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

Performance Evaluation of IEEE 1609 WAVE

and IEEE 802.11p for Vehicular Communications


Sebastian Gräfling, Petri Mähönen and Janne Riihijärvi
Institute for Networked Systems, RWTH Aachen University
Kackertstrasse 9, D-52072 Aachen, Germany
email: {sgr, pma, jar}@inets.rwth-aachen.de

Abstract—In this paper we study the performance of the IEEE IEEE 1609.1 WAVE
Resource Manager
IEEE 1609.2 WAVE
Security Services for Applications
1609 WAVE and IEEE 802.11p trial standards for vehicular and
Management
communications. We have implemented key components of these MIB
IEEE 1609.3 WAVE
Networking Services Messages
standards in a simulation environment also supporting realistic IEEE 1609.4 WAVE
vehicular mobility simulation. We study both the overall capacity Multichannel Operation
(MAC Extension)
of vehicular networks utilizing the said standards, as well as WAVE
Management
delay performance, which is an extremely important performance Entity MLME
IEEE 802.11p
WAVE MAC
metric especially for safety applications. Our results show that the
traffic prioritization schemes selected for the standards work well, PLME
IEEE 802.11p
WAVE PHY
and even in the presence of multi-channel operation implemented
by the IEEE 1609.4 the delay of control messages of highest
priority remains on the order of tens of milliseconds. Thus even Fig. 1. The IEEE 1609 (WAVE) reference architecture and relationship to
the IEEE 802.11p MAC and physical layers.
with high densities of vehicles these standards should yield a
stable platform a variety of vehicular applications can be built
on.
the requirements set forth by the various consortia working
I. I NTRODUCTION
on vehicular communications. In particular, we study how
Vehicular communications has become one of the most well the prioritization schemes implemented can guarantee
active domains of research in the wireless networking com- low delays and high channel access rates for safety-critical
munity. Vehicular networks (VANETs) present a challenging traffic, which still supporting low-priority applications with
environment for protocol and application design due to their high throughput requirements. The novel component in our
potentially very large scale, as well as high degree of dy- study is the integration of both WAVE and 802.11p into the
namism they tend to exhibit. Also the requirements of different same simulation environment. The performance of 802.11p has
applications for the protocol stack vary widely, ranging from been studied in isolation in a number of studies [3], [4], [5],
low overhead delay tolerant infotainment applications all the as has been the performance of regular Wi-Fi for vehicular
way to safety critical applications such as collision avoidance use [6], [7], [8]. However, to the best of our knowledge, the
and traffic management. The IEEE 1609 (WAVE) working present study is the first to explore performance of both WAVE
group [1] has been developing first version of the architecture and 802.11p in an integrated fashion using realistic models for
for the lower layers of the protocol stack specifically tailored vehicular mobility.
towards VANETs. The foundation of the IEEE 1609 family of The rest of the paper is structured as follows. In Section
standards is the IEEE 802.11p standard [2] defining physical II we give a short overview of the relevant draft standards,
and medium access control layers, heavily based on earlier namely both IEEE 1609 as well as IEEE 802.11p. In Section
standards for Wireless LANs. Work in both of these groups III we then outline our simulation scenarios, including the
is currently in draft or trial standard stages, with early experi- mobility models considered. Results from the simulations
mental implementations starting to become available. Thus, at are discussed in detail in Section IV, before drawing the
present time it is critical to begin to obtain experience from conclusions in Section V.
both simulation and laboratory experiments as well as actual
prototype deployments on the performance of the technologies II. OVERVIEW OF IEEE WAVE AND IEEE 802.11 P
being standardized. We will present in this section a short overview of IEEE
In this paper we carry out a comprehensive evaluation 1609 WAVE design, as well as its relationship to IEEE
of the IEEE 802.11p PHY/MAC design together with the 802.11p. Figure 1 depicts the high-level WAVE architecture.
multichannel extensions incorporated through the WAVE draft The different standards belonging to the overall WAVE stan-
standards. We have implemented these protocols in Qualnet, a dards family involve various networking and management
commercial packet-level network simulator. Based on realistic functions, including security, resource management, and sup-
mobility models and our implementation we seek to obtain port for multichannel operation. Additionally, the architecture
understanding on how well both WAVE and 802.11p can meet has logically separated management functions for the medium

978-1-4244-8086-9/10/$26.00 ©2010 IEEE 344 ICUFN 2010


Channel
172 174 176 178 180 182 184
Network Layer
Number

Channel Service Service Service Control Service Service Service


Type Channel Channel Channel Channel Channel Channel Channel

Channel Routing
Traffic Critical Critical Traffic Traffic
Application Non-Safety Non-Safety
Efficiency Safety Safety Efficiency Efficiency

Coordination
Channel
Radio
C2C Medium Medium All Short Short Intersections ...
Range

SCH1
Tx Power
33 dBm 33 dBm 33 dBm 44.8 dBm 23 dBm 23 dBm 40 dBm
Level
SCH0
AC 0 AC 1 AC 2 AC 3
5.855 5.865 5.875 5.885 5.895 5.905 5.915 5.925 Prioritization CCH
AC 0 AC 1 AC 2 AC 3
Frequency (GHz)
AC 0 AC 1 AC 2 AC 3
Fig. 2. The set of channels defined in the WAVE trial standard for multi-

FIFO Queues
channel operation in vehicular networks.

AIFS[AC]

AIFS[AC]
TXOP[AC]

TXOP[AC]

CW[AC]

CW[AC]
AIFS[AC]

AIFS[AC]
CW[AC]

CW[AC]
AIFS[AC]
AIFS[AC]
CW[AC]TXOP[AC]

TXOP[AC]

CW[AC]
CW[AC]
AIFS[AC]

AIFS[AC]
CW[AC]

CW[AC]
access control layer and the physical layer, denoted by MLME

AIFS[AC]

AIFS[AC]

AIFS[AC]

AIFS[AC]
CW[AC]

CW[AC]

CW[AC]
and PLME in figure, respectively. The actual MAC and PHY Internal Contention
used by WAVE are standardized separately by the IEEE
Internal Contention
802.11p group, and are discussed in more detail below.
Internal Contention
The IEEE 1609.4 standard for multichannel operation in
VANETs defines a number of channels, each for differ-
ent application, and with different characteristics, as shown
in Figure 2. Notice especially that in addition to different Medium Contention

frequency being used, different maximum transmit powers


are allotted for different channels. The safety-critical control Fig. 3. Illustration of the relationships between the multichannel operation
and the traffic prioritization with different access categories in the 802.11p
channel has the highest potential transmit power, while non- MAC.
critical applications and short-range safety applications are
to use channels with smaller allowed transmit powers. In
IEEE 1609.4 different channels, from the control channel 58 μs AIFS =
(CCH) to the six service channels (SCH), are utilized in time AC3
CCH 58 μs 39 μs 97 μs
58 μs
AIFSN x Slot Time + SIFS

multiplexed fashion, with the control channel being served SCH 58 μs 39 μs 97 μs


39 μs
every other timeslot. The rest of the timeslots are used by the
Backoff Time =
CCH 71 μs 39 μs 110 μs CWmin x Slot Time
39 μs
different service channels, depending on the actual application AC2
SCH 58 μs 91 μs 149 μs
requirements.
CCH 110 μs 91 μs 201 μs
Below the IEEE 1609.4 multi-channel operation further pri- AC1
SCH 71 μs 195 μs 266 μs
ority scheme is implemented by the IEEE 802.11p MAC layer
in a fashion similar to the IEEE 802.11e Enhanced Distributed AC0
CCH 149 μs 195 μs 344 μs

Coordination Function (EDCA) to regular Wi-Fi networks. For SCH 123 μs 195 μs 318 μs

each channel four Access Categories, denoted by AC0–AC3,


are defined, with AC3 having the highest priority. As for the 50 100 150 200 250 300 350 t

plain EDCA, frames for traffic belonging to different access Fig. 4. The arbitration inter-frame space durations used in the EDCA settings
categories get placed in different queues, which are served for IEEE 802.11p for different channel types and access categories.
based on an internal contention procedure. This procedure
illustrated in Figure 3. Queues for different access categories,
and for different channel types have different timer settings
III. S IMULATION S ETUP AND S CENARIOS
related to the internal contention procedure, as illustrated in
Figure 4. Frame winning the internal contention is then subject As already discussed in the introduction we have imple-
to the regular CSMA/CA contention procedure. mented the WAVE multichannel operation specified by IEEE
The IEEE 802.11p physical layer is OFDM-based, and quite 1609.4, together with the IEEE 802.11p PHY and MAC
similar to the IEEE 802.11a physical layer design. The main layers in Qualnet [9]. The implementations of the latter are
difference is in the overall bandwidth used, which is 10 MHz heavily based on the existing IEEE 802.11a and IEEE 802.11e
for IEEE 802.11p instead of the 20 MHz of IEEE 802.11a. implementations, which mainly needed reparameterization to
Practically all the other changes in the PHY stem from this take into account the changes between the standards. The IEEE
difference. Only other major differences are the changes in 1609.4 implementation on the other hand was created as an
allowed transmit powers and operating frequencies already extension to the existing codebase. We have carefully imple-
encountered in Figure 2. mented all the necessary parts of the standards under study in

345
2500

802.11p SCH0
2250 802.11p SCH1
802.11p SCH2
2000 802.11p CCH
802.11p SCH4
1750 802.11p SCH5
802.11p SCH6
802.11a
1500

Radio Range (m)


802.11b

1250

1000

750

500

250

Fig. 5. The road network within Washington D.C. obtained from the TIGER 0
1 2 3 4 5 6 9 11 12 18 24 27
Data Rate (Mbps)
database used with VanetMobiSim for generating realistic traces of vehicular
mobility.

Fig. 6. Relationships between radio ranges of transmitters of different WAVE


channels and Wi-Fi standards for different data rates (IEEE 802.11p: 3 to 27
order to enable realistic performance evaluation, but simplified Mbps, IEEE 802.11a: 6 to 54 Mbps, IEEE 802.11b: 1.5 to 11 Mbps).
the implementation architecture somewhat compared to the
WAVE reference architecture in order to better match the
Qualnet protocol stack structure.
We considered three distinct scenarios in our simulations. AC3
First was simply a flat open air environment with a number AC2
1,200 AC1
of static vehicles. This scenario was mainly used to test the AC0
behavior of the IEEE 802.11p physical layer, as well as the
Throughput (kbps)

overall scalability of the medium access control solutions. The


800
second scenario considered was an urban one, corresponding
to a region within Washington D.C. Digital form of the
corresponding road network was obtained from the publicly
400
available TIGER dataset [10], and turned into a graph form
shown in Figure 5. This was used as a basis for mobility simu-
lations, which were created using the VanetMobiSim tool [11].
0
Based on the road network data also shown in Figure 5 and SCH0 SCH1 SCH2 CCH SCH4
WAVE Channel
SCH5 SCH6

the Intelligent Driver Model VanetMobiSim generates realistic


mobility traces for any given number of vehicles, which can
then be exported into Qualnet. We used this scenario quite Fig. 7. The saturation throughputs for different WAVE channels and different
access categories.
extensively for determining how the standards under study are
expected to perform under typical usage scenarios. Finally,
the third scenario considered in the simulations corresponds to
roughly linear segment of a long highway, with a number of Figure 7 illustrates the aggregate saturation throughput ob-
road side units (RSUs) placed at varying intervals to generate tained in the static scenario for different channels, with all ac-
traffic towards vehicles passing by. Such a scenario is quite cess categories being used equally for the transmitted frames.
typical for a number of safety applications of various levels of From the figure we see that the prioritization mechanism based
criticality, and thus an important element in any performance on access classes functions as expected, allocating highest
evaluation for VANET technologies. proportion of capacity to the high priority access categories.
We also see that as expected from the time sharing scheme
IV. R ESULTS between the channels, the control channel is allocated the
We shall now discuss the results we have obtained from largest share of capacity. As can be seen from Figure 8, these
the simulations described in the previous section. Figure 6 conclusions also hold in the urban scenario with mobility, and
shows the expected radio ranges for different WAVE channels for a varying number of concurrently transmitting vehicles.
as well as for regular Wi-Fi standards. The impact of the higher We shall next look into the highway scenario together with
allowed transmit power is clearly very large, especially for the the roadside units. Figure 9 shows the obtained throughput
control channel traffic which can be successfully received even over a three-kilometer stretch of the highway, for four different
at a distance of 2.5 km in an open air scenario. Naturally in configurations of the RSUs. From the results shown in the
an urban environment significant reduction of the radio range figure we can infer that in order to receive a steady data stream
would be expected due to heavy shadowing effects. from the RSUs to a moving vehicle, inter-RSU distance should

346
900
SCH0
800 SCH1
SCH2
SCH0-AC3
CCH
SCH0-AC2
700 SCH4 0.25 SCH0-AC1
SCH5
SCH0-AC0
600 SCH6
CCH-AC3

Average End-to-End Delay (s)


CCH-AC2
Throughput (kbps)

0.2
CCH-AC1
500
CCH-AC0

400 0.15

300 Maximum allowed


delay for C2C 0.1
safety messages
200

0.05
100

0 0
2 4 6 8 10 12 14 0.3 0.1 0.03 0.01 0.003 0.001
Number of Vehicles MCBR Inter Departure Time (s)

Fig. 8. The aggregate throughputs of different WAVE channels in the urban Fig. 10. The delay for service and control channels under different traffic
scenario with mobility. loads in the static scenario.

2
5 10
x 10
7
AC0 - 1024 byte
1 RSU (0m)
AC1 - 512 byte
2 RSUs (0m,3000m) 1 AC2 - 256 byte
10
6 3 RSUs (0m,1500m,3000m) AC3 - 128 byte

4 RSUs (0m,1000m,2000m,3000m)

5 Average End-to-End Delay (s) 10


0
Throughput (kbps)

4 -1
10

3
-2
10

2
-3
10

-4
10
802.11a 802.11b 802.11p 802.11a 802.11b 802.11p 802.11a 802.11b 802.11p 802.11a 802.11b 802.11p
0 | 4 nodes | 15 nodes | 30 nodes | 60 nodes |
0 300 600 900 1200 1500 1800 2100 2400 2700 3000
Distance (m)

Fig. 11. The delay for different number of vehicles and different link-layer
Fig. 9. Throughput for transmission from road side units with varying number technologies in the urban scenario with mobility.
and distances to a vehicle in the highway scenario.

urban scenario, and also for usual Wi-Fi technologies.


be in the range of 1000 m–1500 m.
While throughput is certainly a key metric to be studied V. C ONCLUSIONS
for any radio access technology, delay is also of critical In this paper we studied in detail the performance of
importance for many of the expected applications in VANETs, the IEEE 1609 WAVE and IEEE 802.11p trial standards
especially for those related to traffic safety. The commonly for vehicular networks. We implemented key components
agreed upon maximal delay for such applications is 100 ms. of these standards in the Qualnet simulation environment,
Figure 10 shows the end-to-end delay for different traffic loads and included into the simulations realistic vehicular mobility
(measured in terms of the inter-departure times of packets from patterns generated using VanetMobiSim together with actual
the application layer) and for both service and control channel road network data. We studied both the overall capacity of
traffic with different access categories. We see that while the vehicular networks utilizing WAVE and 802.11p, the expected
service channels can suffer from relatively long delays, control communication ranges, as well as the delay performance. The
channel traffic typically experiences delays well below the results showed that the traffic prioritization schemes selected
100 ms bound also illustrated in the figure. Only when the for the standards work well, and even in the presence of multi-
total offered traffic approaches 1000 packets per second the channel operation implemented by the IEEE 1609.4 the delay
delay on the control channel becomes too high for safety- of control messages of highest priority remains on the order of
critical applications. The situation is not substantially affected tens of milliseconds. Only when the total offered traffic within
by mobility as shown in Figure 11, giving the results for the the radio range approached 1000 packets per second the delay

347
started to become excessive. Thus the results indicate that both
WAVE and 802.11p appear to form a solid foundation for
vehicular communication applications, but there still remains
work to be done on higher layers to guarantee that for safety
applications the overall network load remains under control.
ACKNOWLEDGMENT
The authors would like to thank RWTH Aachen University
and the German Research Foundation (Deutsche Forschungs-
gemeinschaft, DFG) for providing financial support through
the UMIC research centre. We would also like to thank the
European Union for providing partial funding of this work
through the ARAGORN and the FARAMIR projects.
R EFERENCES
[1] R. Uzcátegui and G. Acosta-Marum, “Wave: a tutorial,” IEEE Commu-
nications Magazine, vol. 47, no. 5, pp. 126–133, 2009.
[2] D. Jiang and L. Delgrossi, “IEEE 802.11 p: Towards an international
standard for wireless access in vehicular environments,” in Proc. of IEEE
Vehicular Technology Conference (VTC) Spring, 2008.
[3] Y. Wang, A. Ahmed, B. Krishnamachari, and K. Psounis, “IEEE
802.11 p performance evaluation and protocol enhancement,” in IEEE
International Conference on Vehicular Electronics and Safety, 2008.
ICVES 2008, 2008, pp. 317–322.
[4] S. Eichler, “Performance evaluation of the IEEE 802.11 p WAVE
communication standard,” in 2007 IEEE 66th Vehicular Technology
Conference, 2007. VTC-2007 Fall, 2007, pp. 2199–2203.
[5] Y. Zang, L. Stibor, G. Orfanos, S. Guo, and H. Reumerman, “An
error model for inter-vehicle communications in highway scenarios at
5.9 GHz,” in Proceedings of the 2nd ACM international workshop
on Performance evaluation of wireless ad hoc, sensor, and ubiquitous
networks. ACM, 2005, p. 56.
[6] D. Cottingham, I. Wassell, and R. Harle, “Performance of IEEE 802.11
a in vehicular contexts,” in IEEE Vehicular Technology Conference.
Citeseer, 2007, pp. 854–858.
[7] M. Wellens, B. Westphal, and P. Mähönen, “Performance evaluation of
IEEE 802.11-based WLANs in vehicular scenarios,” in Proceedings of
the 65th IEEE Vehicular Technology Conference (VTCÕ07), pp. 1167–
1171.
[8] V. González, A. Santos, C. Pinart, and F. Milagro, “Experimental
demonstration of the viability of IEEE 802.11 b based inter-vehicle
communications,” in Proceedings of the 4th International Conference on
Testbeds and research infrastructures for the development of networks &
communities. ICST (Institute for Computer Sciences, Social-Informatics
and Telecommunications Engineering), 2008, p. 1.
[9] “Qualnet, Scalable Network Technologies,” http://www.qualnet.com/.
[10] “Topologically Integrated Geographic Encoding and Referencing sys-
tem,” http://www.census.gov/geo/www/tiger/.
[11] “VanetMobiSim, framework for mobility modeling,” http://vanet.
eurocom.fr/.

348

You might also like