Content uploaded by Patricia Ruiz
Author content
All content in this area was uploaded by Patricia Ruiz on Jan 23, 2018
Content may be subject to copyright.
X
Survey on Broadcast Algorithms for Mobile Ad hoc Networks
Patricia Ruiz, University of Luxembourg
Pascal Bouvry, University of Luxembourg
Networking at any time and any place paves the way for a large number of possible applications in ad hoc
networks, from disaster relief in remote areas to network extension. Thus, for the last decades, many works
have been proposed trying to make them become a reality. The importance of broadcasting in networking and
the broadcast nature of the wireless medium have encouraged researchers to join their efforts on designing
efficient dissemination algorithms for MANETs. The many different challenges that mobile ad hoc networks
face, e.g., limited network resources, network partitions or energy restrictions, gave rise to many different
approaches trying to overcome one or more of those problems. Therefore, literature reveals a huge variety of
techniques that have been proposed for efficient message dissemination. In this paper, we make an in-depth
review of the existing state-of-the-art techniques, as well as, propose a new taxonomy that provides a global
overview of the most relevant existing techniques.
1. INTRODUCTION
Thanks to technological advances, communicating with a remote person at any time
and any place is a reality and not just a dream. The wireless technology had a major
role in the process by enabling wireless device connections. The miniaturization and
the cost reduction also played major roles in the success of the phenomenon. Indeed,
wireless communications reached a mass-market.
Nowadays, fixed infrastructures establish and manage the majority of communica-
tions between any pair of close or distant nodes. However, researchers envision the
possibility of device-to-device communications without the need of any infrastructure.
This type of networks is known as ad hoc networks.
The term ad hoc is widely used in the scientific community. There are two different
meanings in the American heritage dictionary of English language: (1) ”form for or
concerned with one specific purpose”; (2) ”improvised and often impromptu” [Heritage
2014]. The combination of both of them perfectly describes the objective of this new
type of wireless networks, the ad hoc networks.
Ad hoc networks are mostly created on the fly; i.e., when devices with communica-
tion capabilities meet. There is no need of any infrastructure that allows the com-
munication between devices. The intrinsic characteristics of such networks, as the
self-organisation, heterogeneity of devices, dynamism or the unpredictable behaviour,
bring challenges in their creation with a good performance. Broadcasting is a corner-
stone in networking used by many other protocols and applications (e.g., route discov-
ery, network information update or neighborhood discovery). Thus, researchers have
put much effort on designing efficient broadcast algorithms for MANETs. Not only the
inherent problems of dissemination algorithms must be considered, but also the ones
Author’s addresses: Patricia Ruiz, (Current address) Imatrics Image Technologies, Madrid, Spain; Pascal
Bouvry, Faculty of Science, Technology and Communication, University of Luxembourg, Luxembourg.
Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted
without fee provided that copies are not made or distributed for profit or commercial advantage and that
copies show this notice on the first page or initial screen of a display along with the full citation. Copyrights
for components of this work owned by others than ACM must be honored. Abstracting with credit is per-
mitted. To copy otherwise, to republish, to post on servers, to redistribute to lists, or to use any component
of this work in other works requires prior specific permission and/or a fee. Permissions may be requested
from Publications Dept., ACM, Inc., 2 Penn Plaza, Suite 701, New York, NY 10121-0701 USA, fax +1 (212)
869-0481, or permissions@acm.org.
c
2015 ACM 0360-0300/2015/06-ARTX $15.00
DOI:http://dx.doi.org/10.1145/0000000.0000000
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:2 P. Ruiz and P. Bouvry
inherited from the ad hoc networks, what makes efficient communication not only cru-
cial but also a very complex task.
The related literature in broadcasting techniques for mobile ad hoc networks is enor-
mous. Therefore, a survey that compiles all existing techniques and approaches in an
appropriate way facilitating the familiarization of researchers with the topic is of ex-
treme importance. We can find some surveys on broadcasting protocols in [Vollset and
Ezhilchelvan 2003; Stojmenovi´
c and Wu 2004; Boukerche 2005; Lipman et al. 2009;
Zeng et al. 2009; Garbinato et al. 2010; Mkwawa and Kouvatsos 2011; Koscielnik and
Stepien 2011]. However, none of them presents a complete state of the art covering all
the existing approaches in a clear way. There also exist some relevant classifications in
the field [Williams and Camp 2002; Wu and Lou 2003a; Yi et al. 2003; Stojmenovi ´
c and
Wu 2004] but none of them thoroughly covers all the proposed techniques. The great-
est focus of this paper is to enlighten the reader about what has been done and how it
was done in the field of broadcasting in an organized way. Indeed, in order to achieve
this, we are not only presenting an extensive literature review but also providing a
new classification that better reflects the existing state of the art. Therefore, in this
paper, in order to give researchers a global overview of the topic, we first briefly men-
tion the most relevant existing classifications, then introduce the proposed taxonomy.
We later review the most noticeable broadcasting protocols summarizing their main
characteristics in tabular forms, and finally present some of the most interesting open
problems in the field.
The paper is organized as follows. The next section introduces the broadcast prob-
lem, reviews the existing classifications for broadcasting and proposes a new taxon-
omy. Sections 3 and 4 provide a thorough literature review in the domain. Later in
Section 5, the open issues in the topic are presented and finally, Section 6 presents our
main conclusions.
2. BROADCAST ALGORITHMS
It is important to properly define the broadcast problem before explaining the mecha-
nisms proposed to approach it.
An ad hoc network can be represented as a graph G= (V, E ), where Vis the set of
nodes composing it, and Eis the set of links connecting nodes in range.
Definition 2.1 (Broadcast).
Given a source node vstarting a broadcast process, the broadcast operation mode
sends the message mto a subset V′⊆Vsuch that for all u∈V′there is a (u, v)∈E.
According to [Basagni et al. 2004], in a wireless network, a broadcast is an operation
whereby the message is sent to all neighboring nodes.
Definition 2.2 (Network-wide broadcast problem).
Given a source node vstarting a broadcast process, the network-wide broadcast prob-
lem aims at reaching any u∈Vwith m.
In [Williams and Camp 2002], the network-wide broadcast problem was defined as
an operation where a single node sends a message to every other node in the network.
The different existing techniques for solving the network-wide broadcast problem
are called broadcast algorithms. From now on, we will refer to network-wide broadcast
by ‘broadcasting’ in the reminder of this paper.
For disseminating a message in multi-hop networks, the nodes must forward packets
that are not intended for themselves and act as routers. A straightforward approach
for solving the network-wide broadcast problem consists of retransmitting every re-
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:3
ceived packet. This is known as flooding. Its main drawback is the broadcast storm
problem [Ni et al. 1999], i.e. the higher the network density, the higher the number of
collisions, packet losses, contention of the shared medium, network traffic and network
resources. Such approach is not advisable as it not only leads to network congestions
but also drains the battery of mobile devices. Boukerche showed in [Boukerche and
Sheetal 2003], that even a routing failure causes less problems than the inherent over-
head introduced by blind flooding. However, in high loss rates and sparse networks the
obtained coverage is low even for the flooding approach [Boukerche 2008]. Therefore,
efficient broadcasting algorithms overcoming these challenges are really needed and
still non covered.
Addressing broadcast in mobile ad hoc networks needs a complete different per-
spective with respect to conventional networks. Wireless networks suffer from unique
problems like low throughput, dead spots, inadequate support for mobility, high bit
error rate and varying characteristics over short timescales. Additionally, MANETs
have a huge variety of specific characteristics that differentiate them from any other
instances in the networking paradigm. Literature reveals many different works trying
to overcome those problems and presenting different approaches for efficient dissemi-
nation in MANETs.
Before providing a systematic literature review, we briefly present some of the main
characteristics that will be used for algorithm classification and we also review the
existing classifications. We classify the literature in terms of several features, includ-
ing the existence of a central management node, the network information, the use of
random variables in the algorithm or the location of the forwarding decision.
—Centralized & decentralized systems
In a centralized system there must be a central unit that decides on behalf of the
whole system. It can take decisions in terms of its own information or by also consid-
ering information obtained from different network nodes. Significant coordination,
overhead as well as delays are associated to this architecture. Moreover, the whole
system fails if the central unit fails.
On the contrary, in a decentralized system, nodes can take local decisions and modify
their behavior using only their own information.
Considering a central unit or global knowledge when emulating an ad hoc network
is contrary to its distributed nature.
—Global or local knowledge
An algorithm is said to use global knowledge if the decision making requires infor-
mation of the whole network (e.g. all nodes’ positions).
On the contrary, algorithms using local knowledge only use locally-obtained data.
This includes not only information about the node itself but also from the node’s
neighbors (either eavesdropping, or using beacons).
Unless the knowledge is acquired beforehand (some specific cases in sensor net-
works for example), if there is no central unit, nodes need to exchange and collect
the information from all other nodes in order to gain global knowledge. In very
small networks, that could be achieved using beaconing, but as density grows this
mechanism is not scalable and becomes unrealistic. Thus, in most cases related to
MANETs, nodes are very unlikely to gain global knowledge.
—Deterministic & stochastic process
It is also possible to differentiate algorithms in terms of its predictability: determin-
istic and stochastic approaches.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:4 P. Ruiz and P. Bouvry
On the one hand, a process is considered deterministic when no random decisions
are taken. That is, for given a particular input, the corresponding output is always
the same. Its behavior is predictable.
On the other hand, when there are random choices a process is said to be stochastic.
Two executions of the same process in the same conditions can give different results.
—source-dependent & source-independent techniques
In the context of the broadcasting process, it is possible to distinguish between
source-dependent and source-independent techniques. In the former, the broadcast-
ing strategy depends on source node, it decides the next forwarding nodes from the
1-hop neighbors. While in the latter, the source independent technique, the forward-
ing decision is taken by the receiving node.
Next, we first briefly present some existing classifications and then propose a new
taxonomy.
2.1. Existing Classifications
Williams and Camp classified the broadcasting algorithms into four different fam-
ilies: ”simple flooding, probability-based methods, area-based methods and neighbor-
knowledge methods” [Williams and Camp 2002] .
(1) Simple Flooding is a method for which a source node starts the dissemination pro-
cess and all nodes rebroadcast the message exactly once.
(2) Probability-based methods forward the message with a predefined probability.
(3) Area-based Methods forward the message considering the additional covered area.
(4) Neighbor-knowledge methods use neighbors’ information for deciding whether to
rebroadcast the message or not.
Stojmenovic and Wu proposed another classification using the following charac-
teristics: ”determinism, network information, reliability, hello message content, and
broadcast message content” [Stojmenovi´
c and Wu 2004].
(1) Determinism: whether decisions feature some random aspects or not.
(2) Network information: global or local knowledge. A more detailed classification of
this specific case was proposed in [Wu and Lou 2003a], where four different types of
information are available: global,quasi-global,quasi-local and local information.
(3) Reliability: full coverage guarantee.
(4) Hello message content: includes additional information in the hello message for a
better performance.
(5) Broadcast message content: includes additional information in the broadcast mes-
sage for a better performance.
Additionally, Yi et al. proposed yet another classification: heuristic-based protocols
and topology-based protocols [Yi et al. 2003].
(1) Heuristic-based protocols group the first three categories of Williams and Camp:
”probabilistic, counter-based, distance-based and location-based algorithms”.
(2) Topology-based protocols exploit topological information:
—neighbor topology-based protocols that make the forwarding decision in terms
of 1 or 2-hop neighborhood information;
—source-tree-based protocols construct a tree in the network, for which the source
node is the root;
—cluster-based protocols group nodes into clusters and elect a representative, aka
clusterhead.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:5
Broadcasting Algorithms
No Underlying Topology Underlying Topology
Variable Tx Range Variable Tx Range Fixed Tx Range
Tree CDS Clustering Tree CDS Clustering
Fixed Tx Range
Context
Aware
Context
Oblivious
Probabilistic Counter
Based
Area
Based
Neighbour
Designation
Self-pruning
Neighbour
Knowledge
Fig. 1. Taxonomy of the broadcasting algorithms
Although the already mentioned classifications reflect many existing techniques and
strategies used in broadcasting, none of them completely covers the literature. There-
fore, in this work we propose a new classification that better reflects the current state
of the art.
For overcoming the broadcast storm problem, there are many approaches that create
or assume the existence of a virtual underlying topology in the network, and use this
information to smartly forward the message. However, many other works considers
that the expenses of creating and maintaining this virtual topology are too high, or
sometimes, creating the structure is simply not feasible. Thus, there are mainly two
different approaches; i.e., using or not an underlying topology.
Ad hoc networks are energy constrained networks, thus, one of the main goals of all
the proposed protocols is to reduce the energy consumption. Literature reveals mainly
two different approaches for reducing the energy expenditure when disseminating a
message in the network: (1) trying to reduce the number of rebroadcasts and network
resources, and (2) reducing the transmission range. So, our classification also relies on
whether the transmission power is fixed or adjustable. This variable range is found
in both cases; i.e., using underlying topology and not using it. Algorithms that do not
use any topology but use a fixed transmission range are divided into: (1) context obliv-
ious, (2) context aware, and (3) neighbor knowledge. Additionally, for those relying on
topologies, three different possible classes are considered: (1) connected dominating
sets, (2) tree-based topology, and (3) cluster-based topology. The proposed taxonomy is
illustrated in Figure 1.
There also exist some techniques that deal with specific networks like the delay
tolerant networks [Fall 2003]. Those networks suffer from high latency and continuous
network disruption, thus, devices usually communicate using a store-carry-forward
mechanism. This is a very particular type of dissemination that is not addressed in this
paper. However, interested readers can refer to the Haggle Project for a very interesting
work dealing with content dissemination [hag ] (nodes can carry information for other
nodes), or to some other seminal works tackling the problem [Hui et al. 2011; Bastani
et al. 2012].
Similarly, we can find some approaches that limit the dissemination of a message
to a specific area of the network; i.e. not the complete network: geocast-based routing
algorithms. They mainly use GPS information to locate and limit the targeted area,
use routing mechanisms to get to it, and then broadcast the information within pre-
defined boundaries. This can be seen as a multicast approach more than a broadcast
approach because the message is not intended for every node in the network. This very
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:6 P. Ruiz and P. Bouvry
specific broadcast technique is not covered in this paper. However, interested readers
can refer to [Maihofer 2004; Al-Kanj and Dawy 2011; Allal and Boudjit 2012].
In the following, we review the literature according to the previously introduced
classification. Unless explicitly mentioned, the presented approaches are deterministic
and use local knowledge.
3. ALGORITHMS NOT USING AN UNDERLYING TOPOLOGY
Depending on the application, the benefits of having a virtual structure over the real
network topology is not always worth the cost of its creation and maintenance. In
those cases, for performing an efficient broadcast, the nodes’ forwarding decisions is
enhanced by some information about the current situation (context aware) or by some
knowledge of their neighbors and their related strategies (neighbor knowledge). Nev-
ertheless, there also exist context-oblivious approaches that rely on probabilistic for-
warding decision.
Next, we present the most relevant algorithms that do not assume the existence of
any underlying topology in the network, using either a fixed transmission range or a
variable one.
3.1. Fixed Transmission Range Protocols
As above mentioned, we differentiate algorithms that do not rely on any topological
information and use a fixed transmission range in terms of the knowledge used for
the forwarding decision. Most relevant works in broadcast algorithms can be classified
as: context oblivious, context aware and neighbor knowledge. Next, we review them in
detail.
3.1.1. Context oblivious.Nodes do not periodically exchange information with other
nodes, or eavesdrop the channel in order to make intelligent forwarding. They simply
forward the message with a predefined probability.
In the pure probabilistic approach, nodes do not consider the environment nor the
current network situation. The forwarding decision is made according to a predefined
probability. However, this probability can depend on the node behavior or the neigh-
boring devices’ behaviour. Thus, we can find both context-oblivious and context-aware
probabilistic approaches.
Flooding or blind flooding [Ni et al. 1999] consists of retransmitting the message
the first time it is received and ignoring any other copy of the message. No network in-
formation or neighbor knowledge is required. However, the shortcomings associated to
this algorithm are the high number of collisions, redundancy, and waste of network re-
sources, specially in high density networks. Additionally, this approach does not handle
network partitions or temporary disconnections, i.e. the message only reaches nodes
within a partition or connected component. This is a deterministic approach.
Probabilistic flooding [Ni et al. 1999] reduces the number of forwarding nodes by
assigning a probability to every node receiving a message. When the probability is 1,
the algorithm behaves as blind flooding.
In order to deal with partitions and/or high mobility hyper-flooding was pro-
posed [Viswanath and Obraczka 2002]. Whenever a new neighbor is met, the mes-
sage is retransmitted so that reliability might increase. However, in dense networks,
the reliability could even decrease due to collisions and failures. Hyper-flooding is also
deterministic.
3.1.2. Context aware approaches.In the context-aware approaches, smart forward-
ing decisions are taken using knowledge about the current network state. We differen-
tiate between: (1) probabilistic-based, (2) counter-based (aware of the network traffic),
and (3) area-based (aware of the network density).
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:7
aProbabilistic
The ”weighted p-persistence, slotted 1-persistence, and slotted p-persistence proba-
bilistic and timer-based algorithms” were introduced in [Wisitpongphan et al. 2007]
for vehicular ad hoc networks, VANETs. The three protocols need location informa-
tion (either by GPS or signal strength) for calculating the relative distance to the
source node. In weighted p-persistence, far-away nodes have higher probability of
retransmission. In slotted 1-persistence, the time slot assigned for retransmitting
depends on the distance to the source node, thus, far-away nodes retransmit sooner.
The slotted p-persistence is a mix of the two first ones, the time slot assigned for re-
transmission depends on the distance, but a node will retransmit with a probability
pthat also depends on the distance.
A slightly different approach to the weighted p-persistence protocol is proposed
in [Zhou et al. 2010], the nth power p-persistent broadcasting protocol that efficiently
disseminates messages in dense vehicular ad hoc networks. The main difference re-
sides in the estimation of the probability that is elevated to the nth power. Both,
flooding and weighted p-persistence are particular cases of nth power p-persistent,
with n=0 and n=1, respectively. Authors claim that the correct value of ncan out-
perform both of them, with less delay, fewer hops, lower load and higher throughput.
However, explaining how to obtain this optimal value of nis a complex task that is
not specified in the work.
Another probabilistic approach for vehicular networks is proposed by Slavik
in [Slavik and Mahgoub 2010]. This method focuses on privacy, scalability and
anonymity, because it assumes that drivers are reluctant to share their movements
or maneuvres with anyone. Two schemes are proposed. In the first one, vehicles re-
transmit with a uniform and constant probability. The second one depends on an
estimation of the distance between the receiver and the furthest node.
The speed adaptive probabilistic flooding algorithm (SAPF) [Mylonas et al. 2008],
calculates the adaptive forwarding probability in terms of the vehicle velocity. SAPF
focuses on reducing the delays caused by contention in dense networks. The main
target consists of finding the optimal value of the forwarding probability based on
the vehicle speed.
In the gossip-based broadcast approach, the source node randomly selects kneigh-
bors and sends the broadcast message to them. Any node does the same upon re-
ception of the message for the first time [Kermarrec et al. 2003]. A survey on gossip
protocols can be found in [Leitao et al. 2010].
Probabilistic schemes are generally suitable for MANETs due to following charac-
teristics: low overhead (no infrastructure has to be created and maintained), flex-
ibility to topological changes, resilience to failures and fair energy consumption.
However, reliability is not guaranteed nor the proper functioning of the protocol. Ad-
ditionally, SAPF is suitable for VANETs where cars drive at high speeds, but is not
suitable at all for mobile networks composed of people walking at low speeds [Ruiz
et al. 2012]. A review of probabilistic broadcasting protocols and the most commonly
used performance metrics is included in [Reina et al. 2015].
bCounter-based approach
The stochastic counter-based approach [Ni et al. 1999] refrains the node from re-
broadcasting in case multiple message copies have been received. Upon reception
of the broadcast message, the node waits for a random time before rebroadcasting.
If during this waiting time several copies of the same message are received, the re-
broadcasting is cancelled. In this technique, a node is aware of the network traffic
conditions and uses this information in the forwarding decision. All the counter-
based approaches reviewed next are stochastic.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:8 P. Ruiz and P. Bouvry
A low value for the counter threshold (the maximum number of repeated copies
allowed) provides high saving in retransmission, however reachability highly de-
grades in sparse networks. On the contrary, a high value for this threshold implies
high reachability but low savings. This dilemma is tackled in [Tseng et al. 2002],
where an adaptive and probabilistic counter-based approach is presented. This work
introduces the counter threshold as an adaptive function that varies in terms of the
number of neighboring nodes.
The color-based broadcast algorithm is a variant of the counter-based
scheme [Keshavarz-Haddad et al. 2006]. It colors broadcast messages and all nodes
rebroadcast the message after the random timeout unless it already heard ηcolors
at that time.
Chen et al. proposed DIS RAD in [Chen et al. 2005]. This approach applies the con-
cept of distance-based into the counter-based approach. Nodes closer to the maximal
transmission range have higher probability of rebroadcasting and shorter Random
Assessment Delay, aka RAD. How nodes are aware of the distance to the source
node is not specified.
An adaptive approach that uses two different threshold values, for sparse and dense
networks was proposed in [Al-Humoud et al. 2008]. It compares the current number
of neighbors to the average, and in case it is lower, the network is considered sparse,
otherwise dense (how to calculate the average number of neighbors is not specified).
The random delay and the counter threshold value are set accordingly.
An adaptive probabilistic counter scheme is presented in [Liarokapis and Shahrabi
2009]. In ProbA, a node receiving a message sets a random delay. During this wait-
ing time the node counts the number of received repeated copies, and the forwarding
probability is set in terms of this value. An extension, the fuzzy logic probabilistic
(FLoP) is proposed in [Liarokapis and Shahrabi 2011]. The interval of the hello
message adapts to the variability of the topology. This is estimated in terms of the
difference in the number of neighbors between two consecutive hello messages.
cArea-based approaches
Two other stochastic techniques, that were introduced in [Ni et al. 1999], are
distance-based (DB) and location-based. Both approaches start a random delay
when the message is received for the first time, before rebroadcasting. In the for-
mer, the forwarding decision is taken in terms of the distance to the source node. In
the latter, the decision is taken in terms of an estimation of the potential additional
coverage achieved. In this case, GPS information is needed. Nodes are aware of the
network density by estimating the distance to the 1-hop neighbors (or the additional
expected coverage area).
In Figure 2, the distance-based approach mechanism is shown. Only nodes that are
further away than a predefined value (located in the forwarding area, i.e. light grey
zone) retransmit the message. In this figure, the candidate nodes for retransmitting
the message sent by Node A are: Node E and Node F.
Next, we mention some of the most relevant work based on those approaches.
Two adaptive versions of the DB were presented in [Chen et al. 2002], where nodes
are sorted according to the received signal strength. The first one, DAD-NUM spec-
ifies a predefined number of forwarding nodes. The second one, DAD-PER specifies
a percentage of nodes that will rebroadcast the message.
The difficulty to set up the thresholds is highlighted in [Sun and Lai 2002]. Authors
propose a broadcast algorithm based on a defer time (timeout) before resending.
This defer time is inversely proportional to the distance between the sender and
the receiver nodes. Additionally, an angle-based scheme is proposed to reduce the
number of redundant transmissions.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:9
A
B
C
D
E
F
Fig. 2. Distance-based approach
The area-based beaconless algorithm (ABBA) for 2D and 3D needs precise location
information [Ovalle-Mart´
ınez et al. 2006]. The co-ordinates of the sender are in-
cluded in the header of the broadcast packet (e.g. using a GPS). Nodes calculate
the portion of the transmission range that is not covered, and set a timeout that
is inversely proportional to this value. After the waiting time, if more copies are
received and the transmission range is fully covered, the node cancels the retrans-
mission. Two different approaches for the timeout are proposed; (1) random value
between [0 1] seconds (stochastic approach), and (2) proportional to the already cov-
ered perimeter (deterministic).
Similar beaconless approaches that include the location of the source node in the
header of the message are the optimized broadcast protocol [Durresi et al. 2005] and
the geoflood [Arango et al. 2004]. The former divides the network in a honeycomb.
The receiver calculates the distance to the 2 closest vertices of the honeycomb, and
starts a delay that is proportional to the shortest distance. The latter divides the
transmission area in a cartesian plane. A timeout that is inversely proportional to
the distance is set when receiving the message. As soon as a message is received
from each of the four quadrants, the retransmission is cancelled. Both approaches
are deterministic.
Similar approaches are the six-shot broadcast [Garbinato et al. 2008] and the opti-
mized flooding protocol [Paruchuri et al. 2003] that considers strategic forwarding
positions.
The probabilistic and the distance scheme are combined in [Cartigny and Simplot
2003] to design a stochastic broadcasting algorithm. The forwarding probability de-
pends on the local density of the network and the distance to the source node. How-
ever, no positioning information is used and only the 1-hop neighbors of the sender
are included in the header of the broadcast message. Additionally, the neighbor
elimination scheme is used, i.e. when no new neighbor is expected to be covered, the
node cancels the retransmission.
In [CAO et al. 2007], an enhanced distance-based approach is presented including
the counter-based and the border-aware approaches. The distance to neighboring
nodes is calculated using the free space wireless propagation model. This stochastic
approach also considers the remaining battery level of the nodes.
An adaptive location-based algorithm that also uses 1-hop information was pro-
posed in [Tseng et al. 2002]. The threshold of the covered area is not a fixed value
but an adaptive function that adapts to the neighborhood size.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:10 P. Ruiz and P. Bouvry
The efficient directional broadcast was introduced in [Li et al. 2007]. It uses direc-
tional antennas and repeaters located at intersections for disseminating the infor-
mation. Only the furthest node disseminates the message backwards (i.e., in op-
posite direction to the reception). In the same work a probabilistic scheme is pre-
sented. These protocols are intended for vehicular ad hoc networks and assume the
existence of a network infrastructure.
Transmission failure is mitigated in [Huang et al. 2009] proposing a cooperative
and opportunistic approach for collision warning in VANETs. Information about the
motion is exchanged and the node’s and neighbor’s risk of collision are calculated.
Such that, if a danger ahead is detected and the neighbors do not warn, the node
can broadcast a message with motion information related to the cars involved in the
danger.
In [Gang et al. 2012], a directional distance-based broadcasting algorithm that uses
percolation theory to select the direction of forwarding is presented. It uses direc-
tional antennas for estimating the direction of the source node and the additional
coverage area. The value of this additional coverage area influences the forwarding
probability. Therefore, it is a stochastic model where no beacons are exchanged.
A distance-based algorithm that takes into account the remaining battery level of
devices is proposed in [Kasamatsu et al. 2009]. The broadcasting method consid-
ering battery and distance (BMBD) sets a timeout that is inversely proportional to
the distance and the battery level, i.e. nodes with higher remaining battery and
higher distance from the source node set a lower delay. If a copy of the message
is heard while waiting, the retransmission is cancelled. This method assumes GPS
information available at nodes.
Three different stochastic approaches are presented in [Kokuti and Simon 2012].
The first one, distance-based handshake gossiping (DBHG) is a distance-based pro-
tocol where the source node indicates the forwarding probability of neighboring
nodes in terms of their distance to itself. The second one valency-based handshake
gossiping (VBHG), is adding the nodes’ degree knowledge to the first approach. Fi-
nally, the average valency-based handshake gossiping considers also the past deci-
sions for calculating the forwarding probability.
In [Slavik et al. 2011], the distance-to-mean broadcast method is introduced. The
first time a node receives a message, a delay inversely proportional to the distance
to the source node is set. When the timeout expires, it calculates the spatial mean of
all the neighbors it received the message from, and calculates its distance-to-mean.
If that distance is higher than a predefined threshold, the message is retransmitted.
It requires positional information.
A stateless distance and probability-based approach that does not need to exchange
hello messages is proposed in [Banerjee et al. 2012]. Nodes within 1-hop rebroad-
cast in terms of the distance to the source. The retransmission probability of nodes
located hhops away is calculated and depends on the distance (hops) and network
density.
A distance-based beaconless algorithm DibA is proposed in [Liarokapis et al. 2009].
The distance threshold adapts to the number of retransmissions heard of a specific
packet. It is a combination of distance-based and counter-based approaches.
Many of the studied works include a RAD before retransmitting the broadcast mes-
sage. This technique highly reduces the number of collisions and packet losses as
well as allows enough time to receive multiples copies (to decide whether to rebroad-
cast), but also increases the delay in the broadcasting process. The maximum value
of the RAD highly influences the broadcast time. Indeed, many approaches including
counter-based, area-based, or RAD-based ones rely on thresholds for decision making,
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:11
but finding the appropriate value that fits rapidly evolving mobile ad hoc networks is
a very complex task that highly influences the performance of the protocols.
In Table I, a summary containing the surveyed works that use a fixed transmission
range, and that are not topology-based is presented.
In the next tables, we have classified the protocols in terms of Neigh., Hops, Determ.
GPS and Forward decision, that respectively stand for neighbor knowledge, number of
hops, determinism, need of GPS, and forward decision.
Table I. Classification of non-topology-based algorithms with fixed transmission range: context aware and oblivious
approaches
Neigh. Hops Determ. GPS Forward decision
Simple flooding [Ni et al. 1999] local – yes no source independent
Hyper-flooding [Viswanath and Obraczka 2002]
Probabilistic flooding [Ni et al. 1999]
local – no no source independent
SAPF [Mylonas et al. 2008]
[Banerjee et al. 2012]
[Gang et al. 2012]
DibA [Liarokapis et al. 2009]
[Slavik and Mahgoub 2010]
weighted/slotted 1/p -persistent [Wisitpongphan et al. 2007]
local 1-hop no yes source independent
Rewarn[Huang et al. 2009]
Distance-to-mean [Slavik et al. 2011]
Defer time [Sun and Lai 2002]
gossip-based broadcast [Kermarrec et al. 2003]
local 1-hop no no source independent
nth power p-persistent[Zhou et al. 2010]
Adaptive counter [Tseng et al. 2002]
color-based [Keshavarz-Haddad et al. 2006]
DIS RAD [Chen et al. 2005]
[Al-Humoud et al. 2008]
ProbA [Liarokapis and Shahrabi 2009]
fuzzy logic probabilistic [Liarokapis and Shahrabi 2011]
DBHG/VBHG/AVBHG [Kokuti and Simon 2012]
DAD-NUM/PER [Chen et al. 2002] local 1-hop yes no source independent
BMBD [Kasamatsu et al. 2009]
optimized broadcast protocol [Durresi et al. 2005]
local – yes yes source independent
geoflood [Arango et al. 2004]
six-shot broadcast [Garbinato et al. 2008]
optimized flooding protocol [Paruchuri et al. 2003]
[Cartigny and Simplot 2003]
local 1-hop yes yes source independent[CAO et al. 2007]
[Tseng et al. 2002]
[Gang et al. 2012] local 1-hop yes/no yes source independent
ABBA [Ovalle-Mart´ınez et al. 2006] local – yes/no yes source independent
3.1.3. Neighbor-knowledge approaches.By exchanging hello messages, nodes are
aware of the neighbors within their transmission range. It is possible to obtain 2-hop
neighbors’ information by including in the beacon the list of 1-hop neighbors. The 2-
hop neighbors knowledge gives more topology information, but at the same time, in
highly mobile networks, it is more difficult to have accurate and up-to-date informa-
tion. There are many works that use neighbor knowledge for making smart forwarding
decisions.
Flooding-based on 1-hop neighbors information and adaptive holding (FONIAH) was
presented in [Lee and Ko 2006]. This approach combines neighbor-knowledge-based
and area-based flooding. When receiving a message, a node can hold it for some time
before rebroadcasting. This timeout is inversely proportional to the distance to the
source node. The source node includes in the broadcast message its position and the
distance to the furthest neighbor. If all 1-hop neighbors already received the broadcast
message, the retransmission is cancelled.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:12 P. Ruiz and P. Bouvry
Tonguz et al. present DV-Cast in [Tonguz et al. 2007; Tonguz et al. 2010], a broadcast
protocol that only requires location information obtained using GPS for supporting a
wide range of traffic conditions. Both the broadcast storm problem and network dis-
ruptions are handled in a distributed fashion. DV-Cast specifies the region of interest
(ROI) and disseminates the message towards that direction. The forwarding decision is
taken in terms of the list of 1-hop neighbors and their relative distance. This approach
is intended for alert messages in vehicular networks, such that the broadcast message
is disseminated backwards.
Within the neighbor-knowledge-based approaches, we can also clearly differentiate
between two techniques: self-pruning and neighbor designating approaches. In the for-
mer, the potential forwarding node considers whether it should rebroadcast. In the
latter, the source node decides the forwarding neighbors.
aSelf-pruning
In flooding with self pruning , the header of the dissemination message includes the
list of 1-hop neighbor nodes. If additional neighbors are covered with the retrans-
mission, the node schedules the rebroadcast in RAD seconds. In the meantime, it
calculates the number of additional neighbors with redundant copies. After a RAD,
the message is either sent or discarded [Lim and Kim 2000].
The stochastic neighbor-coverage scheme is proposed in [Tseng et al. 2002]. It uses 2-
hop neighbors information to know if the broadcast message was not received by any
1-hop neighbors. A list with the 1-hop neighbors is created and the random delay is
set the first time the broadcast message is received. All nodes that are potentially
receiving the message are eliminated from this list. If all the 1-hop nodes did receive
the packet, the retransmission is interrupted.
The scalable broadcast algorithm (SBA) that uses 2-hop neighbors knowledge is
presented in [Peng and Lu 2000]. It is similar to flooding with self pruning but the
information is included in the beacons not in the packet header. They also proposed
a RAD that dynamically adjusts to the network conditions. The lower limit of the
RAD interval is 0, and the upper one is proportional to the node maximum degree
divided by the number of neighbors it actually has. See Equation 1:
RAD :r∈[0, α ∗max degree
num neighbors ](1)
Delayed flooding with cumulative neighborhood (DFCN) [Hogie et al. 2004] includes
the 1-hop neighbor list in the packet header. The potential coverage obtained in
case of forwarding is calculated, and if the estimation is larger than a predefined
value, the message is scheduled to be retransmitted after a RAD. Additionally, it
adapts to the local network density: when the number of 1-hop neighbor is lower
than a threshold, the predefine value is set to 1 (it rebroadcasts if there is at least
one neighbor to cover). Moreover, if the node density is considered to be low, and
therefore the forwarding probability 1, whenever a new neighbor is met, the RAD is
stopped and the message is rebroadcast. Therefore, flooding with self pruning [Lim
and Kim 2000] can be considered a special case of DFCN.
In [Khabbazian et al. 2012], a hybrid broadcasting algorithm that uses 2-hop infor-
mation combining the neighbor designating scheme with self pruning is presented.
The source node selects at most one forwarding node from the set of 1-hop neighbors.
At reception, if the node is not selected as a relay, it creates a list of neighbors, and
removes from there the neighbors of the sender and the neighbors of the selected
forwarding node. The message is dropped if the list is empty, otherwise retransmit-
ted.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:13
The lightweight and efficient network-wide broadcast (LENWB) was proposed
in [Sucec and Marsic 2000]. This approach uses 2-hop neighbors’ information, and
includes the degrees of the 1 and 2-hop neighbors in the beacon. The source node
does not select forwarding nodes, but nodes rebroadcast using the knowledge of
which of the 1 and 2-hop neighbors are expected to forward the message. Nodes
have priority according to their degrees, i.e., nodes with more neighbors have higher
priority.
bneighbor designating approaches
It is a source-dependent approach, i.e., the source node decides which neighbors will
retransmit the broadcast message. Different techniques for selecting the forwarding
nodes have been proposed. Next, we will review the most relevant ones.
Dominant pruning [Lim and Kim 2000] includes in the header of the packet a list
with the forwarding nodes. These nodes calculate the new forwarding list using the
Greedy set cover algorithm (neighbors of the source node are considered already
covered; choose the 1-hop neighbor that covers the most 2-hop neighbors; repeat
until all 2-hop neighbors are covered). Its a very promising approach but it does not
prevent redundant transmissions.
Total dominant pruning (TDP) and partial dominant pruning (PDP) are both pro-
posed in [Lou and Wu 2002] for enhancing the Dominant pruning by further de-
creasing the redundancy. In the former, the sender includes in the broadcast packet
the 2-hop neighborhood information, such that the receiver can consider all those
nodes as already covered, with the consequent increase in consumed bandwidth.
The latter induces that information from the 2-hop neighbors’ knowledge, without
piggybacking any information to the broadcast message. In dense networks, this
technique is less efficient than in sparse ones.
In multipoint relay (MPR) [Qayyum et al. 2000], unlike dominant pruning, the list
of relaying nodes is included in the exchanged beacons; i.e. not in the header of the
broadcast message. The selecting mechanism works as follows: first, check the 2-
hop neighbors that can only be reached by one 1-hop neighbors and select them as
MPRs. Then, from the remaining 1-hop neighbors select those that cover the most
2-hop neighbors that are not covered yet. Continue until all 2-hop neighbors are
covered. For a graphical explanation please see Figure 3. Only the blue nodes are in
charge of retransmitting the broadcast messages.
Fig. 3. Multipoint relay mechanism
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:14 P. Ruiz and P. Bouvry
The well-known Optimized Link State Routing Protocol (OLSR) routing protocol is
based on MPR. For a more detailed explanation please refer to the Internet draft
RFC 3626 [Clausen and Jacquet 2003]. A survey on multipoint relays and connected
dominating sets approaches for broadcasting is presented in [Liang 2007].
In the ad hoc broadcast protocol (AHBP) presented in [Peng and LU 2001], the
forwarding nodes are called broadcast relay gateway (BRG). The mechanism for se-
lecting the BGRs is identical to MPR. The differences are: (1) nodes are informed
about becoming a BGR in the header of the packet, not using the beacons. (2) 2-
hop neighbors knowledge is used to determine which neighbors are already cov-
ered. AHBP-EX, an extension of the protocol, is proposed in [Peng and LU 2001] to
support mobility aspects. If a message is received but did not previously exchange
beacons with the source node (i.e. due to the mobility), it assumes it is a BGR.
There are plenty of MPR-based broadcasting algorithms that modify the MPR se-
lection procedure in order to reduce the number of collisions, the cardinality of the
MPR, or to improve the energy management efficiency. We will briefly mention some
of them. In [Mans and Shrestha 2004], the authors propose different models like
the in-degree greedy set cover. It first checks the 2-hop neighbors that can only be
reached by one 1-hop neighbor and selects them as MPRs. Then, the algorithm ran-
domly picks up a node from the non-covered 2-hop nodes, among all 1-hop-neighbor
nodes that can cover this 2-hop node, and have not been selected as MPRs by the
source node S, then selects a node as an MPR that has minimum number of non-
covered 2-hop neighbors. It continues until all 2-hop neighbors are covered. The
same authors also proposed, MPR Selection with Minimum Overlapping, for reduc-
ing the traffic and the risk of collisions. As the others approaches, 1-hop neighbor-
hood nodes are first listed. And then, the node with the minimum coverage ratio is
chosen as an MPR among the 1-hop neighbors that are not selected as MPRs yet.
This is done until all 2-hop neighbors are covered. The coverage ratio is defined as
the ratio of the covered 2-hop neighbors over non-covered 2-hop neighbors of a node.
Moreover, another technique, aka weighted set cover, gives weights to nodes ac-
cording to a desired property, e.g. bandwidth, and selects the 1-hop neighbor with
the largest number of non-covered nodes over the weight ratio. The MPR Selection
with secondary priority that provides priority to the nodes with maximal number of
neighbors is also proposed. This is applied to OLSR, in the case where two nodes
have the same number of non-covered neighbors.
In [Lipman et al. 2002], the authors propose to select the MPR nodes in terms of
the forwarding utility of the 1-hop neighbors. This forwarding utility is computed
as a function of the power utility of each node (remaining battery), and the neighbor
utility (the ratio of non-covered 2-hop neighbors over all the 2-hop nodes that a 1-
hop neighbor node covers). As the previous approaches, the algorithm first checks
the 2-hop neighbors that can only be reached by one 1-hop neighbor and select them
as MPRs. Then it selects as MPR the node with highest forwarding utility value.
The process continues until all 2-hop neighbors are covered. Later in [Lipman et al.
2003], the first steps of the MRP approach were eliminated and MPR nodes are only
selected in terms of the forwarding utility value.
Additional MPR nodes are chosen in order to provide reliable broadcast in [Ahn
and Lee 2011; Clausen and Jacquet 2003]. The redundancy method proposed
in [Ahn and Lee 2011] chooses additional MPR to cover only selected 2-hop nodes.
In [Clausen and Jacquet 2003], the redundant MPRs cover all the subset of 2-hop
neighbors, reducing therefore, the cardinality of MPRs while at the same time in-
creasing the delivery ratio. The idea is that the 2-hop MPR nodes must be covered
at least m-times.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:15
The double-covered broadcast (DCB) algorithm [Lou and Wu 2004] is a source-
dependent protocol where the source node chooses the set of 1-hop neighbors that
covers all 2-hop neighbors and the non-forward 1-hop neighbors are covered by at
least to forwarding nodes. The retransmissions of the forwarding nodes are consid-
ered as acknowledgements, thus when the source does not hear from the forwarders
it will simply re-broadcast the message.
[Bai et al. 2009] introduces the vehicular multi-hop broadcasting protocol (VMP)
that disseminates the message in the opposite direction. VMP designates multiple
delays and forwarding nodes (i.e. it is source-dependent) for disseminating an alert
message in a predefined zone. In order to assure high reachability and avoid unnec-
essary retransmissions, a cooperation schema between nodes and a procedure for
detecting duplicated packets are used.
In Table II, a summary containing the surveyed works that use a fixed transmission
range, and that are not topology-based is presented.
Table II. Classification of non-topology-based and neighbor-based algorithms with fixed transmission range.
Neigh. Hops Determ. GPS Forward decision
FONIAH [Lee and Ko 2006] local 1-hop no yes source independent
DV-CAST [Tonguz et al. 2010]
VMP [Bai et al. 2009] local 1-hop yes yes source-dependent
flooding with self pruning [Lim and Kim 2000] local 1-hop no no source independent
DFCN[Hogie et al. 2004]
neighbor-coverage scheme [Tseng et al. 2002] local 2-hop no no source independent
SBA [Peng and Lu 2000]
TDP/PDP[Lou and Wu 2002] local 2-hop yes no source independent
LENWB[Sucec and Marsic 2000]
[Mans and Shrestha 2004] local 2-hop no no source-dependent
Dominant pruning[Lim and Kim 2000]
local 2-hop yes no source-dependent
multipoint relay [Qayyum et al. 2000]
AHBP [Peng and LU 2001]
[Lipman et al. 2002]
[Lipman et al. 2003]
[Ahn and Lee 2011]
[Clausen and Jacquet 2003]
double-covered broadcast [Lou and Wu 2004]
[Khabbazian et al. 2012] local 2-hop yes no source indep./dep.
3.2. Variable Transmission Range
Energy consumption in MANETs is a key problem because the network degrades with
the device batteries running down. Two different approaches are considered for de-
creasing the energy consumption in MANETs. The first one deals with reducing the
number of message retransmissions. All the above-surveyed papers use this approach.
The second approach considers reducing the transmission power. The first one can be
seen as the fixed power approach that was already surveyed, and the second one is the
variable power approach that we are reviewing next.
In the variable power approach each node can transmit using different transmission
radii; therefore, the number of reached neighbors when broadcasting a message varies
according to the transmission power. There are several works that try to find a com-
mon power level that guarantees a low node degree [Kawadia and Kumar 2005], or
to ensure that the communication graph is connected with at least k-neighbors over
a uniformly distributed network [Blough et al. 2006]. Every node has to communicate
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:16 P. Ruiz and P. Bouvry
with each other for selecting the common transmission power. Such approaches are not
scalable as the overhead increases with the size of the network. However, in [Gomez
and Campbell 2007] it was demonstrated that a variable transmission range provides
better energy savings and network capacity than the fixed transmission range. More-
over, the authors assert that there exists a best-choice transmission range that max-
imizes the nodes’ capacity in presence of mobility. A survey on conserving power by
employing transmission power control can be found in [Guo and Yang 2007].
When considering the variable transmission power of nodes, adjusting each node
with an optimal transmission radius is a key issue, which is addressed by the two fol-
lowing instances: the minimum range assignment and the minimum energy broadcast
problems.
Definition 3.1 (The minimum h range assignment problem).
”Given a finite set S of points (i.e. the stations of a radio network) on the plane and
a positive integer 1 ≤h≤ |S| − 1, the 2 dimensional minimum h range assignment
problem consists of assigning transmission ranges to the stations in order to minimize
the total power consumption provided that the transmission ranges of the stations
ensure the communication between any pair of stations in at most hhops” [Clementi
et al. 2000].
Definition 3.2 (The minimum energy broadcast problem [Cagalj et al. 2002]).
Given a graph G=(V,E), where each node i ∈V is assigned a variable node power pv
i,
we assign a link cost cij : E(G) →R+that is equal to the minimum transmission power
necessary to maintain link (i,j).
Considering a source node vthat wants to broadcast a message, the minimum energy
broadcast problem consists of finding the power assignment vector P= [pv
1,pv
2... pv
|V|]
such that it induces the directed graph G=(V,E’), where E’= (i, j) ∈E : cij ≤pv
i, in which
there is a path from vto any node of V (all nodes being covered), and such that
min X
i∈V
pv
i
As demonstrated in [Clementi et al. 2000] and in [Cagalj et al. 2002], both minimum
assignment and the minimum energy broadcast are NP-hard problems. The main dif-
ference between them, is that in the former, the source of the broadcasting algorithm
is not considered while in the latter, it is. Tree topology has been widely used to solve
these problems. This approach will be addressed in detail in Section 4.
Cartigny et al. presented the RNG broadcast oriented protocol in [Cartigny et al.
2003]. It runs locally at each node but requires knowledge of the distance to all neigh-
bors and between neighboring devices. The RNG graph stands for the Relative Neigh-
borhood Graph. It is an undirected graph in which two points are connected with a
new edge if there is no other node closer to both of them (see Figure 4, where there
cannot be any neighbor in the striped area in order to have two RNG neighbors). If
a packet is received from any RNG neighbor, it is retransmitted using the minimum
transmission range that covers the furthest RNG neighbor that did not received it yet.
Otherwise, it starts a timeout and creates a list with all the RNG neighbors that did
not receive it. If after the waiting time the list is not empty, it rebroadcasts the mes-
sage with the minimum transmission range that covers the furthest RNG neighbor in
the list. A different version where the timeout is set every time a message is received
is proposed in [Cartigny et al. 2005].
The notion of forbidden set in the relative neighborhood graph is incorporated
in [Wang et al. 2010]. This forbidden set prevents nodes with low battery levels to
act as forwarders in the dissemination process. Additionally, when redundant retrans-
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:17
Fig. 4. Relative neighborhood graph
missions are received the node will change of relative neighbor or will ask to be covered
by only one.
In [Chen et al. 2003], the power adaptive broadcasting with local information
(PABLO) is proposed. In PABLO, nodes calculate the transmission power required to
reach 2-hop neighbors. After that, the source node estimates the energy needed to
reach the furthest node directly or through an intermediary node. In case the later
is cheaper in terms of energy consumption, the source node reduces the transmission
power and thus, discards the furthest node. See Figure 5, where node A excludes node
Cfrom the 1-hop neighborhood if the sum of the power node A needs to reach node B
plus the power node B needs to reach node C is lower than the power node A needs to
directly reach node C.
AB
C
Fig. 5. Mechanism of PABLO
In [Karenos et al. 2005], an extension to PABLO is proposed that applies neigh-
borhood pruning, i.e. after performing the optimisation, the algorithm considers the
possibility of excluding the furthest node(s).
The transmission range is set in terms of the local density in [Li et al. 2003]. This
local density is estimated using an analytical model.
The inside-out power adaptive approach (INOP) is presented in [Chiganmi et al.
2008]. It uses 2-hop neighbors knowledge to obtain a good energy utilization for cov-
ering all direct neighbors. The difference with other existing approaches is that each
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:18 P. Ruiz and P. Bouvry
node first sorts the neighbors in terms of the power needed to cover them, and then,
computes the optimal energy strategy starting from the closest neighbor to cover the
next neighbor directly or indirectly. Two different versions for selecting the relay nodes
are proposed: (1) INOP with self pruning that sets a random timeout from an interval
whose range inversely depends on the number of non-covered neighbors. If after a
delay the number of neighbors that were not covered is empty the rebroadcast is can-
celled. And (2) INOP with neighbor designation considers the possibility of increasing
the transmission power of an already selected relay node instead of adding a new re-
lay for covering an non-covered neighbor. Considering that the timeout is randomly
set, this approach is stochastic. The previous approaches do not consider a limit when
decreasing the transmission range, thus, it could happen that the number of hops for
disseminating a message highly increases.
In [Woon and Yeung 2011], different versions of dominant pruning, total dominant
pruning, partial dominant pruning, INOP and PABLO using different termination con-
ditions and variable transmission power are compared.
A distance-based approach (EDB), aiming at reducing the energy consumption that
relies on a cross-layer design and a variable transmission range is introduced in [Ruiz
and Bouvry 2010b]. An adaptive version that targets energy saving in dense networks
by discarding nodes from the 1-hop neighborhood was later presented in [Ruiz and
Bouvry 2010a], AEDB. In this case, the transmission range is not decreased to any
extent but to a predefined value, such that the number of hops does not highly increase.
In [Schleich et al. 2011] community knowledge is used for improving the coverage
achieved by a message dissemination. Nodes that belong to two different communities
always rebroadcast the message although the forwarding policy prevents the retrans-
mission. This is done in order to favor a fast dissemination within different communi-
ties.
[Reumerman and Runi 2005] introduces another technique where the path loss in-
formation is included in the beacons, so that neighboring nodes can be arranged in a
list in terms of average path loss. A predefined number of nodes that must be reached
is set. Thus, a node receiving a packet adjusts its transmission range in order to cover
the target number of nodes.
The efficient reliable 1-hop broadcasting (EROB) algorithm is introduced in [Park
and Yoo 2013]. It considers simultaneous transmission over three different channels:
two control channels, and another for data packets. It proposes the use of a control
packet called BIP (broadcast in progress) to avoid the hidden terminal problem while
broadcasting. Additionally, nodes can use different transmission levels for enhancing
the network throughput and lifetime, as well as reducing the power consumption and
the number of collisions.
In RandomCast [Lim et al. 2009], the algorithm uses the power saving mechanism
of 802.11 PSM [Anastasi et al. 2008] that allows a device to be in a low-power sleep
state if it is not concerned by data transmission. However, nodes in RandomCast con-
sistently operate in PS mode. The transmitter can specify the level of overhearing
for unicast packets. Additionally, the probability of rebroadcasting is also is based on
number of 1 and 2-hop neighbors.
In Table III, a summary containing the surveyed works that use a variable trans-
mission range, and that are not topology-based is presented.
4. TOPOLOGY-BASED ALGORITHMS
In order to make smart forwarding decisions, many works propose the creation or the
use (in case it is already created) of a topology in the network. As we already mentioned
in Section 2, there are mainly three topological structures that have been used for
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:19
Table III. Classification of the non-topology-based algorithms with variable
transmission range
Neigh. Hops Determ. GPS Forward decision
RNG [Cartigny et al. 2003] local 2-hop yes yes source independent
[Cartigny et al. 2005]
[Wang et al. 2010]
PABLO [Chen et al. 2003] local 2-hop no no source independent
INOP [Chiganmi et al. 2008] local 2-hop no no source indep./dep.
[Reumerman and Runi 2005] local 1-hop yes no source independent
EROB [Park and Yoo 2013] local 1-hop no yes source independent
RandomCast [Lim et al. 2009]
local 1-hop no no source independent
EDB [Ruiz and Bouvry 2010b]
AEDB [Ruiz and Bouvry 2010a]
[Schleich et al. 2011]
solving the broadcast problem. These are: connected dominating sets, tree-based and
cluster-based topologies.
Literature reveals that the construction and maintenance of these three different
topologies have been addressed using many different techniques as well as the broad-
casting algorithms working on top of those topologies. Mainly, we can classify them
into approaches that either use a fixed or a variable transmission range. Next, we
briefly review the most relevant works.
4.1. Fixed Transmission range
4.1.1. Connected dominating set.
A Connected dominating set (CDS) is a source independent technique that builds
a sub-network, also called virtual backbone, for covering every single device in the
network. Nodes either belong to the backbone or are adjacent to at least one of the
back-bone nodes. An extensive review can be found in [Yu et al. 2013]. In graph theory,
a connected dominating set is defined as follows:
Definition 4.1 (Connected dominating sets-CDS).
A connected dominating set for G= (V, E )is a subset V′⊆Vsuch that for all
u∈V−V′there is at least a v∈V′for which (u, v )∈Eand the sub-graph induced by
V′,G[V′]has only one connected component.
A connected dominating set contains a group of connected nodes that covers all the
network, forming a virtual backbone. Efficient broadcasting can be achieved by iden-
tifying a connected dominating set, preferably small, and allowing only the nodes in
the CDS to forward the message. Peng and Lu propose an efficient CDS-based broad-
casting algorithm in [Peng and Lu 2001]. The algorithm uses 2-hop knowledge and
includes the information about the dominant nodes (or nodes belonging to the back-
bone) in the message. An important difference is that not only the nodes covered by
the source node are not considered for belonging to the CDS anymore, but also the
nodes covered by the other dominant nodes that were selected before.
A marking process was proposed in [Wu and Li 1999] for efficiently and quickly
determining a CDS. Originally, it was proposed only for undirected graphs using the
concept of dominating set, but it was later extended to support directed graphs by
introducing the notion of absorbent sets[Wu 2002]. Specifically, a node becomes a gate-
way if it has two unconnected neighbors.
Wu et al. proposed in [Wu et al. 2000] a broadcasting and routing algorithm based on
connected dominating sets and on dynamic selection of the dominating nodes. The algo-
rithm alternates the nodes belonging to the CDS when possible because they consume
more energy. Additionally, nodes with higher remaining battery levels are preferred.
Adijh et al. proposed in [Adjih et al. 2004] a source independent MPR approach for
creating and maintaining a CDS in the network before any broadcasting process start
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:20 P. Ruiz and P. Bouvry
using two simple rules. A node belongs to the CDS is the smallest in its neighborhood
or if it was selected as a forwarder by the neighbor with the lowest node ID.
In [Wu et al. 2006], Wu and Lou proposed an enhancement of the source independent
MPR [Adjih et al. 2004] for creating relatively stable CDS. They use 3-hops neighbors
knowledge in order to have complete 2-hops information, that is, the connections be-
tween any 2-hops neighbors. Authors extend the concept of coverage by considering
that a node is covered by another if it is included in its 1-hop neighborhood (directly
covered) or if it is a 2-hops neighbor (indirectly covered). During the construction pro-
cess of the CDS, each node chooses a pair of nodes, a 1-hop and a 2-hop neighbor to
cover all its 2-hop neighbors. The main difference here is that 2-hop nodes can be indi-
rectly selected to cover other 2-hop neighbors.
The Rule k was proposed in [Dai and Wu 2003] to reduce the overhead for creating
connected dominating sets. It is fully decentralized and supports unidirectional links.
This protocol was modified in [Stojmenovi´
c and Wu 2004] to eliminate the exchange of
messages needed to create the CDS.
Stojmenovi´
c et al. proposed a broadcasting algorithm based on dominating sets and
neighbor elimination schemes using 2-hop knowledge [Stojmenovi´
c et al. 2002]. Only
nodes belonging to the CDS are able to rebroadcast a message. Upon reception, nodes
in the CDS set a delay which is inversely proportional to the number of non-covered
neighbors. During the waiting period, if copies of the message are heard, all the neigh-
bors that received the message are also removed from the forwarding list. If there is
no neighbor to cover, the neighbor elimination rule cancels the retransmission.
In order to improve reliability and solve disconnection problems, in [Stojmenovi´
c
et al. 2012] the authors proposed a broadcasting algorithm that works like in [Stojmen-
ovi´
c et al. 2002], but all nodes can rebroadcast a packet. Nodes that are not included
in the CDS set a longer waiting time, and additionally, every time a new neighbor is
met, the broadcast message is sent.
Recently, a general framework for broadcasting that seamlessly (without using any
parameter) adjusts itself to any mobility scenario was introduced in [Stojmenovi´
c
2012]. It is built over several recent algorithms using 2-hop topological or 1-hop po-
sitional knowledge. When the message is received the first time, nodes set a timeout
and also generate 2 lists: Rwith the nodes believed to have received the packet, and
Nwith the nodes that did not receive the message yet. Nodes belonging to the CDS
have shorter timeouts than those not in the CDS. Once the waiting time is finished,
the node retransmits only if Nis not empty, and updates both Rand N. After receiving
each hello message, nodes reevaluate the CDS status (if it is a dominating or a domi-
nated node) and also update N. If Nwas empty and a new neighbor not included in R
is detected, the timeout is reactivated.
Literature reveals yet different decentralized CDS that use local knowledge in dy-
namic ad hoc networks. Two proposals designed to create k-vertex connected m-vertex
dominating set virtual backbones in an asynchronous and computer-effective way are
presented in [Leu and Chang 2011; Schleich et al. 2012]. For a more detailed explana-
tion on CDS please refer to [Schleich 2010].
4.1.2. Tree-based topology.
All the already-mentioned approaches have to store the ID of the broadcast message
in order to just rebroadcast the first reception of the message. The most straightfor-
ward method for avoiding the storage of any ID is to broadcast the packet using an
acyclic subgraph.
A network can be modeled as a graph G=(V,E), where nodes are represented by the
set of vertices Vand the links are the set of edges E. It is possible to define an under-
lying structure in the network, as done with the CDS and the clustering, that forms a
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:21
tree. In order to disseminate a message to all nodes in the network, we can consider
the spanning tree structure where only nodes belonging to the tree and that are not
leaves rebroadcast the message. Next, we define both concepts: tree and spanning tree.
Definition 4.2 (Tree topology).
A tree is an undirected and connected graph that has no cycles.
Definition 4.3 (Spanning tree).
A spanning tree is an undirected and connected subgraph that has no cycles and
contains all nodes. It can be defined as the maximal set of edges of the graph that
contains no cycle, or as a minimal set of edges that connect all vertices of the graph.
The minimum spanning tree (MST) is the minimal set of edges that connect all vertices
of the graph, or in case the graph is weighted, the minimum sum of the weight of all
branches.
In case the network is partitioned, we can talk about spanning forest, where a span-
ning tree is formed in every connected component, see Figure 6.
a) Tree b) Forest
Fig. 6. Tree and forest topologies
The maximum leaf spanning tree problem is equivalent to the minimum connected
dominating set [Fujie 2003] problem. In a tree-based topology, edges are chosen and
play an important role, while in CDS the nodes play the central role. However, the
construction of the spanning tree from a CDS is quite straightforward. In Figure 7, we
can see a graphical representation.
a) Minimum spanning tree b) Minimum connected dominating set
Fig. 7. Minimum spanning tree and and minimum CDS
Tree topology has been widely used in telecommunication networks [for Informa-
tion Technology 1998], but most of these algorithms need a stable network, that is not
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:22 P. Ruiz and P. Bouvry
available in MANETs. Therefore, specific algorithms for mobile ad hoc networks must
be designed.
Creating and maintaining a tree topology in mobile ad hoc networks is challenging
because the nodes have knowledge only about their direct neighbors and the links are
very volatile. Therefore, the data structure and the required maintenance of the tree
must be minimal. It is possible to differentiate between source-dependent and shared
trees. In a source-dependent tree, the root of the tree is located at the source node and
for each pair of source-multicast group a different tree is built. In the shared tree only
one tree is built over the network and the source node just needs to be able to send
data to the root of the tree or to a member.
In [Radhakrishnan et al. 1999], the authors proposed a distributed spanning tree
to forward data packets. However, some operations require centralized designs as for
merging two spanning trees, the root node is deciding which nodes are merging.
Using the overlay of a gossip protocol, the push-lazy-push multicast tree (Plumtree)
is constructed in [Leitao et al. 2007]. The links from which a node receives the mes-
sages are considered to form the spanning tree, and when a packet is received twice,
the node prunes this link. However, for repairing the tree, every time a node receives
a message it sends an IHAVE message to the links that do not belong to the span-
ning tree. Whenever a node requests this message, the link between these two nodes
becomes part of the spanning tree.
A fully distributed and decentralized method, TreeCast is proposed in [Juttner and
Magi 2005]. It is not only capable of dealing with mobility of nodes, but also forces
nodes with high mobility to be leaves, thus, the backbone is composed of the more
static nodes.
A parameterless broadcast algorithm working over a tree topology is BODYF [Ruiz
et al. 2008]. It uses the topology information for a fast dissemination, while does not
prevent neighboring nodes from receiving the message, i.e., it does not multicast the
message to its neighboring tree nodes, but it broadcast the message. In a mobile envi-
ronment, it is very unlikely to have a single tree over the network, but a forest. Thus,
BODYF is able to disseminate the message over different trees. In [Ruiz et al. 2012],
more efficient version of BODYF is proposed and tested over different MANETs and
VANETs scenarios. Additionally, a comparison between different broadcast algorithms
is presented.
MaxCST [Flauzac et al. 2010] is a deterministic and self-establishing algorithm that
builds a cluster in the network with diameter of 2 at most and then, constructs a
spanning tree on the network.
The creation of a tree topology in decentralized systems is a complex task, usu-
ally addressed by computing the local tree and then, merging adjacent trees (e.g. the
LMST [Li et al. 2005]). This approach requires static networks or low mobility. How-
ever, the dynamicity aware graph relabeling system [Casteigts and Chaumette 2005]
is a high level abstraction model for creating and maintaining tree topologies effi-
ciently. It is based on local rules that are able to cope with topological changes. Also
in [Casteigts et al. 0013], a decentralized algorithm that is able to cope with the cre-
ation and maintenance of distributed spanning trees in highly dynamic networks is
presented. Both, the merging and splitting process of two trees are purely locally op-
erated.
Many works also look beyond the topological aspects and attach specific features
to nodes. For example, in [Piyatumrong et al. 2008], nodes are selected for joining
the tree in terms of their level of trust in the network, creating thus, more reliable
topologies. For a more detailed overview of this type of tree-based topologies, please
refer to [Piyatumrong 2010].
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:23
4.1.3. Clustering-based topology.
A clustered network is a 2-level hierarchical network that divides nodes into groups
and elects only one node as representative of the group, the clusterhead (CH). The CH
directly connects to all other members of the cluster. Nodes belong only to one CH, but
sometimes can hear more than two CH. Those nodes declare themselves gateway, see
Figure 8.
The set of clusterheads forms a dominating set, but not a connected dominating set.
In order to build a CDS, clusterheads must elect the gateway as the forwarding nodes
for connecting two neighboring CHs.
CH CH
Gateway
Fig. 8. Two adjacent clusters with a gateway
The main goal of both CDS and trees is to cover the whole network using the mini-
mum number of nodes in the CDS or obtaining the maximum number of leaves. How-
ever, in clustering the main goal is to group nodes in terms of some common features
or a joint goal.
This cluster-based network structure has a straightforward application in hybrid
networks, where some uplinks are established with a backbone infrastructure. Addi-
tionally, clustering nodes by designating a special role for one of them, the clusterhead,
offers many advantages for optimizing energy consumption, network organization,
data aggregation, etc. However, the failure of a CH has big impact on the performance
of the clustered network, as well as the overhead due to re-election and re-clustering
of the network. Also, node mobility plays an important role in clustering as continuous
formation and maintenance of the cluster is required, what leads to big overhead and
less stable networks.
Mainly, there are 2 different approaches in clustering: (1) active clustering, where
nodes cooperate to elect the CH by exchanging information periodically, and (2) pas-
sive clustering where the CH election starts with the ongoing traffic. Information is
propagated in the packets and collected through eavesdropping, such that there is no
induced overhead or setup costs.
In the Lowest ID (LID) clustering algorithm [Ephremides et al. 1987], all nodes are
coloured white, in first place. A white node with the lowest ID among its non clus-
terhead neighbors chooses itself as CH and turns black, and all white neighbors join
this CH and turn themselves grey. The process finishes when there are no more white
nodes. The Highest degree algorithm (HD) [Gerla and Tsai 1995], considers first the
node degree in the clustering decision. In case of tie, the lowest ID node is selected. A
survey on different clustering techniques can be found in [Yu and Chong 2005]. Next,
we present some of the most relevant broadcasting algorithms based on clustering
approaches.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:24 P. Ruiz and P. Bouvry
In [Gerla et al. 2000], a broadcasting algorithm based on passive clustering is pro-
posed. Nodes decide to become either a CH, an ordinary node or a gateway in terms of
some predefined thresholds. Both CHs and gateways rebroadcast the message.
For electing the clusterheads, not only location information but also the battery level
is used in [Li et al. 2004]. The proposed vote-based clustering considers nodes with high
degree and battery level better candidates for becoming CHs. However, this results in
frequently changing topology.
A weighted algorithm called WACA was presented in [Brust et al. 2007]. It is an
application-driven algorithm, i.e., the heuristic weight function depends on the tar-
geted application. The concept of sub-head is considered in this case; i.e., nodes elect a
neighbor as CH, but this one also elects another neighbor as CH.
One of the main challenges in clustering is related to mobility, i.e., mobility causes
very frequent CH changes and thus, leads to an unstable structure. In order to cope
with this problem, the stability of the link is also considered in the node and link
weighted clustering algorithm (NLWCA) [Andronache and Rothkugel 2008]. It uses
the node state (device power and signal strength) and the link stability for electing
CHs. For broadcasting, clusterheads disseminate the message within their own cluster
but also send it directly to all stable neighboring clusterheads by unicast [Andronache
et al. 2008].
Wu and Lou proposed in [Wu and Lou 2003b] a clustered network where each clus-
terhead selects the forwarding nodes inside the cluster so that all clusterheads in the
vicinity are covered. Clusterheads receiving the message are included in the broadcast
message so that CHs can deduce the coverage area by the pruning technique. Instead
of the 3-hop covered area, this paper introduces the novel idea of 2.5-hop coverage.
Every CH only reaches the clusterheads that have members not beyond 2 hops.
Node mobility really influences and degrades the quality of the cluster structure.
Thus, reducing the number of clusters in the network will improve scalability, and
increasing the number of hops inside the cluster reduces the effect of mobility. That
motivates the k-hop clustering algorithms, that allows to choose the diameter of the
network adding also more flexibility. Nocetti et al. generalized the clustering definition
in [Nocetti et al. 2003], where any node in the cluster is at most k-hops from the clus-
terhead. In this work, authors proposed an approach for k-clustering where CHs are
elected in terms of connectivity and lowerID. One main concern in clustering is that a
CH may become a bottleneck, thus in [Fernandess and Malkhi 2002], Fernandess and
Malkhi gives a formal definition of the minimum k-clustering problem in which the
network is divided into clusters without using CHs.
Definition 4.4 (Minimum k-clustering).
Given a unit disk graph denoted by G=(V,E) and a positive integer k, find the small-
est value of lsuch that there is a partition of Vinto ldisjoint subsets V1...Vland diam
(G[Vi]) ≤k for i=1..l.
In this work [Fernandess and Malkhi 2002], authors also proposed an algorithm
which first creates a spanning tree, and then partitions it. However, it does not deal
with mobility or topology changes.
KHOPCA is proposed in [Brust et al. 2008], a k-hop clustering algorithm that only
uses 1-hop neighborhood information for cluster creation and maintenance based on
some rules. Nodes change their status depending on their weight and their direct
neighbors. The algorithms works in a distributed fashion and is highly adaptive to
mobility. It is based on rules like the max-min heuristic proposed in [Amis et al. 2000]
k-hop clustering, where CHs are selected according to the ID of the nodes. The number
of the hops in the cluster determines the number of rounds for exchanging informa-
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:25
tion. Additionally, the system tries to maintain stability in the cluster by re-electing
CHs when possible.
A recent survey dedicated to clustering algorithms in MANETs can be found in [Ben-
taleb et al. 2013].
In Table IV, a summary containing the surveyed works with fixed transmission
range, and that are topology-based is presented.
Table IV. Classification of the topology-based algorithms with fixed transmission range
Neigh. Hops Determ. GPS Forward decision
CDS [Wu et al. 2006] local 3-hop yes no structure
others local 2-hop yes no structure
Trees
PlumTree [Leitao et al. 2007]
local 1-hop no no structure
BODYF [Ruiz et al. 2008]
[Ruiz et al. 2012]
[Radhakrishnan et al. 1999] local/global 1-hop yes no structure
TreeCast [Juttner and Magi 2005] local 2-hop yes no structure
others local 1-hop yes no structure
Clustering
NLWCA [Andronache and Rothkugel 2008]
local 1-hop yes no structure
WACA [Brust et al. 2007]
KHOPCA [Brust et al. 2008]
[Gerla et al. 2000] local – yes no structure
vote-based clustering [Li et al. 2004] local 1-hop yes yes structure
[Wu and Lou 2003b] local 2.5-hop yes no structure
Lowest ID [Ephremides et al. 1987] local 2-hop yes no structure
Highest degree [Gerla and Tsai 1995]
4.2. Using a Variable Transmission Range
Next, we survey different topology-based algorithms where nodes can vary their trans-
mission range.
4.2.1. Connected dominating sets.
A broadcasting algorithm that uses variable transmission range based on connected
dominating set is presented in [Wu and Wu 2003]. As nodes belonging to the CDS
consume more energy, they alternate when possible. Additionally, the transmission
power of every node is reduced during the dissemination with no detriment to the
reachability.
Some distance rules where node’s transmission power might be adjusted based on
the distance are proposed in [Stojmenovi´
c and Wu 2004]. Moreover, nodes with less
distance to their neighbors have higher preference for joining the CDS.
In [Li et al. 2012], two decentralized approaches are proposed for constructing the
connected dominating set that minimize the total communication power. The goal is to
find the CDS that minimizes the sum of the transmission power of CDS nodes.
4.2.2. Tree-based topology.
In order to solve the minimum energy broadcast problem, the well known broadcast
incremental power (BIP) was presented in [Wieselthier et al. 2000]. In BIP, the con-
struction mechanism is as follows. The tree is constructed at the source node, and the
next node to be included in the tree, is the one that is reached using the minimum
power. The sweep procedure is also proposed in the same work for further reducing the
total energy consumption. BIP requires global information.
BAIP, the broadcast average incremental power [Wan et al. 2001] is a variant of
BIP that considers the average incremental cost for selecting a new node that will be
included in the tree. This value is the ratio between the minimum power required by
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:26 P. Ruiz and P. Bouvry
that a node in the current tree needs to reach new nodes, and the number of new nodes
that will be reached.
The hop constrained minimum broadcast incremental power (HC-BIP) is proposed
in [Bulbul et al. 2009]. It ensures that all nodes in the network receive the broadcast
message in less than khops. It requires global-knowledge.
The iterative maximum-branch minimization (IMBM) algorithm was presented
in [Li and Nikolaidis 2001]. It focuses on the construction of a tree that minimize the
energy required for broadcast. In an initial step, the source builds a basic broadcast
tree to all destinations. Then, it tries to reduce the required power by replacing the
most energy-expensive branch by an alternative that uses less power. Liang proposes
in [Liang 2002] an approximation algorithm that uses global knowledge and is based
on Steiner trees.
In [Cagalj et al. 2002], the embedded wireless multicast advantage (EWMA) algo-
rithm starts with an initial solution obtained using the minimum spanning tree (MST),
and then, this solution is improved by replacing branches of the initial tree with new
ones, so that the overall energy used for maintaining the tree is lower. Two EWMA
approaches are proposed, one being centralized and the other is distributed. The dis-
tributed approach, however, requires information from multiple hops away, such that
it might not work properly in frequently changing topologies.
Wieselthier proposed in [Wieselthier et al. 2002] two distributed versions of the well
known BIP: distributed-BIP-all Dist-BIP-A, and distributed-BIP-gateways Dist-BIP-G.
Both use 2-hop neighbors information for constructing trees. In the former, only nodes
that can be reached directly are included in the tree. Each node builds a local tree
that is then assembled in a global tree. In the latter, nodes that can be reached using
a relay are considered gateways. Only, those gateways are in charge of building local
trees. Once the global tree is formed, a centralized sweep operation is applied.
The local minimum spanning tree (LMST) [Li et al. 2003; 2005] needs the local ex-
change of positions in the beacons to build the neighborhood graph. Each node applies
Prim’s algorithm in order to find the local minimum spanning tree. Once, the tree is
constructed it reduces the transmission power to reach the furthest neighbor. The di-
rected LMST broadcast oriented protocol (LBOT) [Cartigny et al. 2004] is based on
LMST but uses directional antennas and lies on the LMST topology. The source node
sends the message, and the message is propagated using the neighbor elimination (or
self-pruning rule) on the LMST.
In [Miyao et al. 2009], the local tree-based reliable topology (LTRT) motivated from
LMST is proposed. It ensures k-edge connectivity of the network in order to ensure
reliable communications.
The broadcast on local minimum spanning tree (BLMST) is proposed in [Li and Hou
2004]. It first constructs an underlying topology using LMST, and then, the source node
broadcasts the message and a node receiving the message from all its neighbors can-
cels the forwarding. This work also includes an analytical study that indicates under
some circumstances it is more efficient to use lower transmission power in a multi-hop
fashion than using longer transmission ranges.
The redundant radius scheme is introduced in [Xu and Garcia-Luna-Aceves 2010]
where two different transmission radii are used. First, a smaller range is considered
for building the broadcast tree in terms of the neighborhood, and then, a longer radius
is used for the actual transmission.
In [Ingelrest et al. 2006], the optimal transmission range that considers both the
number of relays and the energy consumption is calculated. Moreover, two broadcast-
ing algorithms are proposed: the target radius LMST broadcast oriented protocol (TR-
LBOP) and the target radius and dominating set based protocol. The former considers
the neighbor elimination to reduce the subset of direct neighbors and reduces the ra-
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:27
dius to preserve connectivity, increasing it up when possible. The latter computes a
CDS trying to choose the relays nodes as close as possible to the target radius.
4.2.3. Clustering topology.
An autonomous cluster scheme is proposed in [Oda et al. 2010]. It adapts the trans-
mission power of every node in terms of the distance between the node and the neigh-
boring nodes to have a specific number of neighbors.
In [Ni et al. 2010], the node residual energy, the nearby topology, the relative location
and the relative mobility are used for electing the CH. Additionally, cluster members
are able to estimate the distance to the CH and reduce the transmission power accord-
ingly. When the residual energy of the CH is lower than a predefined threshold, the
re-clustering operation is triggered.
In [Ahluwalia and Modiano 2005], 2-hop neighborhood information is used in a dis-
tributed clustering algorithm to divide the network, where clusters may overlap. Then,
a distributed sweep operation is used for finding nodes whose transmission power can
be decreased, while still guaranteeing that every node belongs to at least one clus-
ter. Finally, it runs the DMST [Humblet 1983] algorithm for constructing the directed
minimum spanning tree to join the clusters.
In Table V, a summary containing the surveyed works that use a variable transmis-
sion range, and that are topology-based is presented.
Table V. Classification of the topology-based algorithms with variable transmission range
Neigh. Hops Determ. GPS Forward decision
BIP [Wieselthier et al. 2000]
local global yes no structure
BAIP [Wan et al. 2001]
HC-BIP [Bulbul et al. 2009]
IMBM [Li and Nikolaidis 2001]
[Liang 2002]
EWMA [Cagalj et al. 2002]
Dist-BIP-A/Dist-BIP-G [Wieselthier et al. 2002] local 2-hop/global yes no structure
LMST [Li et al. 2003]
local 1-hop yes yes structure
LBOT [Cartigny et al. 2004]
LTRT [Miyao et al. 2009]
BLMST [Li and Hou 2004]
[Xu and Garcia-Luna-Aceves 2010] local 1-hop yes no structure
TR-LBOP [Ingelrest et al. 2006] local 1-hop no yes structure
[Wu and Wu 2003]
local 1-hop yes no structure
[Stojmenovi´c and Wu 2004]
[Li et al. 2012]
[Oda et al. 2010]
local 2-hop yes no structure
[Ni et al. 2010]
[Ahluwalia and Modiano 2005]
5. OPEN ISSUES AND CHALLENGES
In the previous sections, we have surveyed the existing protocols and we have classified
them. As we have seen, broadcasting in mobile ad hoc networks has been extensively
studied, however, there are still many open issues and challenges to be addressed.
Next, we outline some of the most relevant ones.
— Due to the changing and unpredictable topology in MANETs, most of the algorithms
rely on different thresholds so that they can adapt to topological changes. Those
thresholds are usually experimentally chosen or set for a specific network configu-
ration. Distributed and online algorithms able to tune the value of the parameters
in real time would highly increase the performance of the broadcasting protocols.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:28 P. Ruiz and P. Bouvry
— Broadcasting is a MAC level operation; transmission errors and interference make
broadcasting unreliable. In [Boukerche 2008], authors state that the network-wide
broadcast problem must be viewed as two components: redundancy control and ro-
bustness control. The former aims at reducing the redundancy while maintaining
coverage, while the latter tries to recover from lost messages while maintaining the
coverage. Efficient broadcasting protocols that deal with redundancy, robustness as
well as the overhead produced by the control mechanisms are still challenging the
research community.
— Enhancing the underlying topology intelligence in order to automate the forward-
ing mechanism by reacting to topological changes. The topology must be aware if
the network stability is highly volatile or very sparse and act consequently by for-
warding the broadcast message as soon as a node enters the tree/cluster/CDS or
canceling the RAD. Additionally some pro-activity might be added by retransmit-
ting the message before some topological split happens.
— The higher the number of nodes resending the message, the higher the probability
of reaching the whole network, but the higher the network use. Therefore, the de-
velopment of intelligent distributed methods able to estimate the optimal number
of nodes forwarding the message will significantly help the dissemination process
and optimize the utilization of the network resources.
— MANETs suffer from network partitioning and disruption that makes the broadcast
message dissemination difficult and very often unreliable. Thus, methods aware of
node mobility and network partition that can predict node movements to promote
the message dissemination can significantly improve the broadcast process.
6. CONCLUSIONS AND PERSPECTIVES
Broadcasting is a cornerstone in networking, what means that many work have al-
ready been done, but there is still room for improvement. There are yet many chal-
lenges that have to be solved efficiently. Classifying and compiling existing works is
key for newcomers in the topic so that they easily familiarize and have a global picture
of the literature.
In this paper, we have surveyed the most relevant works that have been proposed
in the last decades for solving the broadcast problem. In order to do that, we first
analyzed the existing classifications and proposed a new taxonomy that better covers
all existing techniques.
According to the proposed taxonomy, we later analyzed in detail the state of the art
in this field and categorized the approaches in terms of the type of centralization of the
system, the knowledge used, the predictability of the system and the role of the source
node.
After having this global view of the broadcast problem, open issues were identified
and, thus, new research lines proposed.
REFERENCES
The Haggle Project. http://www.haggleproject.org. (????). Retrieved February 16th 2014 from
http://www.haggleproject.org.
C. Adjih, P. Jacquet, and L. Viennot. 2004. Computing connected dominated sets with multipoint relays.
Technical Report. http://gyroweb.inria.fr/∼viennot/publis/articles/mpr-cds.pdf
A.S. Ahluwalia and E.H. Modiano. 2005. On the complexity and distributed construction of energy-efficient
broadcast trees in wireless ad hoc networks. IEEE Transactions on Wireless Communications 4, 5 (2005),
2136–2147.
J.-H. Ahn and T.-J. Lee. 2011. A multipoint relay selection method for reliable broadcast in ad hoc networks.
In ICT Convergence (ICTC), 2011 International Conference on. 616–617.
S.O. Al-Humoud, L.M. Mackenzie, and J. Abdulai. 2008. Neighbourhood-Aware Counter-Based Broadcast
Scheme for Wireless Ad Hoc Networks. In GLOBECOM Workshops, 2008 IEEE. 1–6.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:29
L. Al-Kanj and Z. Dawy. 2011. Offloading Wireless Cellular Networks via Energy-Constrained Local Ad Hoc
Networks. In IEEE Global Telecommunications Conference, GLOBECOM. 1–6.
S. Allal and S. Boudjit. 2012. Geocast Routing Protocols for VANETs: Survey and Guidelines. In Sixth In-
ternational Conference on Innovative Mobile and Internet Services in Ubiquitous Computing. 323–328.
DOI:http://dx.doi.org/10.1109/IMIS.2012.133
A.D. Amis, R. Prakash, T.H.P. Vuong, and D.T. Huynh. 2000. Max-min d-cluster formation in wireless ad hoc
networks. In Nineteenth Annual Joint Conference of the IEEE Computer and Communications Societies.
INFOCOM, Vol. 1. 32–41.
G. Anastasi, M. Conti, E. Gregori, and A. Passarella. 2008. 802.11 power-saving mode for mobile computing
in Wi-Fi hotspots: limitations, enhancements and open issues. Wireless Networks 14, 6 (Dec. 2008), 745–
768.
A. Andronache and S. Rothkugel. 2008. NLWCA Node and Link Weighted Clustering Algorithm for
Backbone-Assisted Mobile Ad Hoc Networks. In Seventh International Conference on Networking. 460–
467.
A. Andronache, P. Ruiz, and S. Rothkugel. 2008. Improving Inter-cluster Broadcasting in Ad Hoc Networks
by Delayed Flooding. In MCO. 478–487.
J. Arango, M. Degermark, A. Efrat, and S. Pink. 2004. An efficient flooding algorithm for ad-hoc networks.
In Proceedings of the Second Workshop on Modeling and Optimizations in Mobile Ad Hoc and Wireless
Networks (WiOpt).
S. Bai, Z. Huang, D. Kwak, Sa. Lee, H. Oh, and J. Jung. 2009. Vehicular Multi-Hop Broadcasting Protocol for
Safety Message Dissemination in VANETs. In Vehicular Technology Conference Fall (VTC 2009-Fall),
IEEE 70th. Anchorage, AK, 1 –5.
A. Banerjee, C.H. Foh, C.K. Yeo, and B.S. Lee. 2012. Performance improvements for network-wide broadcast
with instantaneous network information. Journal of Network and Computer Applications 35, 3 (2012),
1162 – 1174. Special Issue on Trusted Computing and Communications¡/ce:title¿.
S. Basagni, M. Conti, S. Giordano, and I. Stojmenovi´c. 2004. Mobile Ad Hoc Networking. Wiley-IEEE Press.
S. Bastani, B. Landfeldt, C. Rohner, and P. Gunningberg. 2012. A Social Node Model for Realising Infor-
mation Dissemination Strategies in Delay Tolerant Networks. In Proceedings of the 15th ACM Interna-
tional Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM ’12).
ACM, New York, NY, USA, 79–88.
A. Bentaleb, A. Boubetra, and S. Harous. 2013. Survey of Clustering Schemes in Mobile Ad hoc Networks.
Communications and Network 5 (2013), 8–14.
D.M. Blough, M. Leoncini, G. Resta, and P. Santi. 2006. The k-Neighbors Approach to Interference Bounded
and Symmetric Topology Control in Ad Hoc Networks. IEEE Transactions on Mobile Computing 5, 9
(2006), 1267–1282.
A. Boukerche (Ed.). 2005. Handbook of Algorithms for Wireless Networking and Mobile Computing. Chap-
man & Hall/CRC Computer and Information Science Series.
A. Boukerche (Ed.). 2008. Algorithms and Protocols for Wireless, Mobile Ad Hoc Networks. Wiley-IEEE Press.
A. Boukerche and V. Sheetal. 2003. A Performance Evaluation of a Dynamic Source Routing Discovery
Optimization Protocol Using GPS System. Telecommunication Systems 22, 1–4 (2003), 337–354.
M. R. Brust, A. Andronache, and S. Rothkugel. 2007. WACA: A hierarchical weighted clustering algorithm
optimized for mobile hybrid networks. In Third International IEEE Conference on Wireless and Mobile
Communications. 23–23.
M. R. Brust, H. Frey, and S. Rothkugel. 2008. Dynamic multi-hop clustering for mobile hybrid wireless
networks. In Proceedings of the 2nd international conference on Ubiquitous information management
and communication. ACM, 130–135.
K. Bulbul, O. Ercetin, and T. Unluyurt. 2009. Hop Constrained Energy-Efficient Broadcasting: Insights from
Massively Dense Ad Hoc Networks. IEEE Transactions on Mobile Computing 8, 2 (2009), 188–202.
M. Cagalj, J-P Hubaux, and C. Enz. 2002. Minimum-energy broadcast in all-wireless networks: NP-
completeness and distribution issues. In MobiCom ’02: Proceedings of the 8th annual international
conference on Mobile computing and networking. ACM, 172–182.
Z-Y CAO, Z-Z JI, and M-Z HU. 2007. An energy-aware broadcast scheme for directed diffusion in wireless
sensor network. Journal of Communication and Computer 4,, 5 (2007), 28–35.
J. Cartigny, F. Ingelrest, D. Simplot-Ryl, and I. Stojmenovic. 2005. Localized LMST and RNG based
minimum-energy broadcast protocols in ad hoc networks. Ad Hoc Networks 3, 1 (2005), 1–16.
J. Cartigny and D. Simplot. 2003. Border Node Retransmission Based Probabilistic Broadcast Protocols
in Ad-Hoc Networks. In Proceedings of the 36th Hawaii International Conference on System Sciences.
189–204.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:30 P. Ruiz and P. Bouvry
J. Cartigny, D. Simplot, and I. Stojmenovi´c. 2003. Localized Minimum-Energy Broadcasting in Ad-Hoc Net-
works. (2003).
J. Cartigny, D. Simplot-Ryl, and I. Stojmenovi´c. 2004. An Adaptive Localized Scheme for Energy-Efficient
Broadcasting in Ad Hoc Networks with Directional Antennas. In Personal Wireless Communications,
Ignas Niemegeers and SoniaHeemstra Groot (Eds.). Lecture Notes in Computer Science, Vol. 3260.
Springer Berlin Heidelberg, 399–413.
A. Casteigts and S. Chaumette. 2005. Dynamicity Aware Graph Relabeling System - a local computation
model to describe MANET algorithms. In Proceedings of the 17th IASTED International Conference on
Parallel and Distributed Computing and System. 198–204.
A. Casteigts, S. Chaumette, F. Guinand, and Y. Pign´e. 20013. Distributed Maintenance of Anytime Available
Spanning Trees in Dynamic Networks. In The 12th International Conference on Ad Hoc Networks and
Wireless, ADHOC-NOW.
C. Chen, C-K Hsu, and H-K Wang. 2005. A distance-aware counter-based broadcast scheme for wireless ad
hoc networks. In IEEE Military Communications Conference (MILCOM). 1052–1058.
X. Chen, M. Faloutsos, and S. Krishnamurthy. 2002. Distance ADaptive (DAD) broadcasting for ad hoc
networks. In Proceedings MILCOM, Vol. 2. 879–883.
X. Chen, M. Faloutsos, and S.V. Krishnamurthy. 2003. Power Adaptive Broadcasting with Local Information
in Ad Hoc Networks. In Conference on Network Protocols. IEEE Computer Society, 168.
A. Chiganmi, M. Baysan, K. Sarac, and R. Prakash. 2008. Variable power broadcast using local information
in ad hoc networks. Ad Hoc Networks 6, 5 (July 2008), 675–695.
T.H. Clausen and P. Jacquet. 2003. Optimized Link State Routing Protocol (OLSR). Technical Report. IETF.
75 pages.
A.E.F. Clementi, P. Penna, and R. Silvestri. 2000. The Power Range Assignment Problem in Radio Networks
on the Plane. In The 17th Annual Symposium on Theoretical Aspects of Computer Science (STACS ’00).
Springer-Verlag, London, UK, UK, 651–660.
F. Dai and J. Wu. 2003. Distributed dominant pruning in ad hoc networks. In IEEE International Conference
on Communications, Vol. 1. 353–357.
A. Durresi, V.K. Paruchuri, S.S. Iyengar, and R. Kannan. 2005. Optimized broadcast protocol for sensor
networks. IEEE Trans. Comput. 54, 8 (2005), 1013–1024.
A. Ephremides, J.E. Wieselthier, and .J Baker. 1987. A design concept for reliable mobile radio networks
with frequency hopping signaling. Proc. IEEE 75, 1 (1987), 56–73.
K. Fall. 2003. A Delay-tolerant Network Architecture for Challenged Internets. In Proceedings of the 2003
Conference on Applications, Technologies, Architectures, and Protocols for Computer Communications
(SIGCOMM ’03). ACM, New York, NY, USA, 27–34. DOI:http://dx.doi.org/10.1145/863955.863960
Y. Fernandess and D. Malkhi. 2002. K-clustering in Wireless Ad Hoc Networks. In Proceedings of the Second
ACM International Workshop on Principles of Mobile Computing (POMC ’02). ACM, New York, NY,
USA, 31–37.
O. Flauzac, B.S. Haggar, and F. Nolot. 2010. Tree and cluster management for MANET. In New Technologies
of Distributed Systems (NOTERE), 2010 10th Annual International Conference on. 73–80.
IEEE Standard for Information Technology. 1998. Telecommunications and Information Exchange
Between Systems - Local and Metropolitan Area Networks - Common Specifications Part 3:
Media Access Control (MAC) Bridges. ANSI/IEEE Std 802.1D, 1998 Edition (1998), i–355.
DOI:http://dx.doi.org/10.1109/IEEESTD.1998.95619
T. Fujie. 2003. An exact algorithm for the maximum leaf spanning tree problem. Computers and Operations
Research 30, 13 (Nov. 2003), 1931–1944.
W. Gang, W. Shigang, L. Cai, and Z. Xiaorong. 2012. Research and Realization on Improved MANET Dis-
tance Broadcast Algorithm Based on Percolation Theory. In International Conference on Industrial Con-
trol and Electronics Engineering (ICICEE). 96–99.
B. Garbinato, A. Holzer, and F. Vessaz. 2008. Six-Shot Broadcast: A Context-Aware Algorithm for Efficient
Message Diffusion in MANETs. In Proceedings of the OTM 2008 Confederated International Confer-
ences, CoopIS, DOA, GADA, IS, and ODBASE 2008. Part I on On the Move to Meaningful Internet
Systems: (OTM ’08). 625–638.
B. Garbinato, A. Holzer, and F. Vessaz. 2010. Context-aware broadcasting approaches in mobile ad hoc
networks. Comput. Netw. 54, 7 (May 2010), 1210–1228.
M. Gerla, T. Kwon, and G. Pei. 2000. On-demand routing in large ad hoc wireless networks with passive
clustering. In Wireless Communications and Networking Confernce, 2000. WCNC. 2000 IEEE, Vol. 1.
100–105.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:31
M. Gerla and J. T. Tsai. 1995. Multicluster, Mobile, Multimedia Radio Network. Journal of Wireless Networks
1 (1995), 255–265.
J. Gomez and A.T. Campbell. 2007. Variable-Range Transmission Power Control in Wireless Ad Hoc Net-
works. IEEE Transactions on Mobile Computing 6, 1 (2007), 87–99.
S. Guo and O.W.W. Yang. 2007. Energy-aware multicasting in wireless ad hoc networks: A survey and dis-
cussion. Computer Communications 30, 9 (June 2007), 2129–2148.
American Heritage. 2014. The American Heritage Dictionary. Definition of the term ad hoc.
http://www.ahdictionary.com/word/search.html?q=ad+hoc&submit.x=0&submit.y=0. (Last accessed De-
cember 2014).
L. Hogie, F. Guinand, and P. Bouvry. 2004. A Heuristic for Efficient Broadcasting in the Metropolitan Ad
hoc Networks. In KES. 529–536.
C-M Huang, L Tu, and C-H Chou. 2009. ReWarn: An opportunistic relay scheme for cooperative collision
warning in VANET. In Personal, Indoor and Mobile Radio Communications, IEEE 20th International
Symposium on. Tokyo, 3030 –3034.
P. Hui, J. Crowcroft, and E. Yoneki. 2011. BUBBLE Rap: Social-Based Forwarding in Delay-
Tolerant Networks. Mobile Computing, IEEE Transactions on 10, 11 (Nov 2011), 1576–1589.
DOI:http://dx.doi.org/10.1109/TMC.2010.246
P.A. Humblet. 1983. A Distributed Algorithm for Minimum Weight Directed Spanning Trees. IEEE Trans-
actions on Communications 31, 6 (1983), 756–762.
F. Ingelrest, D. Simplot-Ryl, and I. Stojmenovic. 2006. Optimal transmission radius for energy efficient
broadcasting protocols in ad hoc and sensor networks. IEEE Transactions on Parallel and Distributed
Systems 17, 6 (2006), 536–547.
A. Juttner and A. Magi. 2005. Tree Based Broadcast in Ad Hoc Networks. Mobile Networks and Applications
10 (2005), 753–762. Issue 5.
K. Karenos, A. Khan, S. Krishnamurthy,M. Faloutsos, and C. Xiaohu. 2005. Local versus global power adap-
tive broadcasting in ad hoc networks. In IEEE Wireless Communications and Networking Conference,
Vol. 4. 2069–2075.
D. Kasamatsu, N. Shinomiya, and T. Ohta. 2009. A Broadcasting Method considering Battery Lifetime and
Distance between Nodes in MANET. In 29th IEEE International Conference on Distributed Computing
Systems Workshops (ICDCS). 406–411.
V. Kawadia and P.R. Kumar. 2005. Principles and protocols for power control in wireless ad hoc networks.
IEEE Journal on Selected Areas in Communications 23, 1 (2005), 76–88.
A.M. Kermarrec, L. Massouli´e, and A.J. Ganesh. 2003. Probabilistic Reliable Dissemination in Large-Scale
Systems. IEEE Trans. Parallel Distrib. Syst. 14, 3 (March 2003), 248–258.
A. Keshavarz-Haddad, V. Ribeiro, and R. Riedi. 2006. Color-Based Broadcasting for Ad Hoc Networks. In
4th International Symposium on Modeling and Optimization in Mobile, Ad Hoc and Wireless Networks.
1–10.
M. Khabbazian, I.F. Blake, and V.K. Bhargava. 2012. Local Broadcast Algorithms in Wireless Ad Hoc Net-
works: Reducing the Number of Transmissions. Mobile Computing, IEEE Transactions on 11, 3 (2012),
402–413.
A. Kokuti and V. Simon. 2012. Adaptive multihop broadcast protocols for ad hoc networks. In 8th Interna-
tional Symposium on Communication Systems, Networks Digital Signal Processing (CSNDSP). 1–6.
D. Koscielnik and J. Stepien. 2011. The methods of broadcasting of information in ad-hoc wireless networks
with mobile stations. In Industrial Electronics (ISIE), 2011 IEEE International Symposium on. 2043–
2048.
S-H Lee and Y-B Ko. 2006. An efficient neighbor knowledge based broadcasting for mobile ad hoc networks.
In Proceedings of the 6th international conference on Computational Science - Volume Part II (ICCS’06).
Springer-Verlag, Berlin, Heidelberg, 1097–1100.
J. Leitao, J. Pereira, and L. Rodrigues. 2007. Epidemic Broadcast Trees. In 26th IEEE International Sympo-
sium on Reliable Distributed Systems. 301–310.
J. Leitao, J. Pereira, and L. Rodrigues. 2010. Gossip-Based Broadcast. In Handbook of Peer-to-Peer Network-
ing, Xuemin Shen, Heather Yu, John Buford, and Mursalin Akon (Eds.). Springer US, 831–860.
S. Leu and R-S Chang. 2011. Simple algorithm for solving broadcast storm in mobile ad hoc network. Com-
munications, IET 5, 16 (2011), 2356–2363.
D. Li, H. Huang, X. Li, M. Li, and F. Tang. 2007. A Distance-Based Directional Broadcast Protocol for Urban
Vehicular Ad Hoc Network. In Wireless Communications, Networking and Mobile Computing, 2007.
WiCom. International Conference on. 1520 –1523.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:32 P. Ruiz and P. Bouvry
D. Li, D. Kim, Q. Zhu, L. Liu, and W. Wu. 2012. Minimum Total Communication Power Connected Dominat-
ing Set in Wireless Networks. In Wireless Algorithms, Systems, and Applications, Xinbing Wang, Rong
Zheng, Tao Jing, and Kai Xing (Eds.). Lecture Notes in Computer Science, Vol. 7405. Springer Berlin
Heidelberg, 132–141.
F. Li and I. Nikolaidis. 2001. On minimum-energy broadcasting in all-wireless networks. In 26th Annual
IEEE Conference on Local Computer Networks. 193–202.
F. Li, S. Z., X. Wang, X. Xue, and H. Shen. 2004. Vote-based Clustering Algorithm in Mobile Ad-hoc Networks.
In International Conference on Networking Technologies for Broadband and Mobile Networks (Lecture
Note in Computer Science). 13–23.
N. Li and J.C. Hou. 2004. BLMST: a scalable, power-efficient broadcast algorithm for wireless networks.
In First International Conference on Quality of Service in Heterogeneous Wired/Wireless Networks,
QSHINE. 44–51.
N. Li, J.C. Hou, and S. Lui. 2003. Design and analysis of an MST-based topology control algorithm. In Twenty-
Second Annual Joint Conference of the IEEE Computer and Communications (INFOCOM), Vol. 3. 1702–
1712.
N. Li, J.C. Hou, and S. Lui. 2005. Design and analysis of an MST-based topology control algorithm. IEEE
Transactions on Wireless Communications 4, 3 (2005), 1195–1206.
X. Li, T. D. Nguyen, and R. P. Martin. 2003. Using adaptive range control to optimize 1-hop broadcast
coverage in dense wireless networks. In SenSys. 314–315.
O. Liang. 2007. Multipoint Relay and Connected Dominating Set Based Broadcast Algorithms for Wireless
Ad Hoc Networks. Ph.D. Dissertation. Melbourne, Australia.
W. Liang. 2002. Constructing minimum-energy broadcast trees in wireless ad hoc networks. In Proceedings
of the 3rd ACM international symposium on Mobile ad hoc networking & computing (MobiHoc ’02).
112–122.
D. Liarokapis and A. Shahrabi. 2009. A probability-based adaptive scheme for broadcasting in MANETs. In
Proceedings of the 6th International Conference on Mobile Technology, Application &Systems (Mobility
’09). Article 46, 8 pages.
D. Liarokapis and A. Shahrabi. 2011. Fuzzy-based probabilistic broadcasting in Mobile Ad Hoc Networks.
In Wireless Days (WD), 2011 IFIP. 1–6.
D. Liarokapis, A. Shahrabi, and A. Komninos. 2009. DibA: An Adaptive Broadcasting Scheme in Mobile Ad
Hoc Networks. In Communication Networks and Services Research. IEEE Computer Society, 224–231.
H. Lim and C. Kim. 2000. Multicast tree construction and flooding in wireless ad hoc networks. In Pro-
ceedings of the 3rd ACM international workshop on Modeling, analysis and simulation of wireless and
mobile systems (MSWIM ’00). ACM, New York, NY, USA, 61–68.
S. Lim, C. Yu, and C.R. Das. 2009. RandomCast: An Energy-Efficient Communication Scheme for Mobile Ad
Hoc Networks. IEEE Transactions on Mobile Computing 8, 8 (2009), 1039–1051.
J. Lipman, P. Boustead, J. Chicharo, and J. Judge. 2003. Resource aware information dissemination in ad
hoc networks. In The 11th IEEE International Conference on Networks. 591–596.
J. Lipman, P. Boustead, and J. Judge. 2002. Utility-based multipoint relay flooding in heterogeneous mo-
bile ad hoc networks. In Proceedings of the Workshop on the Internet, Telecommunications and Signal
Processing. 27–33.
J. Lipman, H. Liu, and I. Stojmenovi´c. 2009. Guide to Wireless Ad Hoc Networks. Springer, Chapter Broad-
cast in Ad hoc Networks. Chapter 6, 121–150.
W. Lou and J. Wu. 2002. On reducing broadcast redundancy in ad hoc wireless networks. IEEE Transactions
on Mobile Computing 1, 2 (2002), 111–122.
W. Lou and J. Wu. 2004. Double-covered broadcast (DCB): a simple reliable broadcast algorithm in MANETs.
In Twenty-third AnnualJoint Conference of the IEEE Computer and Communications Societies (INFO-
COM), Vol. 3. 2084–2095.
C. Maihofer. 2004. A survey of geocast routing protocols. IEEE Communications Surveys Tutorials 6, 2
(Second 2004), 32–42. DOI:http://dx.doi.org/10.1109/COMST.2004.5342238
B. Mans and N. Shrestha. 2004. Performance evaluation of approximation algorithms for multipoint re-
lay selection. In Med-Hoc-Net 2004, The Third An- nual Mediterranean Ad Hoc Networking Workshop.
Bodrum, Turkey, 480–491.
K. Miyao, H. Nakayama, N. Ansari, and N. Kato. 2009. LTRT: An efficient and reliable topology control
algorithm for ad-hoc networks. IEEE Transactions on Wireless Communications 8, 12 (2009), 6050–
6058.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:33
I.-H. M. Mkwawa and D.D. Kouvatsos. 2011. Broadcasting Methods in MANETS: An Overview. In Network
Performance Engineering, DemetresD. Kouvatsos (Ed.). Lecture Notes in Computer Science, Vol. 5233.
Springer Berlin Heidelberg, 767–783.
Y. Mylonas, M. Lestas, and A. Pitsillides. 2008. Speed adaptive probabilistic flooding in cooperative emer-
gency warning. In Proceedings of the 4th Annual International Conference on Wireless Internet. 1–7.
M. Ni, Z. Zhong, K. Wu, and D. Zhao. 2010. An Energy Efficient Clustering Scheme for Mobile Ad Hoc
Networks. In IEEE 71st Vehicular Technology Conference (VTC 2010-Spring). 1–5.
S-Y Ni, Y-C Tseng, Y-S Chen, and J-P Sheu. 1999. The broadcast storm problem in a mobile ad hoc network.
In 5th annual ACM/IEEE international conference on Mobile computing and networking. 151–162.
F. G. Nocetti, J. S. Gonzalez, and I. Stojmenovic. 2003. Connectivity based k-hop clustering in wireless
networks. Telecommunication systems 22, 1-4 (2003), 205–220.
R. Oda, T. Ohta, and Y. Kakuda. 2010. A Transmission Power Control Adaptive to Variation of Node Density
in Mobile Ad Hoc Network Clustering. In Second International Conference on Communication Software
and Networks. 141–146.
F.J. Ovalle-Mart´ınez, A. Nayak, I. Stojmenovi´c, J. Carle, and D. Simplot-Ryl. 2006. Area based beaconless
reliable broadcasting in sensor networks. In Proceedings of the Second international conference on Al-
gorithmic Aspects of Wireless Sensor Networks. Springer-Verlag, 140–151.
S. Park and S-M Yoo. 2013. An efficient reliable one-hop broadcast in mobile ad hoc networks. Ad Hoc
Networks 11, 1 (2013), 19 – 28.
V. Paruchuri, A. Durresi, and R. Jain. 2003. Optimized Flooding Protocol for Ad hoc Networks. CoRR
cs.NI/0311013 (2003).
W. Peng and X. Lu. 2000. On the reduction of broadcast redundancy in mobile ad hoc networks. In Mobile
and Ad Hoc Networking and Computing, 2000. MobiHOC. 2000 First Annual Workshop on. 129–130.
W. Peng and X. Lu. 2001. Efficient Broadcast in Mobile ad hoc Networks Using Connected Dominating sets.
12, 4 (2001), 1265–1275.
W. Peng and X-C LU. 2001. AHBP: An efficient broadcast protocol for mobile Ad hoc networks. Journal of
Computer Science and Technology 16 (2001), 114–125. Issue 2.
A. Piyatumrong. 2010. Efficient and Robust Spanning Forest for Topology Management in DT-MANETs: A
Tree-Based Decentralized and Multi-Objective Approach. Ph.D. Dissertation. University of Luxembourg,
Luxembourg.
A. Piyatumrong, P. Bouvry, F. Guinand, and K. Lavangnananda. 2008. Trusted Spanning Tree for Delay
Tolerant MANETs. In EUC (2). 293–299.
A. Qayyum, L. Viennot, and A. Laouiti. 2000. Multipoint Relaying: An Efficient Technique for flooding in
Mobile Wireless Networks. Technical Report 3898. INRIA.
S. Radhakrishnan, G. Racherla, C.N. Sekharan, N.S.V. Rao, and S.G. Batsell. 1999. DST-A routing protocol
for ad hoc networks using distributed spanning trees. In IEEE Wireless Communications and Network-
ing Conference (WCNC). 1543–1547 vol.3.
D.G. Reina, S.L. Toral, P. Johnson, and F. Barrero. 2015. A survey on probabilistic broadcast schemes for
wireless ad hoc networks. Ad Hoc Networks 25, Part A, 0 (2015), 263 – 292.
H. Reumerman and M. Runi. 2005. Distributed power control for reliable broadcast in inter-vehicle commu-
nication systems. In 2nd International Workshop on Intelligent transportation.
P. Ruiz and P. Bouvry. 2010a. Distributed Energy Self-Adaptation in ad hoc Networks. In In proc. of IEEE
Int. workshop on Management of Emerging Networks and Services (MENS), in conjunction with IEEE
Globecom. Miami, USA, 539–543.
P. Ruiz and P. Bouvry. 2010b. Enhanced Distance Based Broadcasting Protocol with Reduced Energy Con-
sumption. In Workshop on Optimization Issues in Energy Efficient Distributed Systems (OPTIM), part of
the 2010 International Conference on High Performance Computing and Simulation (HPCS). 249–258.
P. Ruiz, B. Dorronsoro, P. Bouvry, and L. J. Tard´on. 2012. Information Dissemination in VANETs based upon
a Tree Topology. Journal of Ad hoc Networks 10 (2012), 111–127. Issue 1.
P. Ruiz, B. Dorronsoro, D. Khadraoui, P. Bouvry, and L. Tard´on. 2008. BODYF–A Parameterless Broadcast-
ing Protocol Over Dynamic Forest. In Proceedings of the High Performance Computing & Simulation
Conference (HPCS 08). Nicosia, Cyprus, 297–303.
J. Schleich. 2010. Robust Dominating Set based Virtual Backbones for Wireless Ad hoc Networks. Ph.D.
Dissertation. University of Luxembourg, Luxembourg.
J. Schleich, G. Herbiet, P. Ruiz, P. Bouvry, J. Wagener, P. Bicheler, F. Guinand, and S. Chaumette. 2011.
Enhancing the Broadcast Process in Mobile Ad hoc Networks using Community Knowledge. In Proceed-
ings of the 1st ACM International Symposium on Design and Analysis of Intelligent Vehicular Networks
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
X:34 P. Ruiz and P. Bouvry
and Applications in conjunction with the 14th ACM International Conference on Modeling, Analysis and
Simulation of Wireless and Mobile Systems. Miami, 23–30.
J. Schleich, A. Le Thi Hoai, and P. Bouvry. 2012. Solving the minimum M-dominating set problem by a
continuous optimization approach based on DC programming and DCA. Journal Combinatorial Opti-
mization 24, 4 (2012), 397–412.
M. Slavik and I. Mahgoub. 2010. Stochastic Broadcast for VANET. In Consumer Communications and Net-
working Conference (CCNC), 7th IEEE. Las Vegas, 1 –5.
M. Slavik, I. Mahgoub, and F.N. Sibai. 2011. The distance-to-mean broadcast method for vehicular wireless
communication systems. In International Conference on Innovations in Information Technology (IIT).
371–374.
I. Stojmenovi´c. 2012. A General Framework for Broadcasting in Static to Highly Mobile Wireless Ad hoc,
Sensor, Robot and Vehicular Networks. In IEEE 18th International Conference on Parallel and Dis-
tributed Systems (ICPADS). 8–13.
I. Stojmenovi´c, A. Khan, and N. Zaguia. 2012. Broadcasting with seamless transition from static to highly
mobile wireless ad hoc, sensor and vehicular networks. International Journal of Parallel Emergent and
Distributed Systems 27, 3 (June 2012), 225–234.
I. Stojmenovi´c, M. Seddigh, and J. Zunic. 2002. Dominating sets and neighbor elimination-based broad-
casting algorithms in wireless networks. IEEE Transactions on Parallel and Distributed Systems 13, 1
(2002), 14–25.
I. Stojmenovi´c and J. Wu. 2004. Broadcasting and activity scheduling in ad hoc networks. Eds. S. Basagni
and M. Conti and S. Giordano and I. Stojmenovi´c. 105–139 pages.
J. Sucec and I. Marsic. 2000. An efficient distributed network-wide broadcast algorithm for mobile ad hoc
networks. Technical Report 248. Rutgers University, CAIP.
M.-T. Sun and T-H Lai. 2002. Location aided broadcast in wireless ad hoc network systems. In IEEE Wireless
Communications and Networking Conference (WCNC), Vol. 2. 597–602.
O.K. Tonguz, N. Wisitpongphan, and Fan B. 2010. DV-CAST: A distributed vehicular broadcast protocol for
vehicular ad hoc networks. Wireless Communications, IEEE 17, 2 (april 2010), 47 –57.
O. Tonguz, N. Wisitpongphan, F. Bai, P. Mudalige, and V. Sadekar. 2007. Broadcasting in VANET. In Mobile
Networking for Vehicular Environments. 7 –12.
Y.-C. Tseng, S.-Y. Ni, and E.-Y. Shih. 2002. Adaptive Approaches to Relieving Broadcast Storms in a Wireless
Multihop Mobile Ad Hoc Network. IEEE Trans. Comput. 52 (2002), 545–557.
K. Viswanath and K. Obraczka. 2002. An adaptive approach to group communications in multi hop ad hoc
networks. In Seventh International Symposium on Computers and Communications. 559–566.
E. Vollset and P. Ezhilchelvan. 2003. A survey of reliable broadcast protocols for mobile ad-hoc networks.
Technical Report CS-TR-792. University of Newcastle. http://books.google.lu/books?id=yprkMgEACAAJ
P-J Wan, G. Calinescu, X.-Y. Li, and O. Frieder. 2001. Minimum-energy broadcast routing in static ad hoc
wireless networks. In INFOCOM 2001. Twentieth Annual Joint Conference of the IEEE Computer and
Communications Societies. Proceedings. IEEE, Vol. 2. 1162–1171.
H-C Wang, J-B Lin, F-C Kuo, and K-C Ting. 2010. Combination of Relative Neighborhood Graph and For-
bidden Set in the Design of Distributed Broadcast Algorithms for Wireless Ad Hoc Networks. In In-
ternational Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery (CyberC).
94–100.
J.E. Wieselthier, G.D. Nguyen, and A. Ephremides. 2000. On the construction of energy-efficient broadcast
and multicast trees in wireless networks. In The nineteenth Annual Joint Conference of the IEEE Com-
puter and Communications Societies (INFOCOM), Vol. 2. 585–594.
J.E. Wieselthier, G.D. Nguyen, and A. Ephremides. 2002. The energy efficiency of distributed algorithms
for broadcasting in ad hoc networks. In Wireless Personal Multimedia Communications, 2002. The 5th
International Symposium on, Vol. 2. 499–503.
B. Williams and T. Camp. 2002. Comparison of Broadcasting Techniques for Mobile Ad Hoc Networks. In
Proceedings of the ACM International Symposium on Mobile Ad Hoc Networking and Computing (MO-
BIHOC). 194–205.
N. Wisitpongphan, O.K. Tonguz, J.S. Parikh, P. Mudalige, F. Bai, and V. Sadekar. 2007. Broadcast storm
mitigation techniques in vehicular ad hoc networks. Wireless Communications, IEEE 14, 6 (December
2007), 84 –94.
W. Woon and K.L. Yeung. 2011. Variable power broadcasting based on local information for source-dependent
broadcasting protocols. In IEEE Wireless Communications and Networking Conference (WCNC). 796–
801.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
Survey on Broadcast Algorithms for MANETs X:35
J. Wu. 2002. Extended dominating-set-based routing in ad hoc wireless networks with unidirectional links.
IEEE Transactions on Parallel and Distributed Systems 13, 9 (2002), 86–881.
J. Wu, M. Gao, and I. Stojmenovic. 2000. On Calculating Power-Aware Connected Dominating Sets for Effi-
cient Routing in Ad Hoc Wireless Networks. IEEE/KICS Journal of Communications and Networks 4
(2000), 59–70.
J. Wu and H. Li. 1999. On calculating connected dominating set for efficient routing in ad hoc wireless
networks. In Proceedings of the 3rd international workshop on Discrete algorithms and methods for
mobile computing and communications (DIALM ’99). 7–14.
J. Wu and W. Lou. 2003a. Forward-node-set-based broadcast in clustered mobile ad hoc. Wireless Communi-
cations and Mobile Computing 2, 3 (2003), 155–173.
J. Wu and W. Lou. 2003b. Forward-node-set-based broadcast in clustered mobile ad hoc networks. Wireless
Communications and Mobile Computing 3, 2 (2003), 155–173.
J. Wu, W. Lou, and F. Dai. 2006. Extended Multipoint Relays to Determine Connected Dominating Sets in
MANETs. IEEE Trans. Comput. 55, 3 (March 2006), 334–347.
J. Wu and B. Wu. 2003. A transmission range reduction scheme for power-aware broadcasting in ad hoc
networks using connected dominating sets. In IEEE Vehicular Technology Conference (VCT2003.
H. Xu and J.J. Garcia-Luna-Aceves. 2010. Efficient broadcast for wireless ad hoc networks with a realistic
physical layer. Ad hoc Networks 8 (2010), 165–180.
Y. Yi, M. Gerla, and T-J Kwon. 2003. Efficient Flooding in Ad hoc Networks: a Comparative Performance
Study. In IEEE International Conference on Communications (ICC’03). 1059–1063.
J.Y. Yu and P. H J Chong. 2005. A survey of clustering schemes for mobile ad hoc networks. IEEE Commu-
nications Surveys Tutorials 7, 1 (2005), 32–48.
J. Yu, N. Wang, G. Wang, and D. Yu. 2013. Connected dominating sets in wireless ad hoc and sensor networks
– A comprehensive survey. Computer Communications 36, 2 (2013), 121–134.
H. Zeng, M. Li, H. Liu, and X. Jia. 2009. Efficient Flooding in Mobile Ad Hoc Networks. In Ad Hoc Networks:
New Research, J.N. Turner and C.S. Boyer (Eds.). Nova Science Publishers.
L. Zhou, G. Cui, H. Liu, Z.Wu, and D. Luo. 2010. NPPB: A Broadcast Scheme in Dense VANETs. Information
Technology Journal 9, 2 (2010), 247 –256.
ACM Computing Surveys, Vol. X, No. X, Article X, Publication date: June 2015.
