Highway and Rural Propagation Channel Modeling
for Vehicle-to-Vehicle Communications at 5.9 GHz
Lin Cheng*(1), Benjamin E. Henty(1), Fan Bai(2) and Daniel D. Stancil(1)
(1) Carnegie Mellon University, Pittsburgh, PA, 15213
(2) General Motors Research, Warren, MI, 48090
E-mail:
[email protected] Introduction
Recent advances in computing and wireless communication technologies have
increased interest in outdoor vehicular networks. Dedicated Short Range
Communications (DSRC) systems have been proposed to support safety and
service operations for vehicular networks. In North America, the Federal
Communications Commission has allocated 75 MHz of spectrum at 5.9 GHz for
DSRC based information exchange between vehicles [1]. Because vehicular
networks bring significant potential for a wide range of services and applications,
there exists a need to gain a detailed understanding of various vehicle-to-vehicle
(V2V) propagation channels that cover diverse and rapidly changing on-road
environments.
This paper reports channel measurements made using an on-road vehicular
testbed equipped with programmable laboratory instruments. In particular, this
location-aware channel sounding platform allows us to conduct statistical
measurement campaigns for the outdoor V2V propagation channel as a function
of location. Relevant Antenna effects are also discussed.
Reported experimental studies include narrow-band measurements conducted at
5.2 GHz described in [2], joint Doppler-delay power profile measurements at 2.4
GHz presented in [3], and vehicle-to-vehicle channel measurements and modeling
at the 5.9 GHz band reported in [4].
Measurement Campaigns
Simultaneous mobility and RF measurements are desired as the vehicles traverse
different environments. The RF platform we developed in [5] was improved by
using Rubidium clocks to prevent the carrier drift. This is particularly useful for
long time measurements. The experiments used roof mounted antenna on both
vehicles. Fig. 1 depicts snapshots of the highway and rural environments during
the measurement campaigns. The highway data presented here was measured on
interstate highway route 79 located north of Pittsburgh, Pennsylvania. This
segment of interstate highway consists of two and three lane segments with
moderate traffic. During the experiments, the two vehicles were moving steadily
at high speed, along with other vehicles with similar speeds. There exist cases
978-1-4244-2042-1/08/$25.00 ©2008 IEEE
Authorized licensed use limited to: TELECOM Bretagne. Downloaded on February 26,2010 at 07:22:06 EST from IEEE Xplore. Restrictions apply.
when the two vehicles changed lanes and overtook each other. There were in
general no nearby buildings, but occasional overpasses were observed. The rural
data was acquired as the vehicles traversed the rolling country side north of
Pittsburgh, PA. This area is characterized by sections of open road interspersed
with towns and intersections. The traffic was very light on these 2-lane roads, and
the vehicles did not pass one another. Nearby clutter consisted of different kinds
of low-height vegetation. Remote trees and hills were also commonly observed.
(a) Highway (b) Rural
Fig. 1: Snapshots of the highway and rural environments.
Channel Modeling
Owing to the complexity of the real environment, empirical channel models are
desired in large-scale vehicle-to-vehicle network simulations to ensure realistic
results. Here we provide parameters for a dual-slope log-normal pathloss model
extracted from our measurements. While the statistical variations in V2V
channels are generally not normally distributed owing to the line-of-sight
component [5], the normal distribution is an attractive approximation owing to its
simplicity.
The received signal strength was computed from the frequency spectrum of the
signals. We obtained the signal strength in each sweep by taking the difference
between the total power in a fixed bandwidth with and without a signal present.
For implementation details we refer the readers to [5].
We now describe the large-scale pathloss and shadowing model as a function of
distance between the transmitter and receiver. We fit the measured pathloss to a
log-distance pathloss model, as shown in Fig. 2. The slopes of the first segment
are close to 2. For the second segment, the slopes are observed to be around 4.
Given the transmit and receive antenna heights, the breakpoint distance calculated
using the first Fresnel zone at 5.9 GHz is dc = 225m. Referring to Fig. 2, the
breakpoint distances that gave the best results are very near this value. It is
interesting to compare these results with our previous suburban studies in [7],
where a breakpoint of around 100m gave the best fit to the data. The shorter
critical distance may be caused by more densely distributed objects like vehicles
Authorized licensed use limited to: TELECOM Bretagne. Downloaded on February 26,2010 at 07:22:06 EST from IEEE Xplore. Restrictions apply.
and pedestrians on the road, creating reflections from points higher than the
ground [5]. Since there are fewer pedestrians and other objects on the road in rural
or highway areas compared to suburban areas, the likelihood of having reflections
from objects above the ground is much smaller compared to the suburban case.
(a) Highway (b) Rural
Fig. 2: Pathloss measurements and approximate large-scale models. (a) Highway.
Dual-slope model with γ1=1.9, σ1=2.5 dB; γ2=4.0, σ2=0.9 dB; and dc=220 m. (b)
Rural. Dual-slope model with γ1=2.3, σ1=3.2 dB; γ2=4.0, σ2=0.4 dB; and dc=226m.
Here γi , σi are the pathloss exponent and standard deviation for the ith segment,
and dc is the breakpoint distance.
Next consider the spread of measured values about the average. We observe that
in the suburban environment, the σ2 values (standard deviations beyond the
breakpoint) reported in [5] are larger than highway and rural environments
reported here. This difference may result from a greater number of nearby objects
such as houses and buildings in the suburban environment than in rural or
highway environments.
In addition to the above comparison and analysis, it is interesting to note that in
the highway data, we observe data points with much shorter vehicle separations
than observed in the rural case (as short as 2 meters). These correspond to the case
when the vehicles pass one another on the two-lane or three-lane highway
segments. This situation is not observed in the rural measurements with one lane
in each direction.
In the case of vehicles passing, variations can also arise from the radiation pattern
of the antennas. Fig. 3 shows the measured antenna pattern above a simple ground
plane for the actual antennas used for these measurements. In the on-road
experiments, the roof-mounted antennas were placed with their direction of low-
gain to the right side of the vehicle to ensure the best gain when the two vehicles
were in car-following scenarios (majority of the cases). However, when a vehicle
passes another, the gain of the passing vehicle antenna can vary rapidly as the
angle between the vehicles changes. This suggests that measurements made
during vehicle passing should be corrected for changes in antenna gain before
inferring changes with distance.
Authorized licensed use limited to: TELECOM Bretagne. Downloaded on February 26,2010 at 07:22:06 EST from IEEE Xplore. Restrictions apply.
Conclusion
In this article, we have reported experimental studies of signal strength as a
function of vehicle separation for outdoor vehicle-to-vehicle propagation at 5.9
GHz. Statistical measurement campaigns were conducted in highway and rural
driving environments. These measurements were used to obtain parameters for a
dual-slope log-normal propagation model. The effects of antenna pattern
variations for passing vehicles were also discussed.
Fig. 3: Measured antenna pattern above a ground plane (not mounted on vehicle).
References
[1] Standard specification for telecommunications and information exchange
between roadside and vehicle systems - 5GHz band dedicated short range
communications (DSRC) medium access control (MAC) and physical layer
(PHY) specifications.
[2] J. Maurer, T. Fugen, and W. Wiesbeck, “Narrow-band measurements and
analysis of the inter-vehicle transmission channel at 5.2 GHz,” IEEE
Vehicular Technology Conference, vol. 3, pp. 1274 – 1278, 2002.
[3] Guillermo Acosta, Kathleen Tokuda, and Mary Ann Ingram, “Measured joint
doppler-delay power profiles for vehicle-to-vehicle communications at 2.4
GHz,” GLOBECOM ’04. IEEE, vol. 6, pp. 3813 – 3817, 2004.
[4] Guillermo Acosta and Mary Ann Ingram, “Doubly selective vehicle-to-
vehicle channel measurements and modeling at 5.9 GHz,” Wireless Personal
Multimedia Communications Conference, Sept, 2006.
[5] Lin Cheng, Benjamin Henty, Daniel Stancil, Fan Bai and Priyantha Mudalige,
‘‘Mobile vehicle-to-vehicle narrow-band channel measurement and
characterization of the 5.9 GHz dedicated short range communication (DSRC)
frequency band,’’ IEEE Journal on Selected Areas in Communications, vol.
25, no. 8, pp. 1501 --- 1516, 2007.
Authorized licensed use limited to: TELECOM Bretagne. Downloaded on February 26,2010 at 07:22:06 EST from IEEE Xplore. Restrictions apply.
