experimental Analysis of DSRC

for Radio Signaling at Grade Crossings


Junsung Choi Vuk Marojevic Christopher R. Anderson Carl B. Dietrich
CCS Graduate school of Green Electrical and Computer Electrical Engineering Bradley Department of Electrical
Transportation Engineering Department Department and Computer Engineering
KAIST Mississippi State University US Naval Academy Virginia Tech
Daejeon, Republic of Korea Mississippi State, MS, USA Annapolis, MD, USA Blacksburg, VA, USA
choijs89@kaist.ac.kr vuk.marojevic@msstate.edu canderso@usna.edu cdietric@vt.edu

Abstract—Despite the breakthroughs of wireless technology, functionality in mind, we evaluate the feasibility of DSRC as a
only a few systems have been proposed for improving the safety protocol for protecting the safety of users near grade crossings.
at railroad crossings. We propose using and adapting the Communications systems deployed on locomotives,
Dedicated Short-Range Communications (DSRC) protocol for
railroad crossing protection to improve the safety of both trains
vehicles, and crossings have different characteristics than
and vehicles. This paper analyzes the radio frequency (RF) systems for typical V2V or V2I environments. The visibilities
propagation channel and the DSRC system performance based of train operators and vehicle drivers are typically narrower
on measurements at railroad crossings on a test track in wide- due to the environment surrounding the crossing. Also, the
open spaces and artificial shadowing environments. These time for a train to come to a full stop differs significantly from
environments assimilate typical rural and urban settings. Our that of road vehicles because of important differences in
results show that the channel around railroad crossings has an
approximately 3 to 5 dB lower Rician K factor and 2 higher path
speeds, sizes, and weights. In nearly all cases, it is not feasible
loss exponent when compared to typical Vehicle-to-Vehicle or for a locomotive or train to stop quickly enough to avoid a
Vehicle-to-Infrastructure environments; the RMS delay spread collision with an unexpected object on the railroad; sending a
in the shadowing environment is similar to that of a tunnel or warning message to vehicle drivers to make a complete stop,
non-line of sight highway scenario. For the DSRC performance on the other hand, is feasible. The relative speed differences
evaluation we use the packet error rate of the warning messages between train-to-vehicle and vehicle-to-vehicle can by itself
transmitted by the approaching train and received by cars near
the railroad crossing. We find that warning messages are reliably
be the reason for differences in the communication system
received before the minimum notification distance in a wide-open performance. Therefore, using a communication system
space regardless of the train speed and in an artificial shadowing designed for road vehicles to protect grade crossings requires
environment only when LoS condition provided. careful analysis and extensive testing.
We propose the following architecture: The DSRC
Keywords—DSRC, crossing safety, warning application, RF transmitter deployed on an approaching train transmits to
propagation channel.
DSRC receivers near the crossing. Such receivers can be
I. INTRODUCTION Roadside Units (RSUs) that are mounted on the infrastructure
Collisions between trains and road vehicles are the most near the crossings or On-Board Units (OBUs) carried by
likely accidents that happen near railroad crossings, vehicles. This is illustrated in Fig. 1. The role of the RSU
accounting for approximately 95% of the reported railroad would be to trigger warning signals for vehicles and other
accidents [1]. Such collisions usually happen because of the grade crossing users that do not have a DSRC radio or to relay
vehicle driver’s lack of awareness about the approaching train.
About 80% of U.S. railroad crossings are classified as
unprotected and do not have lights, warnings, or crossing
gates [1]. The existence of safety features at crossings is
critical and can benefit from advances in communications
systems.
Dedicated Short-Range Communications (DSRC) is a
major commercial protocol for Intelligent Transportation
Systems (ITSs). The ITS band is centered at 5.9 GHz with
seven 10 MHz channels between 5.850 to 5.925 GHz. DSRC
provides services in the form of Vehicle-to-Vehicle (V2V) or
Vehicle-to-Infrastructure (V2I), and sends warning messages
to prevent potential collisions among road vehicles. With this Fig. 1. DSRC transmitter and receiver locations for grade crossing protection.

978-1-7281-3616-5/19/$31.00 ©2019 IEEE

Authorized licensed use limited to: Peking University. Downloaded on June 24,2022 at 14:38:52 UTC from IEEE Xplore. Restrictions apply.
the warning messages at obstructed crossings. In this paper, measurements. Both campaigns are collocated and are
we analyze the train-to-OBU scenario and present DSRC and conducted simultaneously.
channel measurement results in wide-open spaces and
artificial shadowing environments. We use these 1) DSRC Performance Measurements
measurements to evaluate the feasibility of DSRC for the The purpose of the DSRC performance measurements is to
proposed application of early warning. By evaluating the determine the feasibility of using DSRC for grade crossing
location of the train transmitter and the position of the vehicle protection by assessing the reliable reception distances in
receiver and matching them to the system performance, we representative grade crossing environments. The performance
assess whether the warning message can be reliably received metric is the packet error rate (PER) as a function of train
for the distances where a potential collision can be avoided. distance from the crossing. The distance is calculated using
Global Positioning System (GPS) coordinates; the PER is
We define the methodology, measurement setup, and provide
computed by comparing the received packets with the
measurement results at crossings of the Transportation
transmitted packets.
Technology Center, Inc. (TTCI) in Pueblo, Colorado, USA.
The rest of the paper is organized as follows: Section II
2) RF Propagation Channel Measurements
presents the related work. Section III introduces the
measurement campaign for DSRC performance evaluation and The purpose of these measurements is to model the
channel sounding. Section IV provides and analyzes the propagation channel around railroad crossings where the
measurement results and Section V derives the conclusions. DSRC performance measurements are taken. Because of the
DSRC performance measurements use 5.875–5.885 GHz, a
II. RELATED WORK direct-sequence spread spectrum (DSSS) channel sounder [6]
A feasibility study of DSRC is presented in [2], where the is used for the RF propagation channel measurements with a
authors demonstrate the feasibility of using DSRC in the range of 5.86–5.91 GHz. More precisely, the channel sounder
agricultural domain with realistic harvesting scenarios. The uses a continuous waveform with a 2,047 long spreading code
authors of [3] introduce the concept of Vehicle-to-Train (V2T) sequence, which is clocked at a rate of 25 MHz (50 MHz RF
communications and present the design of a warning system bandwidth). Both transmitter and receiver are implemented as
as well as measurements performed in Australia; yet the software-defined radios, using GNU Radio software and Ettus
presented measurements are limited and more data is needed USRP B210 hardware. Continuous channel data is saved with
for system evaluation. The authors of [4] and [5] introduce GPS coordinates at intervals of 0.5 s.
antennas that have optimal patterns for train related B. Measurements Sites
communications. Propagation channel models in railroad As shown in Fig. 2 (a), the first scenario is a wide-open
environments are presented in [6] for 5.8 GHz. Similarly, [7- space except for a parking lot and the building next to it. Fig.
11] present channel measurements for modeling and analyses. 2 (b) depicts the second scenario, the artificial shadowing
Those works measure and model the channel for V2V and V2I environment. It has seven railroad freight cars, each being
and here we compare our propagation channel results with approximately 4 m high and 10 m long and there is an
theirs. The authors of [12] present a V2T early warning approximately 10 m separation between them. These railroad
system architecture and DSRC performance and channel cars are on the track between our test track and the OBU. It
propagation measurements. Similarly, [13] provides additional creates a rich multipath propagation environment with
data, extending the results and analyses of [12]. References shadowing.
[12-13] report experiments conducted in a production
environment and do not analyze the best and the worst-case
scenarios and the effect of different train speeds. This paper
extends those measurement campaigns by considering two
extreme railroad environments and a wide range of train
speeds at a configurable test track and compares the results
with previous findings. From the comparisons, we observe
DSRC reception range is depending on the environment rather
than speed and the channel around railroad crossings has an
approximately 3 to 5 dB lower Rician K factor and factor of 2
higher path loss exponent then typical V2V or V2I
environments presented in other papers.
III. MEASUREMENT CAMPAIGN
A. Measurement Types
We combine two measurement campaigns: DSRC
performance measurements and RF propagation Fig. 2. TTCI tracks for wide-open space (a) and artificial shadowing
environment (b). The train direction is from top to bottom.

Authorized licensed use limited to: Peking University. Downloaded on June 24,2022 at 14:38:52 UTC from IEEE Xplore. Restrictions apply.
3) Configurations
The DSRC OBU receivers use a 6 dBi omnidirectional
antenna. The DSRC transmitter uses a 12 dBi omnidirectional
antenna with a 23 dBm power level, which is the highest
transmit power that Cohda MK5 DSRC supports. The antenna
pattern of the 12 dBi omnidirectional antenna is as same as
that of the 6 dBi antenna in the horizontal plane. The
modulation scheme is QPSK for all transmissions. The
channel sounder operates at a 46 dBm output power with 6
dBi omnidirectional antennas installed at the transmitter and
receiver.

4) Train Operations
The train makes several passes through the measurement
region; the train track and locations of the DSRC OBUs are
shown in Fig. 2. For each pass, the train starts at about 1 km
before the crossing, accelerates to the desired test speed of 20,
50, or 79 mph, where 79 mph is the highest speed a
locomotive is permitted to run in the U.S. For all cases, the
locomotive maintains a constant speed through the test area,
and then decelerates to a full stop after passing the
measurement distance of 600 m beyond the crossing.
For both the DSRC and channel measurements, the train
Fig. 3. Operation of train, scenario setup, and DSRC OBU locations for wide-
open space (a) and artificial shadowing environment (b). passes the vehicle mounted radios, which are installed near the
tracks as shown in Fig. 2. For the DSRC measurements, PER
For these two scenarios, we evaluate the feasibility of is evaluated for a single pass. For the channel measurements,
DSRC for grade crossing safety. They represent the best and the channel characteristics are derived from three train passes.
worst-case scenarios for communications, a wide-open space
and a shadowing environment, simulating rural and urban IV. RESULTS AND ANALYSES
settings.
A. Propagation Channel Measurements
C. Measurement Settings Table I provides the propagation channel parameters: the
1) DSRC Radios RMS delay spread (τrms), the mean path loss exponent (n) and
For the DSRC performance measurements, we use Cohda its standard deviation (), and the Doppler spread (fDop) for the
MK5 DSRC radios. One DSRC OBU, which acts as the wide-open space and artificial shadowing scenarios.
transmitter, is installed on the locomotive engine and uses The RMS delay spread for the wide-open space is much
channel 174 (centered at 5.87 GHz). The DSRC OBU antenna lower than any reported value in [9-11]. For the shadowing
is installed on top of the train’s long hood, immediately environment it matches the values in tunnels [9] and highway
behind the engineer’s cabin. The height from the ground to the Non-Line-of-Sight (NLoS) [10] environments. For a more
tip of the antenna is approximately 5 m. detailed evaluation of the impact of multipath, we may need to
The DSRC OBU that acts as the receiver is on the roof of a analyze and compare results with directional antennas as
bucket truck, with the antenna being mounted at a height of presented in [6] and omnidirectional antennas which we
about 1.7 m from the ground. For the wide-open space setup, present in this paper.
the DSRC OBU is placed about 50 m from the crossing as TABLE I. Propagation channel parameters
shown in Fig. 3 (a). For the artificial shadowing setup, the for wide open space and artificial shadowing
DSRC OBU is placed about 3 m from the stationary railroad
cars on the opposite side of the operating track. This is Propagation Channel Parameters
illustrated in Fig. 3 (b). Train
Scenario Speed
(mph) τrms fDop
2) Channel Sounder (ns)
n 
(Hz)
The channel sounder receiver is on top of the train engine,
next to the DSRC transmitter. The channel sounder transmitter 20 0.53 597
Wide Open
is placed next to the DSRC OBU receiver. We scan the 5.875- 50 3.68 3.08 11.5 1085
Space
5.885 GHz range in order to avoid co-channel interference 79 7.75 789
between the two systems. We assume channel reciprocity for 20 132 763
Shadowing
justifying this transmission-reception setup opposed to the 50 56.1 2.76 12.3 826
Environment
DSRC system. 79 73.3 752

Authorized licensed use limited to: Peking University. Downloaded on June 24,2022 at 14:38:52 UTC from IEEE Xplore. Restrictions apply.
The Ricean K-factor lies in the range between 9.8 and 11.3
for the two environments. It matches the values presented in
[6], whereas [8] measures 14.2 for rural and 14.6 on a bridge.
Except for [6] and [8], our obtained Ricean K-factors are
higher than the values reported in other works. With the
observed values, we conclude that minimal fading exists in the
considered clear-to-moderately cluttered railroad
environments.
We observe a path loss exponent of 3.08 for the wide-open
space and 2.76 for the shadowing environment, whereas [7]
measures it as 1.12 for suburban and 1.98 for urban vehicular
environments. Hence, the train-to-vehicle environment has a
higher path loss than the V2V communication environment. (a)
The Doppler spread matches the V2V values obtained in
highway LoS environments as reported in [11].
We conclude that overall the railroad crossing environment
may face less fading, higher path loss, less delay, and higher
Doppler spread than the obtained results for V2V/V2I links
[7-11].
B. DSRC Performance Measurements
We evaluate the DSRC performance using the metric of
PER where 0 represents no errors and close to 1 many packet
errors. We can then observe the distribution of correctly
received packets and its Cumulative Distribution Function (b)
(CDF) relative to the distance from the crossing. From these Fig. 5. DSRC performance for different speeds for the artificial shadowing
plots, we evaluate where the DSRC system provides environment: PER (a) and CDF of correctly received packets (b).
appropriate system performance for the proposed application
a grade crossing.
of radio signaling from an approaching train to a receiver near
References [12-13] suggest a 15 s advance notice as
adequate for a vehicle to receive the warning before it would
reach the crossing. In this paper, we consider 10 s as the
minimum necessary notification time. Then, the vehicle must
receive the packet at train distances of more than 88.8, 222.2,
and 351.1 m before the crossing for speeds of 20, 50, and 79
mph. We examine the PER for different distances from the
crossing to evaluate the system suitability for providing a
timely warning.
Fig. 4 shows the DSRC performance results for the wide-
open space scenario. We find that there are almost no packet
errors for all three speeds. As shown in Fig. 4 (a), the
coverage range after the crossing is shorter than before the
(a) crossing; this is likely because of the existence of an
obstruction on the top of the locomotive located behind the
position of the antenna or because of the building placed next
to the parking lot (Fig. 2 (b)), creating NLoS conditions
beyond 150 m past the crossing. Interestingly, we observe that
the performance of DSRC is not affected by the speed of the
train in the wide-open space scenario. But, even for a lightly
obstructed environment, as is the case here beyond the150 m
mark, the coverage region differs for different speeds (Fig. 4
(a)). About 100% of all error-free packets are received 400 m
before reaching the crossing for low, medium, and high-speed
U.S. trains.
Fig. 4 (b) shows linear increases of the proportions of
(b)
correctly received packets between -600 and +200 m, which at
Fig. 4. DSRC performance for different speeds for the wide-open space: PER a constant speed and periodic transmission means that the
(a) and CDF of correctly received packets (b).
communication system performance is constant. The lower

Authorized licensed use limited to: Peking University. Downloaded on June 24,2022 at 14:38:52 UTC from IEEE Xplore. Restrictions apply.
curves, representing speeds of 20 and 50 mph, indicate a wider Our measurements are only valid for the wide-open space
range of successful packet reception. environment and the artificial shadowing environment with
Fig. 5 shows the DSRC performance results for the stationary receivers. Also, our measurements only use one
artificial shadowing environment. The shadowing effect is transmitter and one receiver; more transmitters that transmit
clearly observable. For all speeds, we observe a performance the same warning messages or more transceivers that acts as
degraded gap, and the reason for this is the shadowing created relays can improve the DSRC performance for this warning
by the railroad cars. This shadowing environment affects application. The conducted measurements confirm the
performance; PER values are lower for higher speeds, but the feasibility of DSRC as a train warning system. This is in line
number of correctly received packets at open sites, near -350m, with prior works, yet more complex scenarios and larger-scale
are similar for three speeds. Only about 130 m before the measurements are needed for a careful evaluation of the
crossing reliable reception is observed. Due to similar number DSRC protocol for improving grade crossing safety via radio
of packets are received near -350m where we assume a LoS
signaling.
condition, we could conclude as lower speed can provide wider
reception range as we observe from Fig. 4 (a) and LoS ACKNOWLEDGMENT
condition is more effective to DSRC performance.
Taking together all results, the two environments and three This work was in part supported by the Federal Railroad
locomotive speeds, we observe that a LoS radio link is more Administration under Grant DTFR53-13-C-00064. The authors
relevant for good DSRC performance than low speed. We find would like to thank Jared Withers and FRA. Also, the authors
that DSRC can perform without any errors and can also be would like to thank Transportation Technology Center, Inc.
employed for the proposed application in both wide-open space staff for their help in planning and conducting the
and a shadowing environment when sufficient LoS condition is measurements.
provided. Repeaters for higher speeds are one option to REFERENCES
improve the service in shadowing environments.
We observe that the large-scale RF propagation channel [1] Federal Railroad Administration, “FRA Office of Safety Analysis,” July
2019. [Online]. Available: http://safetydata.fra.dot.gov/OfficeofSafet
statistics do not have a correlation with the distance-based y/Default.aspx [Accessed July 2019]
DSRC performance results. Note that the channel sounder uses [2] F. Klingler, J. Blobel, and F. Dressler, “Agriculture meets IEEE 802.11p:
a wider bandwidth of 50 MHz and we look at site general a feasibility study,” in Proc. IEEE International Symp. On Wireless
channel parameters and site specific DSRC performance. A Commun. Systems, 2018, pp. 1-6.
specific point-by-point or distance-by-distance comparison [3] J. Singh et al., “Cooperative intelligent transport systems to improve
between the two measurement types might reveal a greater safety at level crossing,” in Proc. 19WKITS World Congr., 2012.
correlation between the channel parameter and the DSRC [4] X. Ma et al., “Analysis of directional antenna for railroad crossing safety
performance data. applications,” in Proc. Consum. Commun. Netw. Conf., 2017, pp. 1-6.
[5] X. Ma et al., “Prototypes of using directional antenna for railroad
V. CONCLUSIONS AND DISCUSSION crossing safety applications,” in Proc. IEEE Consum. Commun.
Networking Conf., 2017, pp.594-596.
We conducted a measurement campaign and analyzed the [6] T. W. Tedesso et al., “Propagation measurements at 5.8 GHz for railroad
DSRC performance and RF propagation channel in two intelligent transportation systems,” in Proc. 2017 IEEE Wireless
scenarios, a wide-open space and an artificial shadowing Commun. and Networking Conf., 2017, pp. 1-6.
environments. Our results show that the railroad crossing [7] M. Yang et al., “Path loss characteristics for vehicle-to-infrastructure
propagation channel characteristics are quite different from channel in urban and suburban scenarios at 5.9 GHz,” in Proc. General
Assembly and Scientific Symp. Int. Union of Radio Science, 2017, pp. 1-
typical V2V/V2I channels. Furthermore, the DSRC link can 4.
perform error-free for all separation distances and train speeds [8] L. Bernado et al., “Time- and frequency- varying K-factor of non-
of interest only when operating in a completely unobstructed stationary vehicular channels for safety-relevant scenarios,” IEEE Trans.
environment for the relevant distances and train speeds. DSRC Intell. Trans. Syst., vol. 16, no. 2, pp. 1007-1017, 2015.
performance was significantly limited in the artificial [9] L. Bernado et al., “Delay and Doppler spreads of nonstationary
vehicular channels for safety-relevant scenarios,” IEEE Trans. Veh.
shadowing environments, requiring very LoS sites to generate Technol., vol. 63, no. 1, pp. 82-93, 2014.
error-free performance. We observe that the effect of the [10] P. Alexander, D. Haley, and A. Grant, “Cooperative intelligent transport
environment surrounding the DSRC radios has a higher systems: 5.9-GHz field trials,” Proc. of the IEEE, vol. 99, no. 7, pp.
impact on the performance than the locomotive speeds. Using 1213-1235, 2011.
the TCCI test track to emulate the best and the worst [11] I. Tan et al., “Measurement and analysis of wireless channel
propagation environments that a communication system can impairments in DSRC vehicular communications,” in Proc. IEEE Int.
Conf. on Commun. 2008, pp. 4882-4888.
face, we identify DSRC as a feasible option for improving
[12] J. Choi et al., “Measurement and configuration of DSRC radios for
grade crossing safety using radio signaling. Its performance Vehicle-to-Train (V2T) safety-critical communications,” IEEE Commun.
can be further improved in challenging radio environments by Lett., vol. 7, no. 3, pp. 428-431.
using DSRC repeaters or directional antennas pointed towards [13] J. Choi, V. Marojevic, and C. B. Dietrich, “Measurements and analysis
the critical region of the road that crosses the railroad. of DSRC for V2T safety-critical communications,” in Proc. IEEE
Connected and Automated Vehicles Symp., 2018.

Authorized licensed use limited to: Peking University. Downloaded on June 24,2022 at 14:38:52 UTC from IEEE Xplore. Restrictions apply.

You might also like