Professional Documents
Culture Documents
Abstract—Multi-hop wireless broadcast is an important component in means of communication, so solving the broadcast storm
vehicular networks. Many applications are built on broadcast communi- problem in an efficient way is critical.
cations, so efficient routing methods are critical for their success. Here The distribution of nodes in VANETs can be highly
we develop the Distribution-Adaptive Distance with Channel Quality
variable. Network density can range from one or two
(DADCQ) protocol to address this need and show that it performs well
compared to several existing multi-hop broadcast proposals. neighbors to very high densities. Dedicated Short Range
The DADCQ protocol utilizes the distance method to select forward- Communications (DSRC), a layer 1 and 2 standard
ing nodes. The performance of this method depends heavily on the based on IEEE 802.11 designed for vehicle-to-vehicle and
value of the decision threshold, but it is difficult to choose a value vehicle-to-infrastructure communications, allows trans-
that results in good performance across all scenarios. Node density, mission ranges up to 1000m. Even with a more modest
spatial distribution pattern, and wireless channel quality all affect the
transmission radius of 200m, the number of vehicles in
optimal value. Broadcast protocols tailored to vehicular networking must
be adaptive to variation in these factors.
range of any point on a 4-lane road with cars separated
In this work we address this design challenge by creating a decision by 20m is 80.
threshold function that is simultaneously adaptive to the number of The distribution pattern, independent of density, can
neighbors, the node clustering factor, and the Rician fading parameter. also vary. In isolated highways, nodes will be con-
To calculate the clustering factor, we propose using the quadrat method strained along one dimensional paths. In an urban area,
of spatial analysis. The resulting DADCQ protocol is then verified with nodes may appear more uniformly distributed in two
JiST/SWANS and shown to achieve high reachability and low bandwidth
consumption in urban and highway scenarios with varying node density
dimensions. Road crossings and suburban areas exhibit
and fading intensity. many combinations in between. Broadcast protocols
should be able to function equally well in all these
Index Terms—Wireless broadcast, statistical broadcast, broadcast scenarios.
storm, VANET Another problem in wireless communications gener-
ally and VANET in particular is loss of packets as they
traverse the medium. Multi-path fading causes a signal
1 I NTRODUCTION to interfere with itself as it splits into multiple paths
Multi-hop wireless broadcast plays an important role in due to being reflected off objects in the environment.
the operation of wireless systems. Many point-to-point Packets can also be lost when different transmitted
ad-hoc networking protocols, such as AODV and DSR, signals interfere with each other, a phenomenon called
use broadcast to discover routes. Research in this area a collision. These effects vary in intensity across space
has demonstrated that blindly retransmitting broadcast and time and can work against the design of multi-hop
packets (flooding) can lead to an explosive growth of broadcast protocols. If a broadcast protocol is designed
traffic called the broadcast storm problem [1], that attenu- and tested in an environment that assumes perfectly
ates broadcast performance as a result of collisions and reliable communications, it could break down when
congestion. fading and collisions are introduced.
Further, emerging classes of networks use broadcast as Often, broadcast protocols must accommodate packet
a primary data delivery mechanism. Vehicular Ad-hoc communication failures by increasing the number of
NETworks (VANETs) are a class of wireless networks nodes that are rebroadcasting source messages. Ideally,
in which mobile vehicles communicate with each other this increased usage of bandwidth is the minimum nec-
in an ad-hoc fashion. VANET applications such as traf- essary to retain the level of connectivity required by the
fic data dissemination use broadcast exclusively as the application. Practical VANET broadcast protocols should
accommodate fading and collisions while minimizing
M. Slavik and I. Mahgoub are with the Department of Computer and
Electrical Engineering and Computer Science, Florida Atlantic University, redundant source message retransmissions.
Boca Raton, FL, 33431 USA e-mail: mslavik@fau.edu, mahgoubi@fau.edu. Wireless broadcasting methods can be classified into
at least statistical and topological methods [2]. Topo- It is possible to design a statistical protocol, as we do
logical protocols use the network topology (knowledge in this work, that shows high efficiency while achieving
of the identity of nodes in the 2-hop neighborhood) to desired reachability. The key to accomplishing this is in
select nodes to forward broadcast messages. The key the design of the threshold value. If the threshold value
advantages of topological protocols are that they reduce is too aggressive, the protocol will give low coverage.
broadcast redundancy in a way that is agnostic to node If it is too conservative, the broadcast efficiency will
distribution and they can be proven to not degrade be low. An optimal value of the threshold must be
reachability compared to flooding, provided communi- found that gives the best efficiency possible without
cations are perfectly reliable [3]. sacrificing reachability. Further, the optimal value of the
However, topological protocols have disadvantages. threshold varies with node density, distribution pattern,
The density and speed of vehicles in VANET means that and channel quality. A main objective of this work is to
the network topology will be highly variable. Consider demonstrate how to design a threshold function that is
the rather modest scenario of a single 2-lane street in adaptive to a wide range of these factors. This threshold
isolation with vehicles traveling in both directions at function is then used in the proposed DADCQ protocol.
a speed of 50 kph (32 mph) and separated by 20m. The contributions of this work are summarized as
Half of each node’s neighbors (the ones traveling in the follows:
same direction) are roughly constant. However, the set
of neighbors traveling in the opposite direction changes • DADCQ Protocol: The primary contribution of this
every 715 ms. If a protocol needs accurate neighborhood work is the proposed multi-hop broadcast proto-
information to operate properly, messages used to dis- col DADCQ. DADCQ combines local spatial dis-
cover the topological data will need to be sent at a high tribution information and other factors with the
rate to keep the local neighborhood information up-to- distance method heuristic to select rebroadcasting
date. nodes. Previous broadcast protocols proposed for
Statistical broadcast protocols typically do not use this ad-hoc that make use of the distance method use
rapidly changing neighborhood information. Statistical less comprehensive supplemental information. A
protocols measure the value of one or more locally key insight proposed here is a methodology for
available variables and make a decision to rebroadcast incorporating more information into the protocol.
based on the measured value and a cutoff threshold. For This extra information is used to make the protocols
example, the statistical method used in this work, the adaptive to more networking scenarios than many
distance method, measures the distance to the nearest previous proposals.
neighbor from which a node has received the broadcast • Adaptive Threshold Function Design: This paper gives
message. If that distance is greater than a threshold a novel design strategy for a decision threshold
value, then the node rebroadcasts the message. The function. Threshold functions are a critical com-
threshold value is calculated as a function of more slowly ponent of many multi-hop broadcasting methods,
changing topological factors such as node density and such as stochastic broadcast (gossiping), the counter
spatial distribution pattern. method, the distance method, and the location
Another qualitative advantage of statistical methods method [4]. The proposed design scheme builds
is with respect to privacy. Privacy is a critical aspect a threshold function using three independent in-
of vehicular networking systems. Data broadcasted in put variables chosen to allow the threshold to be
VANET applications, such as vehicle speed and location, adaptive to the environmental conditions of pri-
must not be traceable. Therefore security mechanisms mary interest. These variables measure local node
need to be built in to obscure or hide node identifica- density, the distribution pattern of nearby nodes,
tion. Multi-hop broadcast protocols built using statistical and the wireless channel quality. Thus the resulting
methods ease the development of these mechanisms, threshold function of these three inputs causes the
because statistical methods generally do not require protocol to operate efficiently across a broad range
any neighborhood information and thus node identifiers of conditions.
are not required to be sent with data and overhead • Quadrat Method for Spatial Distribution Characteriza-
messages. By contrast, topological methods use node tion: One of the factors used to compute the rebroad-
identifiers extensively. Privacy preserving schemes may casting decision threshold in DADCQ is local node
still be required when using statistical methods, but the distribution pattern. This work is the first to pro-
statistical method will complement and reinforce the pose that the quadrat method of spatial statistics be
privacy mechanism. used to characterize the spatial distribution of nodes
The downside of statistical protocols is that they are for use in a multi-hop broadcast protocol. Because
stochastic in nature and usually cannot be proven to al- distribution pattern may affect the behavior of many
ways give full reachability. Instead, simulation modeling multi-hop broadcast methods, this contribution may
is done to verify that the protocols exhibit acceptable per- be applicable in a wider context as well.
formance in a target percentage of randomly distributed • Analysis of Behavior with Respect to Threshold Value:
networks. This paper presents original results addressing the
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
IEEE TRANSACTIONS ON MOBILE COMPUTING
3
nature of the threshold value used in the distance gives the best performance so we use it as part of our
method. We show that system performance exhibits evaluations here. A more comprehensive study of these
a phase transition with respect to the threshold methods in highway scenarios is rolled into a scheme
value and suggest that an optimal threshold value called DV-CAST [16]. DV-CAST has been shown to
should cause the system to operate in the supercrit- have performance issues in urban environments, so the
ical region of performance as close to the transition authors propose a variant called UV-CAST designed for
region as possible. urban environments [17]. Our work focuses on building
In Section 2 we will discuss related work to describe a multi-hop broadcast protocol that performs well in a
where this work fits into the literature. Section 3 de- wide range of node distribution patterns.
scribes the distance method and statistical broadcast Bako et al adopted a stochastic broadcast method orig-
protocol design generally. Sections 4 and 5 show how inally proposed in [18] for VANET [19]. From that body
to build adaptivity to distribution pattern and channel of work we use Advanced Adaptive Gossiping (AAG)
quality into a protocol that uses the distance method. in our evaluations. AAG is a hybrid protocol that uses
Section 6 describes multi-hop broadcast protocols that 2-hop topological data to set forwarding probabilities
will be used for comparison. Section 7 discusses the used as in a traditional stochastic broadcast algorithm.
collection of supplemental information used by protocols p-persistence and AAG are described in more detail in
to make rebroadcasting decisions. Section 8 presents sim- Section 6. Simulation results comparing these protocols
ulation results and Section 9 gives concluding remarks. to the proposed solution are given in Section 8.
Osafune et al propose MHVB [20] and E-MHVB [21],
which use the distance method with constant threshold
2 R ELATED W ORK value to suppress rebroadcasts. The proposed protocol,
Multi-hop broadcast in VANET mainly supports two DADCQ, also makes use of the distance method, but
types of broadcast applications. One type is safety ap- with a highly adaptive threshold value. Finally, Korkmaz
plications such as Collective Collision Avoidance (CCA) et al propose Urban Multi-hop Broadcast (UMB) [22],
where in the event of an emergency an alert message is which like the CCA broadcast protocols is fundamentally
propagated as quickly as possible into oncoming traffic. uni-directional, but does contain provisions for branch-
Broadcast protocols for these applications are usually ing at intersections to cover wider areas.
designed to propagate messages in one or two directions Multi-Point Relaying (MPR) [23] is a well-known
with high reliability and minimum latency. Examples of broadcast protocol in MANET that uses network topol-
these proposals are [5], [6], and [7]. ogy information to select rebroadcasting nodes. An ex-
The focus of this work is the second type of VANET tension of MPR that is more accommodating to high
broadcast application such as traffic data dissemination. mobility called E-MPR is proposed in [24]. We use
These applications require data to be disseminated to a another variation of MPR, called Double Covered Broad-
large area as efficiently as possible, with less stringent cast (DCB) [3], in our evaluations here because it was
reliability and delay requirements than CCA-type appli- designed to accommodate transmission errors, which is
cations. Protocols to support these applications are more one of the topics of this paper.
like traditional MANET broadcast protocols, except they
must handle high node mobility, variable node distribu-
tion, and poor channel quality of VANET environments.
3 S TATISTICAL B ROADCAST P ROTOCOL D E -
Multi-hop broadcast protocols can generally be clas- SIGN
sified into statistical and topological protocols by the The following three sections (3, 4, and 5) guide the reader
type of information used to select forwarding nodes. through the design of the proposed protocol, DADCQ.
The fundamental statistical methods (stochastic, counter, First, we give background information on a heuristic
distance, and location), are described in [1]. Several pro- used by DADCQ called the distance method, including
tocols have been built using these methods that are adap- an original analysis of system behavior with respect to
tive to node density. Several papers have addressed the the forwarding decision threshold value. Also included
need to make stochastic broadcast adaptive to density in this discussion is an overview of the general threshold
[8] [9]. Density-adaptive versions of counter are given value design procedure and WiBDAT, a specialized tool
in [10] [11]. Density-adaptive versions of both counter we designed for the purpose of statistical broadcast
and location methods are given in [4]. We described a protocol design [25].
method using the link between continuum percolation Wireless broadcasting methods can be divided into at
and wireless broadcast to make stochastic broadcast least statistical and topological methods [2]. Statistical
adaptive to density in [12]. methods use a priori knowledge of the impact of an
For VANET in particular, several proposals for broad- external variable (examples listed below) present at each
cast protocols have been made. Tonguz et al, have pro- node to decide whether or not to rebroadcast. These
posed forwarding node selection algorithms based on methods share a common framework where nodes mea-
the fundamental statistical methods mentioned above sure the value of the variable at their location, calculate a
[13] [14] [15]. Of these, we have found p-persistence threshold value, and make a rebroadcast determination
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
IEEE TRANSACTIONS ON MOBILE COMPUTING
4
based on whether or not the value of the variable exceeds 4) When the backoff expires, rebroadcast if D > Dc
the value of the threshold. Several local variables have The key to any statistical broadcast protocol is the
been proposed for use with this framework, most due to decision cutoff, in this case Dc . If the value is set too
Ni et al [1]. Here we use a technique called the distance high, reachability will be degraded. If the value is set
method that is designed to select nodes to rebroadcast too low, the protocol will not prevent many nodes from
that will add the most coverage area (see Section 3.2). rebroadcasting. Thus protocol behavior exhibits a phase
transition relative to Dc [26]. In the supercritical phase,
3.1 Outline of Design Methodology when Dc is less than the critical value, reachability is
nearly one in all simulation runs. Around the critical
In the following sections we will present the design of value, the reachability quickly transitions from one to
an adaptable multi-hop broadcast protocol built around zero and the variance of the reachability spikes, indicat-
a statistical broadcast method. We begin with the basic ing the reachability is highly variable from run to run. In
distance method heuristic, then build in adaptability to the subcritical phase, message propagation is suppressed
node distribution pattern and channel fading intensity so that only nodes near the source will receive the
to create the DADCQ protocol. The behavior of distance message.
method protocols is largely determined by a particular The design trade-off is with bandwidth consumption.
protocol parameter called Dc . The following sections will When Dc is set deep into supercritical phase, the protocol
develop a function to map locally measured variables will propagate messages to the entire network with high
to experimentally determined good values for Dc . This reliability, but it comes at a cost of decreased broadcast
function is referred to throughout the paper as the efficiency. Our goal then is to set Dc as high as possible
threshold function. To reduce design time, we use a high- while maintaining acceptable reachability.
level simulator called WiBDAT [25] (discussed further
This optimal value for Dc varies with node density
below). It is important to note that this tool produces less
(among several factors). At low densities, Dc must be
accurate results than a full network simulator such as ns-
lower to allow the message to propagate to the entire
2 (which in turn produces less accurate than real-world
network. When the node density is high, we can set
experiments), so all values of Dc presented below are
Dc more aggressively to eliminate excess transmissions.
approximations only. Further, the functional forms and
Therefore, we use the following procedure to find Dc as
function parameters selected to model those values of Dc
a function of N , the number of neighbors of the node
(e.g., Equation 1) are themselves only approximations,
(also referred to as local density).
typically chosen to balance computational and curve
fitting complexity with accuracy in fitting the estimated 1) Find the optimal value of Dc at many values of
values. The result is that the final threshold function for the global density λ. The optimal value of Dc is the
DADCQ is clearly only a rough model of the “true” maximum value for which the reachability remains
optimal values of Dc . In the simulation results section acceptable
(Section 8), we use JiST/SWANS, a full network simula- 2) Plot the measured optimal values and find suitable
tor, to verify the protocol in as accurate an environment approximation function Dc (λ).
as possible and gauge the performance of the threshold 3) Substitute local density N for global density λ to
function built using these approximations. get Dc (N ).
4) Test Dc (N ) to ensure reachability is acceptable.
5) Adjust Dc (N ) as necessary.
3.2 Distance Method Here we define acceptable reachability as broadcast
At its core, the proposed DADCQ protocol utilizes the messages are propagated to at least 99% of all nodes
distance method to select forwarding nodes. The dis- 95% of the time. In previous work, we found that using
tance method uses the minimum distance from sender an exponential curve for Dc (N ) provided good perfor-
to receiver (1-hop distance) as the variable of discrim- mance across the range of N [2] [27]. This determination
ination between rebroadcasters and non-rebroadcasters. was made by empircally finding optimal values of Dc vs
The method appeals to the intuition that if a node has N while holding node distribution pattern and channel
received a message from another node very close to it, quality constant. For the range of node distributions and
there is little benefit in terms of additional coverage channel qualities reasonable in VANET (see Section 8)
achieved by rebroadcasting. Nodes then should favor we find that the curve of Dc vs N roughly follows an
rebroadcasting when this distance is large. exponential form. This indicates that for low node den-
The algorithm is simple and is given here. sities, the threshold should be low to allow many nodes
1) Initialize D = 1 to retransmit. As the density increases, the threshold can
2) When a message is received, set d to the distance quickly become more aggressive to prevent unnecessary
to the sender, D = min{D, d/r}, and set a random rebroadcasts. We therefore use the following general
backoff timer form for the threshold function.
3) If a message is received during the backoff, repeat
2 Dc (N ) = Dmax − βeαN (1)
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
IEEE TRANSACTIONS ON MOBILE COMPUTING
5
Fig. 2. Illustration of quadrat method statistic Q in 2-D fields. Small values of Q indicate even spacing while large
values of Q indicate clustering.
by that value to get the final α. Observe that the value among scattered multi-path signals. Note that others
of Q for 1-D networks closely follows this equation. have disputed using a standard fading model such as
Rician or Nakagami fading in VANET [32]. Clearly,
Q1D ≈ 1 + 0.125N (3) movement of vehicles causing temporary disruptions
in line-of-sight communications can have large fading
We can use this approximation to create a linear
effects that are not captured by these standard models.
mapping (Equation 4) that will range from α2D in pure
However, for our purposes we need to adopt a single
2-D networks to α2D /3 pure 1-D networks. Here α2D is
model of fading so that the fading intensity can be
the optimal α from the 2-D experiments (Table 1).
quantified and used to adapt the broadcast threshold
−1 value. The Rician model allows the fading intensity to be
2(Q − 1)
α = α2D +1 (4) easily quantified and has been accepted in the literature
0.125N
as a reasonable model of fading in VANET [31] [33].
When the network is 2-D distributed, Q ≈ 1 and A parameter, K, quantifies the relative strength of
α ≈ α2D . When the network is 1-D distributed, Q ≈ the line-of-sight signal in Rician fading. When K = 0,
0.125N + 1 and α ≈ α2D /3. These values roughly match also called Rayleigh fading, there is no strong line-of-
the desired values in Table 1. Other values of Q will be sight signal. When K = ∞, fading has no effect thus
mapped linearly between the pure 1-D and 2-D value. all nodes within transmission range of each other can
Using WiBDAT, we cannot easily generate random node reliably communicate.
distributions with arbitrary values of Q, so we do not Several studies have estimated the value of K in
have samples between the pure 1-D and pure 2-D cases. VANET scenarios. In [31], the value is measured in the
Thus the linear form is chosen to fill the gap between range 5 to 11 for communications in the 900 MHz band.
pure 1-D and pure 2-D due to its simplicity. The resulting At the 5.3 GHz carrier, researchers have measured K
equation for Dc is shown by Equation 5. between 8 and 13 [33]. Standardization of VANET com-
−1 munications is coalescing around IEEE 802.11p (DSRC),
Dc (N, Q) = 0.90 − 1.2e−0.099( 0.125N +1)
2(Q−1)
N
(5) which uses the 5.85 - 5.925 GHz band. Our evaluations
will use Rician fading with K from 0 to 20.
5 A DAPTATION TO C HANNEL Q UALITY
Finally, this section incorporates adaptability to channel 5.2 Designing a Threshold Function using K
quality into the proposed protocol. The design described In Section 4, we found good values of Dmax , α, and β
in the previous section assumes that the wireless channel when nodes were laid out in a line and in a rectangle,
and the medium access control protocol deliver mes- then fit the curve of Dc to these cases using Q to
sages between nodes located within transmission range differentiate them. We will extend that process here,
of each other with perfect reliability. In practice, wireless finding good values of Dc in 1-D and 2-D networks for
signals in the system interfere with themselves and with various values of K.
each other in unpredictable ways, leading to apparently Previously, we found that α in the 2-D scenario is
non-deterministic message reception. roughly 3 times the value in the 1-D scenario. We will
When two nodes transmit messages at the same time, reuse this result here and restrict the value of α in each
the wireless signals may interfere and cause one or both case to follow this pattern and then fit the value of Dmax
of the messages to be lost at a destination node. Fading, and β using least squares optimization. Table 2 shows the
the phenomenon where multiple parts of the same signal resulting values of the threshold function parameters for
traveling along different paths interfere with each other, average neighbors ranging from 8 to 30 in the 2-D case
degrades communications even when only a single node (Q ≈ 1) and 15 to 70 in the 1-D case (Q ≈ 3 to 10).
is transmitting. Next we need to find expressions for Dmax , α, and β
Multi-hop wireless broadcast protocols must be able to in terms of Q and K. Finding a perfect fit is difficult,
operate effectively even when communication reliability so we make two simplifications. First, the value of Dmax
is poor. What follows in this section is the design of a is roughly the same for the 1-D and 2-D cases, so Dmax
decision threshold function that is adaptive to both node can be independent of Q. Second, β is about 3 times
distribution pattern and channel quality, culminating in larger in the 2-D case than the 1-D case, so we can use
the Distribution-Adaptive Distance with Channel Qual- a similar linear mapping between them as we use for
ity (DADCQ) protocol. α. These simplifications allow us to use a simple one-
dimensional least squares curve fit to find expressions
5.1 Modeling Fading of Dmax and β. Results are shown in Equations 6 - 8.
First, we need a way to model multi-path fading in −1
2(Q − 1)
WiBDAT so that we can find good threshold values in the α = −0.200 +1 (6)
0.125N
presence of fading. In VANET, fading may be modeled as −1
Rician fading [31]. In Rician fading, the received signal is −0.1K 2(Q − 1)
β = 6.25e +1 (7)
assumed to have a strong component from line-of-sight 0.125N
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
IEEE TRANSACTIONS ON MOBILE COMPUTING
8
Fig. 4. Illustration of DADCQ rebroadcast decision surface. Subfigure caption indicates the parameter being held
constant in the plot.
value of one. When a node receives a message, it first simple greedy approach called the Forwarding Node Set
sets a backoff timer and waits to see if the message is Selection Process (FNSSP).
received again. If it is, the retransmission is canceled. If
not, the node rebroadcasts. 1) Initialize X = H1 (v), U = H2 (v), and F = ∅
Calculation of the backoff period involves two other 2) Find w in X with maximum degree deg(w) =
statistical methods. First, nodes decide if they are high |N1 (w) ∩ U |
priority rebroadcasters according to a fixed probability p. 3) Update F = F ∪ {w}, U = U − N1 (w), and X =
In the evaluations that follow, we follow Wisitpongphan X − {w}
et al [15] and use p = 0.5. Nodes that do not elect them- 4) Repeat 2,3 while U = ∅
selves as high priority rebroadcasters set a backoff time
of T = Ns τ , where Ns and τ are protocol parameters.
Finally, nodes that elect themselves to have high prior- DCB uses the FNSSP procedure with modified values
ity use one-hop distance to the sender to bias the backoff for X and U to achieve various goals. X can be seen as
timer, similar to the distance method. Specifically, the the set of nodes from which v chooses forwarding nodes,
backoff time is calculated as T = Ns (1 − d/r)τ , where and U is the set of nodes that need to be covered. Since
d is the distance to the node from whom the message v will cover H1 (v) by broadcasting itself, only nodes
was received and r is the transmission radius. in H2 (v) need to be covered by the forwarding nodes
designated by v. Note that as long as the network is
connected, this algorithm will guarantee all nodes in the
6.2 Double-Covered Broadcast network receive the broadcast message, provided there
The next two protocols, DCB and AAG, use network are no transmission errors.
topology data to select forwarding nodes. To describe
In practice, of course, there are often transmission er-
these topological protocols, we need to provide some
rors. DCB attempts to account for link errors by making
basic definitions (this notation is adopted from [3]).
the protocol more reliable. It does this in two ways.
Nk (v) is the set of all nodes within k hops of node v,
First, nodes use a double-coverage version of FNSSP to
including v itself. Hk (v) is the set of nodes exactly k
select forwarding nodes. Node v sets X = H1 (v) and
hops from v.
U = N2 (v) − {v} and then executes the FNSSP, which
Double-Covered Broadcast (DCB) [3] is based on
results in all nodes in H2 (v) being covered once by a
Multi-Point Relaying (MPR) [23]. MPR is a dominant
broadcast and all nodes in H1 (v) being covered twice.
pruning broadcast algorithm used as the broadcasting
Since each node will receive the message at least twice,
mechanism in the popular Optimized Link State Routing
they have a better chance of receiving it.
(OLSR) protocol [34]. In dominant pruning, forwarding
nodes, beginning with the source, select which of their Second, DCB introduces a retry mechanism. When
neighbors should be the next forwarding nodes (this forwarding node v rebroadcasts a message P , it will
stands in contrast to self pruning, where nodes decide have selected a set of forwarding nodes F (v) ⊆ H1 (v).
for themselves whether or not to rebroadcast). This After it transmits P , v starts a timer and waits to receive
forwarding node list is attached to each data message, messages. If it has not overheard the rebroadcast of
so when a node receives a message, it first checks if each of the forwarding nodes in F (v) by the time the
it is in the forwarding node list. If it is, it calculates timer expires, it retransmits the message P again. It will
which neighbors should also be forwarding nodes and continue this process until it overhears all the nodes in
retransmits the message. Otherwise, it does nothing. F (v) rebroadcast or it reaches the maximum of R retries.
In MPR, when forwarding node v receives a message, Following the suggestions of the authors of DCB, we use
it calculates the next set of forwarding nodes F using a the value R = 1 in all evaluations.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
IEEE TRANSACTIONS ON MOBILE COMPUTING
10
6.3 Advanced Adaptive Gossiping used here, AAG and DCB, require 2-hop neighborhood
Advanced Adaptive Gossiping (AAG) [19] is a hybrid information, so they transmit their ID and their most
between pure topological protocols like MPR and DCB recent list of neighbors in the beacon.
and a stochastic broadcast protocol. AAG uses 2-hop With DADCQ, beacons contain a location, which we
topological information to calculate a rebroadcast prob- define to be 12 bytes. We assume network layer ad-
ability p. dresses are four bytes long, so in AHBP and DCB, bea-
First, when node v receives a message from node u, cons contain 4 bytes for the node’s address and 4N bytes
let C(v) = H1 (v)−H1 (u) be the set of neighbors of v that for the addresses of all the node’s neighbors. Additional
are not covered by u, called the children of v. For each bandwidth is used by MAC and network layer packet
child node c ∈ C(v), v then determines the parent set of framing.
c, defined as P (c) = H1 (c) ∩ H1 (u), the set of nodes that If the network is static, the beacon interval Tb can be
could potentially cover c by rebroadcasting. very long and nodes will still have reasonably accurate
Next, for each child node ci , v calculates a forwarding topological data. However, if nodes are mobile, the bea-
probability pi = prob(|P (ci |) where prob is defined con interval needs to be reduced so that beacons received
below. The final forwarding probability used to make near the beginning of an interval are not stale by the end
the rebroadcasting decision is then calculated as p = of the interval. VANETs of course are not static, so Tb
max{pi }. Like any stochastic broadcast protocol, AAG needs to be short enough that vehicles maintain accurate
rebroadcasts the message if a uniformly random number information.
between 0 and 1 is less than p. We have found that a good value of Tb for DADCQ is
The prob(n) function used to calculate forwarding 30 seconds while for AAG and DCB is 1 second. The
probabilities in AAG is shown in Equation 9. This func- difference stems from the varying temporal nature of
tion depends on three parameters: τarp , δ, and D. τarp the external data needed by the protocols. AAG and
is an application defined reachability target, which we DCB use topological information, so they need up-to-
set to be 0.99. δ is the network diameter (in number of date data on the identities of nodes in their 2-hop
hops) and D is the number of neighbors of v. neighborhood. DADCQ requires measurements of the
node distribution pattern in the form of N and Q. We
find node distribution pattern changes more slowly than
1/δ 1/n
0.5 network topology, so overhead messaging in statistical
prob(n) = 1 − 1 − τarp − (9) protocols such as DADCQ can be lower than in topo-
1 + e−0.3(D−14)
logical ones like AAG and DCB.
Finally, it is currently unknown whether the over-
7 OVERHEAD B EACONING head messaging used by multi-hop broadcast protocols
Next we will describe the mechanism for communicating in VANET will be unique to the broadcast protocol
topological information between nodes, called beacon- or whether it will be shared across many components
ing. The algorithm is very simple, but results in efficient in the platform. If this information is available from
transfer of information in the network. Nodes broadcast other subsystems, then beacons should not be considered
a short message every Tb seconds called a beacon. Unlike additional overhead of the broadcast protocol. If the
the multi-hop broadcast messages that get propagated to information is not available, either because other systems
the entire network, beacons are broadcast from a node do not use the same amount of information (ex. 2-hop
to its neighbors only and are never retransmitted. Nodes topological data) or other systems use different transmit
send beacons at a constant fixed rate, so in any interval power, then beacons should be understood as overhead.
of Tb seconds, every node in the network will send Evaluations that follow present results to accommodate
exactly one beacon. The timing of beacons is in no way both perspectives.
synchronized between nodes, except that all nodes use
the same value of Tb . 7.1 Protocol Adaptability Rate
Individual nodes set a timer for Tb seconds and track Another important issue when considering the choice of
all the beacons received from neighbors during that time. Tb is the rate at which the protocol can adapt to sudden
At the end of the interval, they compile all the beacons changes in node density and distribution pattern. Nodes
and transmit their own beacon, which may contain continue to use collected topological information that is
information from the most recently received beacons. up to Tb seconds old, so in the worst case a node will not
The information contained in beacons depends on the be fully aware of a sudden change in topology for Tb sec-
propagation algorithm. p-persistence requires no addi- onds. However, beacon reception is randomly staggered
tional information and so does not send beacons at all. between all nodes, so if a vehicle transitions into an area
DADCQ needs to calculate N and Q, so nodes transmit with a dramatically different node distribution pattern, it
their current location in the beacon. Nodes estimate N will begin to receive updated information immediately.
by simply counting the number of beacons received in Its internal aggregate measurements of topology will
an interval and calculate Q using the relative locations slowly be updated and improve in accuracy until after
of all the received beacons. The topological protocols Tb seconds the topological information is up-to-date.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
IEEE TRANSACTIONS ON MOBILE COMPUTING
11
Fig. 6. Performance vs number of nodes of the broadcast protocols in the urban scenario
Fig. 7. Performance vs number of nodes of the broadcast protocols in the highway scenario
scenario, AAG and DCB select the most rebroadcasters achieves better than 95% reachability across all network
and transmit the most bytes. On the highway, DCB scenarios while showing efficient use of bandwidth. In
selects too few rebroadcasters and so the reachability most cases, protocols used for comparison either fail to
is low. p-persistence and DADCQ show similar band- match this level of reachability or do so using a higher
width consumption, with p-persistence being a little level of bandwidth consumption. There are some sce-
better across all results. Considering the reachability and narios where DADCQ performance could be improved,
bandwidth results together, DADCQ has an advantage, notably in very poor channel quality, and its use requires
showing high reachability while using a similar or less positional information, but overall it is shown to be
amount of bandwidth than the other protocols. highly adaptable and efficient.
The performance of protocols such as DADCQ that
8.2 Effect of Fading Intensity use statistical broadcasting methods like the distance
method is highly dependent on the rebroadcasting deci-
The second set of results investigates the performance of
sion threshold function. In the past, this dependency has
the broadcast protocols with different levels of fading,
been a key disadvantage of statistical protocols, as it is
specifically the Rician fading parameter K. Here the
difficult to choose a value for the decision threshold that
number of nodes is fixed at 250, the urban scenario
works well across a wide range of network scenarios.
maximum speed is set to 15 m/s, the highway maximum
In this work we addressed this challenge directly and
speed is set to 30 m/s. The fading parameter K is varied
proposed a methodology to design a decision threshold
from 0 (most intense fading) to 20.
surface dependent on several external factors so the re-
Figures 8 and 9 show the protocol performance in the
sulting protocol is adaptive to many scenarios. Bridging
two mobility scenarios. Results are in the same pattern
this statistical broadcast protocol design gap is important
as the density results. The reachability of DCB and the
for VANET as it now opens up a new class of multi-hop
rebroadcast efficiency of AAG are poor. DADCQ and p-
broadcast protocols that are well-suited to the challenges
persistence perform well in both measures, although p-
of VANET.
persistence has fairly low reachability in the highway
scenario. Interestingly, the reachability of p-persistence
seems to be better when fading is more intense. ACKNOWLEDGMENTS
Here, we see that DADCQ is highly adaptive to the This work is an extension of papers presented at
intensity of fading, which is expected since we designed GLOBECOM 2010 (adaptivity to channel quality [2]) and
the DADCQ decision surface to account for it. When the WiMOB 2010 (adaptivity to node distribution pattern
fading is most disruptive (small values of K), DADCQ [27]).
selects more nodes to rebroadcast in order to compensate
for the reduced connectivity. When K = 0, DADCQ R EFERENCES
selects even more nodes to rebroadcast than AAG in
[1] S.-Y. Ni, Y.-C. Tseng, Y.-S. Chen, and J.-P. Sheu, “The broadcast
the highway scenario. As the fading intensity reduces storm problem in a mobile ad hoc network,” in MobiCom ’99:
(K gets larger), the decision threshold becomes more Proceedings of the 5th annual ACM/IEEE international conference on
aggressive and the bandwidth consumption is reduced. Mobile computing and networking. New York, NY, USA: ACM,
1999, pp. 151–162.
[2] M. J. Slavik and I. Mahgoub, “Statistical broadcast protocol design
9 C ONCLUSION for unreliable channels in wireless ad-hoc networks,” in Proceed-
ings of IEEE Globecom 2010 - Wireless Networking Symposium (GC10
We have proposed the Distribution-Adaptive Distance - WN), Miami, Florida, USA, December 2010.
with Channel Quality (DADCQ) protocol for multi-hop [3] W. Lou and J. Wu, “Toward broadcast reliability in mobile ad
hoc networks with double coverage,” IEEE Transactions on Mobile
broadcasting in VANET. VANETs exhibit wide variabil- Computing, vol. 6, no. 2, pp. 148–163, 2007.
ity in node density, distribution pattern, and channel [4] Y.-C. Tseng, S.-Y. Ni, and E.-Y. Shih, “Adaptive approaches to
quality. In response, we have shown how to design a relieving broadcast storms in a wireless multihop mobile ad hoc
network,” Computers, IEEE Transactions on, vol. 52, no. 5, pp. 545–
broadcast protocol utilizing the distance heuristic that 557, May 2003.
is adaptive to these factors. In the past, the distance [5] S. Biswas, R. Tatchikou, and F. Dion, “Vehicle-to-vehicle wireless
method has been made adaptive to a range of node communication protocols for enhancing highway traffic safety,”
IEEE Communications Magazine, vol. 44, no. 1, pp. 74–82, Jan. 2006.
densities, but not to be efficient across a range of [6] F. Ye, R. Yim, J. Guo, J. Zhang, and S. Roy, “Prioritized broadcast
these three factors simultaneously. DADCQ adopts the contention control in vanet,” in Communications (ICC), 2010 IEEE
quadrat method of spatial analysis to characterize the International Conference on, May 2010, pp. 1 –5.
[7] Y. Bi, L. Cai, X. Shen, and H. Zhao, “A cross layer broadcast
distribution pattern at each node. The resulting metric, protocol for multihop emergency message dissemination in inter-
the Q statistic, is combined with local node density N vehicle communication,” in Communications (ICC), 2010 IEEE In-
and the Rician fading parameter K to create a decision ternational Conference on, May 2010, pp. 1 –5.
[8] J. Cartigny, D. Simplot, and J. Carle, “Stochastic flooding
surface used by the distance method. broadcast protocols in mobile wireless networks,” LIFL
Simulation results with JiST/SWANS demonstrate Univ. Lille 1, Tech. Rep., May 2002. [Online]. Available:
DADCQ provides high reachability and efficient use http://citeseer.ist.psu.edu/525199.html
[9] X.-Y. Li, K. Moaveninejad, and O. Frieder, “Regional gossip
of bandwidth in both urban and highway scenarios routing for wireless ad hoc networks,” Mob. Netw. Appl., vol. 10,
with varying node density and fading intensity. DADCQ no. 1-2, pp. 61–77, 2005.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
IEEE TRANSACTIONS ON MOBILE COMPUTING
14
[10] S. al Humoud, L. Mackenzie, and J. Abdulai, “Neighbourhood- Conference Record of the Twenty-Eighth Asilomar Conference on, vol. 1,
aware counter-based broadcast scheme for wireless ad hoc net- oct. 1994, pp. 470 –474 vol.1.
works,” in GLOBECOM Workshops, 2008 IEEE, 30 2008-Dec. 4 [32] M. Boban, T. Vinhoza, M. Ferreira, J. Barros, and O. Tonguz,
2008, pp. 1–6. “Impact of vehicles as obstacles in vehicular ad hoc networks,”
[11] A. Mohammed, M. Ould-Khaoua, L. Mackenzie, and J.-D. Abdu- Selected Areas in Communications, IEEE Journal on, vol. 29, no. 1,
lai, “Dynamic probabilistic counter-based broadcasting in mobile pp. 15 –28, january 2011.
ad hoc networks,” in Adaptive Science Technology, 2009. ICAST [33] A. Domazetovic, L. Greenstein, N. Mandayam, and I. Seskar,
2009. 2nd International Conference on, 2009, pp. 120–127. “Propagation models for short-range wireless channels with pre-
[12] M. Slavik and I. Mahgoub, “Stochastic broadcast for vanet,” dictable path geometries,” Communications, IEEE Transactions on,
Proceedings of the 2010 Consumer Communications and Networking vol. 53, no. 7, pp. 1123 – 1126, jul. 2005.
Conference, January 2010. [34] T. Clausen and P. Jacquet, “Optimized link state
[13] O. Tonguz, N. Wisitpongphan, J. Parikh, F. Bai, P. Mudalige, and routing protocol (olsr),” October 2003. [Online]. Available:
V. Sadekar, “On the broadcast storm problem in ad hoc wireless http://www.ietf.org/rfc/rfc3626.txt
networks,” Broadband Communications, Networks and Systems, 2006. [35] M. Fiore, J. Harri, F. Filali, and C. Bonnet, “Vehicular mobility
BROADNETS 2006. 3rd International Conference on, pp. 1–11, Oct. simulation for vanets,” in Simulation Symposium, 2007. ANSS ’07.
2006. 40th Annual, mar. 2007, pp. 301 –309.
[14] O. Tonguz, N. Wisitpongphan, F. Bai, P. Mudalige, and V. Sadekar, [36] O. K. Tonguz, W. Viriyasitavat, and F. Bai, “Modeling
“Broadcasting in vanet,” in 2007 Mobile Networking for Vehicular urban traffic: a cellular automata approach,” Comm. Mag.,
Environments, May 2007, pp. 7 –12. vol. 47, pp. 142–150, May 2009. [Online]. Available:
[15] N. Wisitpongphan, O. Tonguz, J. Parikh, P. Mudalige, F. Bai, and http://dx.doi.org/10.1109/MCOM.2009.4939290
V. Sadekar, “Broadcast storm mitigation techniques in vehicular [37] Z. Jin, N. Yan, and L. Bing, “Reliable on-demand
ad hoc networks,” Wireless Communications, IEEE, vol. 14, no. 6, geographic routing protocol resolving network disconnection
pp. 84–94, December 2007. for vanet,” in Proceedings of the 5th International
[16] O. Tonguz, N. Wisitpongphan, and F. Bai, “Dv-cast: A distributed Conference on Wireless communications, networking and
vehicular broadcast protocol for vehicular ad hoc networks,” mobile computing, ser. WiCOM’09. Piscataway, NJ, USA:
Wireless Communications, IEEE, vol. 17, no. 2, pp. 47 –57, april IEEE Press, 2009, pp. 3963–3966. [Online]. Available:
2010. http://portal.acm.org/citation.cfm?id=1737966.1738429
[17] W. Viriyasitavat, F. Bai, and O. Tonguz, “Uv-cast: An urban [38] G. Wolny, “Modified dmac clustering algorithm for vanets,”
vehicular broadcast protocol,” in Vehicular Networking Conference in Proceedings of the 2008 Third International Conference on
(VNC), 2010 IEEE, dec. 2010, pp. 25 –32. Systems and Networks Communications. Washington, DC, USA:
[18] P. Kyasanur, R. Choudhury, and I. Gupta, “Smart gossip: An IEEE Computer Society, 2008, pp. 268–273. [Online]. Available:
adaptive gossip-based broadcasting service for sensor networks,” http://portal.acm.org/citation.cfm?id=1473252.1474527
IEEE International Conference on Mobile Adhoc and Sensor Systems [39] IEEE, “Dedicated short range communica-
Conference, vol. 0, pp. 91–100, 2006. tions working group.” [Online]. Available:
[19] B. Bako, F. Kargl, E. Schoch, and M. Weber, “Advanced adap- http://www.leearmstrong.com/DSRC/DSRCHomeset.htm
tive gossiping using 2-hop neighborhood information,” in Global
Telecommunications Conference, 2008. IEEE GLOBECOM 2008. IEEE,
nov. 2008, pp. 1 –6.
[20] T. Osafune, L. Lin, and M. Lenardi, “Multi-Hop Vehicular Broad-
cast (MHVB),” in ITS Telecommunications Proceedings, 2006 6th
Michael Slavik is a Computer Engineering
International Conference on, jun. 2006, pp. 757 –760.
Ph.D. Candidate in the Department of Computer
[21] M. Mariyasagayam, T. Osafune, and M. Lenardi, “Enhanced
and Electrical Engineering and Computer
multi-hop vehicular broadcast (mhvb) for active safety appli-
Science, at Florida Atlantic University. He
cations,” in Telecommunications, 2007. ITST ’07. 7th International
was previously a software engineer for
Conference on ITS, jun. 2007, pp. 1 –6.
Motorola, Inc in the iDEN and WiMAX handset
[22] G. Korkmaz, E. Ekici, and F. Ozguner, “Black-burst-based multi-
development group. His research interests
hop broadcast protocols for vehicular networks,” Vehicular Tech-
include vehicular networking, multi-hop wireless
nology, IEEE Transactions on, vol. 56, no. 5, pp. 3159 –3167, sep.
broadcast, wireless sensor networks, distributed
2007.
computing, and machine learning. His email
[23] A. Qayyum, L. Viennot, and A. Laouiti, “Multipoint relaying for
address is <mslavik@fau.edu>.
flooding broadcast messages in mobile wireless networks,” Hawaii
International Conference on System Sciences, vol. 9, 2002.
[24] C. Wu, K. Kumekawa, and T. Kato, “A novel multi-hop broadcast
protocol for vehicular safety applications,” Journal of Information
Processing, vol. 18, pp. 110–124, 2010.
[25] M. J. Slavik, I. Mahgoub, and M. Rathod, “Statistical broadcast
protocol design with wibdat: Wireless broadcast design and Imad Mahgoub is a full professor of Computer
analysis tool,” in Proceedings of IEEE Wireless Communications and Science and Engineering and director of
Networking Conference (WCNC 2010), Cancun, Mexico, March 2011. Mobile Computing Laboratory at Florida Atlantic
[26] M. J. Slavik and I. Mahgoub, “Designing statistical multi-hop University. His research interests include mobile
wireless broadcast protocols using confidence levels from stochas- computing, vehicular ad hoc networks, wireless
tic models of reachability,” in The International Wireless Communi- sensor networks, and parallel and distributed
cations and Mobile Computing Conference (IWCMC-2011), Istanbul, systems. He has over 100 publications including
Turkey, 7 2011. four books. He received research funds from
[27] ——, “Adapting statistical broadcast to linearly oriented networks federal government agencies and the industry.
for VANETs,” in Proceedings of IEEE International Conference on He is in the editorial board of the International
Wireless and Mobile Computing, Networking and Communications Journal of Communication Systems. He served
(WiMob’2010), Niagara Falls, Canada, Canada, October 2010. on the editorial board of the International Journal of Computers
[28] I. S. Institute, “The network simulator - ns-2,” July 2006. [Online]. and Applications and the Encyclopedia of Wireless and Mobile
Available: http://www.isi.edu/nsnam/ns/ Communications. He is a member of the steering committee of the
[29] R. Barr, “Swans - scalable wireless ad hoc network International Symposium on Performance Evaluation of Computer and
simulator user guide,” March 2004. [Online]. Available: Telecommunication Systems (SPECTS). He served as vice chair, track
http://jist.ece.cornell.edu/ chair, posters chair, publicity chair, and program committee member for
[30] R. Thomas, An Introduction to Quadrat Analysis. Geo Abstracts several international conferences and symposia. He is a senior member
Ltd., 1977. of the IEEE. He is also a member of the IEEE Computer Society, and
[31] I. Davis, J.S. and J. Linnartz, “Vehicle to vehicle rf propagation the ACM. His email address is <mahgoubi@fau.edu>.
measurements,” in Signals, Systems and Computers, 1994. 1994