Congestion Control Techniques in VANETs: A Survey

Mustapha Younes Taleb Salah Merniz Saad Harous

MISC Laboratory MISC Laboratory College of Information Technology
Constantine2 University Constantine2 University United Arab Emirates University
Constantine, Algeria Constantine, Algeria United Arab Emirates
[email protected] [email protected] [email protected]

Abstract— During these last years, Intelligent Transport Systems Figure 1 shows the VANET protocol stack which
(ITS) have experienced a great growth in both areas: academic represent all algorithms and protocols that are involved in the
and industrial. ITS which aim to increase safety and comfort for VANET technology.
users’ transport, are essentially governed by Vehicular Ad hoc
Networks: VANETs. In such networks, nodes represent smart
vehicles that can communicate either between themselves to
exchange traffic information or with roadside infrastructure to
disseminate or request useful information. Therefore, the
congestion control remains one of the most challenging problems
of these networks. This paper surveys congestion control
techniques, which are divided into three categories: Rate
adaptation, Media access control (MAC) and trajectory based
schemes. For each technique we give its principle, its merits and
its limits. A comparative study with respect to some relevant
metrics is given as well.
Keywords—VANETs, congestion control, routing protocols, rate
adaptation, MAC layer.

I. INTRODUCTION
Vehicular networks are a sub-category of mobile networks
developed over smart vehicles that play the role of mobile
nodes. It consists of a system of connections between smart Figure 1: VANET Protocol Stack
vehicles equipped with sensors. Vehicular networks are
characterized by: highly dynamic topology, exchange between
heterogeneous nodes, lack of energy constraints (storage II. VEHICULAR NETWORKS
capacity, computing power) and important information relay. A vehicular network is a communication network between
We consider three kinds of messages in vehicular networks: intelligent vehicles equipped with on-board units allowing
Data Messages, Periodic Messages and Critical Event vehicle-to-vehicle as well as vehicle-to-infrastructure
Messages [3]. The last two are the ones that are taken into communication over 802.11p MAC protocol.
consideration in the majority of congestion control algorithms.
In fact, Critical Event Messages, called also Event-Driven VANETs are part of the family of Mobile Ad hoc
Safety Messages (ED) or Warning Messages, are sent when a Networks (MANET) that operate in point-to-point links
vehicle or the road infrastructure discovers a critical situation. without infrastructure. In a Vehicular network, the nodes are
They must be relayed with no delay, because they carry time- composed of the intelligent vehicles belonging to the network
critical information of high importance, while Periodic and the road side units.
Messages, called also Beacon Messages, are broadcasted
periodically (every 2 seconds) by all nodes to send and receive There are mainly three kinds of vehicular networks:
information about the network.
• Vehicular infrastructure networks
These aspects frequently cause heavy traffic on the network,
• Vehicular ad hoc networks
to the point of overloading it. To solve this problem,
researchers have proposed several approaches, at different • Hybrid Vehicular Networks
levels, and developed a number of schemes and algorithms for
congestion control. VANETs have mostly the same characteristics as
MANETs, but they are distinguished by the following points:

978-1-5090-4372-9/17/$31.00 ©2017 IEEE 484

 • Don’t suffer from energy resources, computation Results and discussion: simulation tests were done by
power and storage space constraints sending bursts of ED Messages by one node and measuring the
• High mobility (high vehicle’s speed) percentage of vehicles located within the transmission area of
• Highly dynamic topology the sender node that received an ED Message correctly.
• Frequent disconnection (du to rapid topology Results show that this rate of vehicles is higher when applying
changes) this control congestion scheme by 15% at least.
However, this effectiveness is observed only when one or
several nodes generate many ED Messages in a very short
III. CONGESTION CONTROL MECHANISM time, which only happens in very rare cases. It doesn’t
We can categorize congestion control techniques that exist improve the performance of the general network transmissions.
into three major groups namely: Rate adaptation and Media 2) Performance Evaluation of Beacon Congestion Control
access control and trajectory based schemes [1], but we focus Algorithms for VANETs [3]:
in this paper only on the two first groups. This method is also related to Periodic Messages and it
In the rate adaption category, we find algorithms that involves three phases: monitoring, estimation and adjustment.
regulate the flow rate according to the state of the network, in Its principle is to adjust the rate of Periodic Messages
order to generate a new flow rate away from congestion zone. according to the received ED Messages using three different
The media access control category algorithms work methods.
differently. These algorithms use different techniques of Rate Control (RC): this method aims to estimate the
media access according to the state of the network. sending rate of Periodic Messages using the following
metrics:
• CBTth: channel busy time threshold
IV. COMPARATIVE STUDY • Ni(t): number of nodes that have node i in their
carrier-sense range (CR)
A. Rate adaptation schemes
• Psize: Periodic Message size
In this category we find mainly techniques which deal with • C: positioning constant
the rate of sending Periodic Messages and the priority of The adapted rate is estimated by the following equation:
Event-Driven Messages (ED Messages) in order to avoid or
reduce the network congestion.
1) Distributed Rate Control Algorithm for VANETs 1 .
1
(DRCV)[2]: This scheme is a distributed algorithm, it
runs on all network nodes. It is concerned with the control Power Control (PC): unlike the first method, this one
of Periodic Messages, it is composed of three phases. The seeks to control the Periodic Messages transmission power
first one is a monitoring phase of the transmission by assessing the nodes communication range (CR), which
channel to determine the following network parameters : corresponds to the range where transmitted packets can be
• Number of neighbors received with a high probability. With high (CR), the
• Data Rate Package (DPR) transmitted packets will be received by more distant nodes,
• Chanel Busy Time (CBT) but with higher probability of collision due to signal
The second phase is the phase of estimating the future interferences. Conversely, if the (CR) is small, transmitted
sending rate, where the algorithm determines what speed packets will be received by fewer nodes, but with smaller
should be used for next transmission. The third phase is the probability of collision.
action phase, where the algorithm decides whether to send
Periodic Messages with the determined flow rate in the Joint Power + Rate Control (PRC): this third approach is
previous phase or wait for a period before sending. The the merging of the two previous methods, to perform -at the
decision is based on the received ED Messages. In other words, same time- a rate adaptation and a power control of sending
if a node receives an ED Message at time t, it will not send any messages. In other words, to avoid the network congestion,
Periodic Message until t+Δt. After this delay it sends Periodic the flow rate is adjusted by the equation above, and the
Messages with the estimated rate in phase 2. Figure 2 power transmission is decreased.
illustrate these three phases. Results and discussions: simulation tests were performed by
measuring three parameters (Beaconing rate, ED Messages
rate and CBT rate) in many situations; three intersection size
levels (small, midsize and large), for each one by two vehicles
density levels (low and high). The expected results are: a
significant increase in ED Messages rate against an important
decrease in beaconing rate and CBT rate. This was the case in
most of simulation tests, but it was not with small intersection
sizes. In fact, in the case of a network with small intersections,
Figure 2: Phases of DRCV scheme ED Messages always reach destination. It depends only on the

485
distance between nodes, and this algorithm -in terms of results Results and discussion: simulation tests highlight the
- will be reduced in an adaptation of Beacon Messages. contribution of this algorithm by measuring only the warning
delay (delay of ED Messages). This warning delay is greatly
3) Congestion Control Algorithm in VANETs [4]: lesser compared to the reference model (absence of congestion
This algorithm is concerned with both ED Messages and control), but making a comparison of models according to one
Periodic Messages. It aims to ensure high reliability and criterion is a little unbiased.
timely delivery of disseminating these messages. It is divided There is more than one shortcoming in this method. In fact, we
into two independent modules: Measurement Based Detection can see that the first module (Measurement Based Detection)
and Event Driven Detection. Each module is related to one type maintains a static threshold (5 packets in the queue). Once this
of messages. threshold is reached, messages will be discarded. This
Measurement Based Detection: this part will monitor the condition is too strong, especially when the network
Control Channel (CCH) to determine when the CCH is congestion is concentrated in a small zone (not all vehicles).
congested. In this system the CCH is considered to be Furthermore, we notice in the second module (Event
congested when the number of packets in the queue reaches Driven Detection) that the whole queue of Periodic Messages
five. After this threshold, all incoming Periodic Messages is frozen when an ED Message arrives. This condition is also
will be ignored (they will be discarded). too strict because it prioritizes ED Messages totally, which is
Event Driven Detection: this module will monitor ED not fair, it would have been better to propose a weighted
Messages to lunch the congestion control procedure prioritization.
immediately when an ED Message is detected. The In addition, when trying to reduce a lot the measured
congestion control procedure will freeze all transmission criterion in the simulation (warning delay) at the expense of
queues except the ED queue to allow a high transmission Beacon Messages, this could easily get the opposite effect.
reliability for ED Messages.

These two parts of the congestion control scheme are B. Media access control
illustrated in figure 3. 1) Multi-priority supported medium access control in
VANETs (MP-MAC)[5]
This algorithm uses a technique, which defines different
priorities for packet transmission, for both ED messages and
Beacon messages. It uses a multi-priority Markov process to
optimize the use of the channel according to the network state.

Besides, it implements a p-persistent MAC scheme to

reduce the probability of collisions during the transmission

This algorithm implements a strategy that defines several

levels of priorities according to the type of packets. Priorities
of high level are devoted to ED Messages and Periodic
Messages. It implements also a multi-priority Markov process
to use the transmission channel more efficiently according to
the network traffic.

2) Cooperative MAC Protocol in VANETs(VC-MAC)[6]

This protocol uses the concept of cooperative
communication tailored for VANETs. So as to maximize the
network throughput, the broadcast is made by the access point
based on the assumption that under the information-
downloading scenario, all vehicles are interested in the same
messages. During the transmission, a set of vehicles may not
receive the right messages, this is due to the unreliability of
the wireless channel. Then, the vehicles that received the
messages will be selected to relay to their neighbors.
Therefore, in order to reduce the collisions or interference
Figure 3: Congestion control algorithm probability, the protocol does not use all network vehicles, but
it uses a set of good relays created by only some vehicles.

486
 • Kind of messages that the algorithm is concerned
V. COMPARISON OF THESE TECHNIQUES (ED Messages or Periodic Messages)
• The Improved parameters
In this section, we present a comparison of the congestion • The performance of the technique compared with
control techniques discussed above. Tables 1 and 2 represent a the reference model (802.11p)
summary of our study. For each technique we present the • The weaknesses
following criteria:
• The category of the algorithm

Parameters
Types of
Algorithm Approach ED Msg. PER Msg. Results Weaknesses
messages
throughput throughput
Media
Periodic Not Throughput increased, Design only to lead with
VC-MAC access Improved
Messages improved and collision reduced broadcast scenario.
control
Both
Periodic
Media High reliability of ED
Messages Not suitable for
MP-MAC access Improved Improved Msg., and collision
and Event Multi-hop communication.
control reduced
Driven
Messages

Table1: Comparison of Media Access Control schemes

Parameters
Types of Periodic
Algorithm Approach ED Msg. PER Msg. Queue Results Weaknesses
messages messages
throughput throughput threshold
size
Distributed Number of
Rate Not taken Not taken vehicles that
Event Scenario very
Control Rate Not into into receive ED Msg.
Driven Improved rare. Performance
Algorithm adaptation improved considera considerat sent by a node
Messages is not improved
for tion ion (by bursts) is
VANETs higher
The performance
Taken Not taken
is insignificant
Rate Periodic into into A decrease in
PEBCCAV Improved Improved when the number
adaptation Messages considera considerat CBT throughput
of intersections is
tion ion
small
Both
Congestion Queue threshold
Periodic Not taken Warning delay
Control Taken into is static. ED Msg.
Rate Messages Not Not into (delay of ED
Algorithm considerat are totally
adaptation and Event improved improved considera Msg.) is greatly
In ion prioritized on
Driven tion lesser
VANETs PER Msg.
Messages

Table2: Comparison of Rate Adaptation schemes

487
 [2] Drigo, M., Festag, A., Baldessari, R., & Zorzi, M. (2009).
For the first class algorithms, the second scheme is the one
Distributed Rate Control Algorithm for VANETs
that improves the most the performance of the network in
( DRCV ) Categories and Subject Descriptors. In
most of cases, except for road networks with small
Proceedings of 6th ACM Workshop on VehiculAr
intersections, in which case, the performance is insignificant.
InterNETworking (VANET) (pp. 3–4). Beijing, China.
For the second class algorithms, the first scheme is the one [3] Le, L., Baldessari, R., Salvador, P., Festag, A., & Zhang,
that improves the performance of the network in most of cases. W. (2011). Performance Evaluation of Beacon
In fact, this scheme reduces message collision and provides Congestion Control Algorithms for VANETs. In The
high reliability of ED messages. IEEE Global Communications Conference (p. 6). Texas,
USA.
VI. CONCLUSION [4] Darus, M. Y., & Bakar, K. A. (2013). Congestion control
algorithm in VANETs. World Applied Sciences Journal,
Congestion control for VANETs is a challenge and a very 21(7),1057–1061.
intensive research topic. Throughout this paper we surveyed http://doi.org/10.5829/idosi.wasj.2013.21.7.242.
the most important congestion control techniques proposed to [5] Shao, C., Leng, S., Zhang, Y., & Fu, H. (2014). A multi-
reduce congestion on VANETs, while we proposed a priority supported medium access control in Vehicular
categorization of these techniques into three classes: Rate Ad Hoc Networks, 39, 11–13.
Adaptation, Media Access Control and Trajectory based [6] Zhang, J., Member, S., Zhang, Q., Member, S., Jia, W.,
schemes. & Member, S. (2009). VC-MAC : A Cooperative MAC
Throughout this study, we also identified a number of Protocol in Vehicular Networks. IEEE
parameters that must be considered when elaborating a TRANSACTIONS ON VEHICULAR TECHNOLOGY,
congestion control scheme. These parameters are grouped into 58(3), 1561–1571.
two classes: The class of observable parameters that provide [7] Fracchia, R., Meo, M., & Rossi, D. (2006). VANETs :
information about the state of the network, and the control To Beacon or not to Beacon ? In 1st IEEE Workshop on
parameters to control the network congestion. Automotive Networking and Applications (p. 9). San
Throughout this survey, three main research axes emerged. Francisco, USA.
The DRCV algorithm varies the packet rate of periodic [8] Aslam, B., Amjad, F., & Zou, C. C. (2013). PMTR :
messages between two static values, while it would be better Privacy-enhancing Multilayer Trajectory-based Routing
to manage the two thresholds values dynamically and Protocol for Vehicular ad hoc Networks. In 2013 IEEE
independently by each node using a learning method. The Military Communications Conference PMTR: (p. 6). San
Congestion Control Algorithm in VANETs presented in [4] Diego, USA. http://doi.org/10.1109/MILCOM.2013.154
showed a reduction of ED messages by fully prioritizing them [9] Lin, W., Li, M., Lan, K., & Hsu, C. (n.d.).
over periodic messages (because it remains unknown while it Communication: a measurement study, (1), 1–13.
stops transmitting Periodic Messages). It would be more [10] Labiod, H., Ababneh, N., & García de la Fuente, M.
interesting to use a weighted priority between periodic (2010). An Efficient Scalable Trajectory Based
messages and ED messages with a dynamic queue threshold. Forwarding Scheme for VANETs. In The IEEE
The third issue is to merge these approaches according to International Conference on Advanced Information
the network state, for example using a weighted priority of ED Networking and Applications (pp. 1–7). Perth, Australia.
messages over periodic messages for a period of time, [11] Yusof, M., Darus, B., & Bakar, K. A. (2011).
thereafter, restoring default priorities followed by a rate Congestion Control Framework for Disseminating
adaptation of periodic messages. Safety Messages in Vehicular Ad-Hoc Networks
( VANETs ). International Journal of Digital Content
Technology and Its Applications, 5(2), 173–180.
http://doi.org/10.4156/jdcta.vol5.
[12] M. van Eenennaam, A. van de Venis, & G. Karagiannis.
(2012). Impact of IEEE 1609 . 4 channel switching on
REFERENCES the IEEE. In 2012 IFIP Wireless Days (WD) (p. 8).
Dublin, Ireland.
[1] Aslam, B., Amjad, F., & Zou, C. C. (2013). A [13] Wischhof, L., & Rohling, H. (2005). Congestion Control
Probability Based MAC Channel Congestion Control in Vehicular Ad Hoc Networks. In 2005 IEEE
Mechanism for VANET. In 2013 IEEE 77th Vehicular International Conference on Vehicular Electronics and
Technology Conference: VTC2013-Spring (p. 5). Safety (pp. 1–6). Xian, China.
Dresden, Germany.

488