You are on page 1of 6

Timer-based distributed dissemination protocols for

VANETs and their interaction with MAC layer


Pierpaolo Salvo

, Mario De Felice

, Andrea Baiocchi

, Francesca Cuomo

and Izhak Rubin

DIET at University of Roma Sapienza, Rome, Italy Email: (salvo, defelice, baiocchi, cuomo)@diet.uniroma1.it

Electrical Engineering Department, UCLA, Los Angeles, CA, USA Email: rubin@ee.ucla.edu
AbstractA key paradigm enabled by Vehicular Ad hoc NET-
works is the support of location aware push-mode info-tainment
services, besides the safety services that motivate the deployment
of the technology in the rst place. Message and advertisements
are spread via multi-hop communications from the originating
Road Side Units connected to the Internet. Support of such
services relies on robust and efcient dissemination protocols.
We dene a timer-based vehicular backbone network protocol,
where each vehicle can take forwarding decisions only based on
the information read in the message header, its current state
and local measurements. We analyze its performance taking into
account the effect of the IEEE 802.11p MAC layer. A comparison
with other literature dissemination protocols is carried out in a
highway setting.
Index TermsVehicular Ad Hoc Networks; data dissemina-
tion; IEEE 802.11p MAC; message forwarding
I. INTRODUCTION
Vehicular Ad-Hoc Networks (VANET) are a topic of grow-
ing interest both as a research eld and a promising technology
to enhance safety and improve the driving experience by
providing infotainment services (e.g., electronic toll collec-
tion, urban sensing, parking directions, advertisements). An
emerging standard in this eld is the DSRC (Dedicated Short-
Range Communications) [1] that allows the communications
among vehicles (V2V) through ad-hoc mode and between a
Road Side Unit (RSU) and a vehicle (V2I). In these networks,
information dissemination is crucial. The simplest solution,
message ooding, is known to incur the so called broadcast
storm problem in dense scenarios [2]. This problem can be
mitigated by designing a smart dissemination protocol.
Several proposals have appeared in the literature to provide
message spreading through a VANET. One of the most known
is the Distance Defer Transmission (DDT) [3], which assigns
message forwarding to the node farthest away from the sender
among those that have received the message. This is achieved
by setting a timer in each node A whose value is decreasing
with the distance between sender and A. The other key feature
of the protocol is that a node receiving a second copy of a
message gives up forwarding it and resets its timer. The Road
Oriented Dissemination (ROD) protocol [4] disseminates data
separately in each direction, and optimizes data dissemination
at the intersection. The nodes are assumed to get their GPS
positions, like in DDT. RSU service area extension has been
investigated in other papers, like in Urban Multi-hop Broadcast
(UMB) on [5] that also selects for re-transmission the most
distant node with respect to the transmitter and uses physical
repeaters at intersections to retransmit a message and to
overcome the problem of large buildings obstructing a message
path. The aim of UMB is to avoid MAC collisions caused
by hidden nodes, to make efcient use of the channel, and
getting reliable transmissions, thus spreading messages in all
directions at the intersection. In [6] a so called Triangle Rule
is introduced to improve dissemination in all the directions
at intersection, without further use of infrastructure, like in
UMB. The problem of nding the best relay node is also
faced in [7], with a signal burst process to select the best
relay node to disseminate the packet. An interesting work is
presented in [8], where the Authors provide a protocol that
relays packets using vehicle trajectories and travel statistics to
make the packet arrive at a precise point along the interested
vehicle trajectory, exactly when it passes by. It is an attractive
approach, even if it needs many parameters to be properly
tuned. In [9] a clustering algorithm is proposed, where vehicle
mobility patterns are used to set vehicle clusters with high
stability. This solution needs beacon exchange among vehicle,
which is quite a big overhead.
Our aim is to dene a fully distributed dissemination pro-
tocol that does not need beaconing and is suitable to support
high rate message ows without collapsing. This is obtained by
realizing, in a distributed way, a backbone network of vehicles
relaying messages. The relay election is carried out by means
of local timers at the vehicles. We show through our models
and simulations, that our algorithm meets those goals. We
account for the MAC delay and nite capacity. In particular,
we highlight that the timer quantization effect due to MAC
message delivery delays can bring to dissemination abruptly
stopping if the protocol is not properly designed.
The rest of the paper is structured as follows: Section II
describes the proposed dissemination protocol. In Section III
we provide a theoretical analysis to nd out the trade-off
between information spreading distance and end to end delay,
in Section IV we evaluate the timer dimensioning with respect
to the MAC delay. Simulation results of the full stack in a
highway scenario are reported in Section V. Finally in Section
VI we draw our conclusions.
II. TIMER-BASED DISSEMINATION PROTOCOL PRINCIPLE
The protocol architecture of nodes participating in the
VANET comprises PHY and MAC layers, based on the IEEE
802.11p standard, and a Forwarding Layer (FL) above, that is
in charge of the dissemination logic. We refer to infotainment
push mode services, where a continuous ow of advertise-
ments comes out of the RSU directed to all vehicles in a given
area around the RSU (typically several square km). Since the
target area is larger than the area directly covered by the RSU,
multi-hopping is used to spread the messages.
We refer to the Vehicular Backbone Network (VBN) dis-
semination protocol. With VBN the messages sent out by the
RSU are forwarded by the nodes that are closest to nominal
relaying positions, spaced out by D. The distance D is chosen
to as to provide a SINR adequate for the intended message
ow rate (see [10] for a full analysis of the centralized VBN).
In the following we assume nodes know the value of D, that
can be optimized and sent to all nodes on a much longer time
schedule than message ow delivery (e.g., hourly).
Here we present a new fully distributed version of VBN,
based on local timers at the FL. The distributed version of the
VBN, namely Timer-based Backbone Network dissemination
protocol (TBN) is stated with reference to a linear road
scenario, e.g., a highway. If the RSU location is taken as
the origin, nominal relaying positions are at
1
(x
k
, y
k
) =
(kD, 0), k Z.
The key idea is that each vehicle receiving a message starts
a timer, the smaller the closer the vehicle to a nominal relaying
position. The vehicle closest to the nominal relay position will
forward the message and by doing so will cause all other
vehicles in the same forwarding interval to cancel the message
before they forward it as well (inhibition rule).
Let us assume that each message carries a sequence number
s and sender node coordinates (x
S
, y
S
). Node A maintains a
state composed of three elements.
1) the index of the closest nominal relay position, k
A
=
arg min
kZ
d
A,k
=

(x
A
x
k
)
2
+ (y
A
y
k
)
2
,
where (x
A
, y
A
) are the coordinates of A.
2) a list T
A
of integers, representing the sequence numbers
of the messages that the node has processed (received,
possibly forwarded, payload passed on to upper layers)
3) a set T
A
= s
1
, . . . , s
m
of integers, representing
the sequence numbers of the m messages currently
handled by node A (pending messages); if no message is
currently under processing by node A then T
A
is void.
4) a set of timer values, one for each pending message:
T =
1
, . . . ,
m
.
Let a message with header values s, (x
S
, y
S
) be received
by a node A and let k
S
denote the index of the nominal relay
position closest to the sender node S. If s T
A
, the message
is discarded and no further action is taken. If s , T
A
but
s T
A
, then let s = s
j
. Station A checks whether k
A
,= k
S
.
In that case, the received message is a duplicate sent by a node
in a different relaying interval than A, so the original message
is kept with its timer running. If instead k
A
= k
S
, then some
station has forwarded that message before station A; hence
station A has to drop message s
j
and give up forwarding it
(inhibition rule). Finally, if s , T
A
and s , T
A
, this is a
1
Ordinate value of 0 is taken as the center of the road, namely the line
separating opposite direction lanes.
new message. Its payload is passed onto to the upper layer.
If [k
A
k
S
[ = 1, i.e., S and A belong to adjacent relaying
intervals, then m is increased by one and a new entry is
added to the sets T
A
and T
A
with values s
m+1
= s and

m+1
= T
max
2d
A,kA
/D, where T
max
is the maximum delay
the FL imposes to a message before forwarding. The received
message is schedule to be forwarded when the timer of value

m+1
expires.
III. ANALYSIS OF DISSEMINATION PROTOCOLS
We analyze the performance of message dissemination
with VBN, under an ideal MAC (no delay, no bandwidth
limitation). We denote the overall covered distance from the
RSU via multi-hop with Y , while W is the overall delay,
built up as the sum of the forwarding layer timer values.
Let us assume vehicles are scattered along a linear road span
according to a Poisson process with average spatial density .
Let us assume that the nominal relay distance D be such
that D R/2, where R is the radio coverage distance.
Then, wherever the forwarder is located inside an interval
of length D, it can cover the entire next interval of length
D. A message sent from the (k 1)-th interval to the k-th
interval will be successfully forwarded provided that there is
at least one station in the k-th interval. The probability of
successful forwarding is 1 e
a
, where a = D. Therefore,
the probability of M = m successful hops is Q
m
= T(M =
m) = (1 e
a
)
m
e
a
for m 0. It can be easily found that
E[M] =

m0
mQ
m
= e
a
1.
The covered distance is Y = D/2 + (M 1)D + S + R,
where S is the displacement of the last relay mode with respect
to the left extreme of the M-th interval. Since there is no bias
and the timer function has a symmetric law with respect to
the center of the interval, then E[S] = D/2. So, E[Y ] =
E[M]D + R = (e
a
1)D + R.
As for the timer, provided there is at least one station, the
probability that the timer is larger than t is the probability of
the event that no vehicle is found in the central part of the
interval within a distance d = (D/2)(t/T
max
) of the center
of the interval. We use the variable N[I] to denote the number
of vehicles in the interval I and let I() [D/2, D/2].
Then, we obtain:
T( > t) =

k=1
T(N[I(t/T)] = 0, N[I(1)] = k)
T(N[I(1)] > 0)
=
1
1 e
a

k=1
e
at/T
[a(1 t/T)]
k
k!
e
a(1t/T)
=
e
at/T
e
a
1 e
a
, t [0, T]
We can let = V T
max
, where V is a random variable in
[0, 1] with pdf
f
V
(v) =
ae
av
1 e
a
, v [0, 1] (1)
The mean value of V is E[V ] = [e
a
1 a]/[a(e
a
1)].
The overall delay is W =

M
k=1

k
= T
max

M
k=1
V
k
. Since
M and the V
k
s are independent, the mean delay is
E[W] = T
max
E[M]E[V ] = T
max

e
a
1
a
1

(2)
If we dene the normalized quantities

Y = (Y R)/R and

W = W/T
max
, then a = log(1 + E[

Y ]R/D) and
E[

W] =
E[

Y ]R/D
log(1 + E[

Y ]R/D)
1 (3)
This is the trade-off between average dissemination delay
and average dissemination coverage. As the latter increases,
also the former increases. With VBN the trade-off is quasi-
linear for a given ratio between the interval length D and the
coverage distance R. The analysis above holds for R/D 2.
In the following Sections we account for the effect of MAC
layer, rst as for the time quantization implied by non null
delivery times through the MAC interface, even for light load.
Then, by means of simulation we assess the effect of different
trafc loads as the message ow rate varies.
IV. EFFECT OF MAC DELAY ON TIMER DIMENSIONING
At light loads, MAC frame delivery times in the IEEE
802.11p are in the order of 1 ms, depending on frame payload
and air bit rate. In case two vehicles are located so that their
respective timers differ by an amount less than the MAC
delivery delay, say , the inhibition rule will not work. Note
that this issue affects any timer-based dissemination protocol.
In the following we analyze this effect on TBN, again with
Poisson vehicle spatial distribution.
Conditional on the number k of vehicle in a nominal
relay interval of length D, vehicle positions are i.i.d. uniform
random variables over (0, D). Since the location of each of
the k vehicles is uniform over (0, D), it will be in a segment
of length with probability q = /D. Let us divide the
interval into n equal size segments, namely = D/n, so
that q = 1/n. The timer is a map between the interval
and time, i.e., : i 1, . . . , n (0, T
max
). We
consider a generic nominal relay interval, say (D/2, D/2],
with nominal position 0. Then, the timer is quantized to n
values, the i-th of which is (i) and it is associated to the set
(i
D
2n
, (i 1)
D
2n
] ((i 1)
D
2n
, i
D
2n
], i = 1, . . . , n.
We are then in a similar situation as with a random back-
off mechanism where k competing nodes select at random
one out of n slots to attempt transmission. In our case, each
vehicle selects at random segment i and schedules the
relaying of the packet with a timer value (i). Thanks to the
Poisson distribution properties, the selection of each vehicle
is independent of all other vehicles. We assume (i) = i, so
that a node that sets the timer value (i) has time enough to
pass the message on to the MAC layer and have it delivered
to the other nodes in the road interval before any other vehicle
that has selected a timer value (j), j > i can release its copy
of the message to the MAC layer. In other terms, is no less
than the MAC resolution time.
Let us denote f(i) = T(node selects (i)) and G(i) =
T(node selects (j), j i), i = 1, . . . , n. Let q
k
be the
probability that there is no forwarding superposition, i.e., a
single node out of the k forwards the packets and all others
become aware of the forwarding action on time so as to cancel
their own scheduled forwarding. Then
q
k
=
n

i=1
kf(i)G(i + 1)
k1
(4)
In our case f(i) = 1/n and G(i) = (n i + 1)/n,
i = 1, . . . , n, so q
k
= (k/n)

n1
j=0
(j/n)
k1
. To recover the
unconditional probability of no forwarding superposition, q,
we weight q
k
with the probability that there are k vehicle in
the considered interval of length D, conditional on there being
at least one vehicle in the interval, i.e.,
a
k
e
a
/k!
1e
a
for k 1,
with a = D. Therefore
q =

k=1
e
a
1 e
a
a
k
k!
q
k
=
n

i=1
f(i)
ae
a
1 e
a

k=0
[aG(i + 1)]
k
k!
=
ae
a
1 e
a
n

i=1
f(i)e
aG(i+1)
(5)
In our special case:
q =
ae
a
1 e
a
n

i=1
1
n
e
a(ni)/n
=
a/n
e
a/n
1
(6)
The probability of superposition q = 1 q tends to 0 as
n increases since q 1/[1 + a/(2n) + o(1/n)] as n .
For a = 10, we obtain q 0.075 for n 65. Note that the
maximum value of the timer is T
max
= n. In the numerical
example above, it is T
max
97.5 ms for = 1.5 ms.
Concurrent forwarding of two or more vehicles in the same
nominal interval due to MAC induced timer quantization can
have a catastrophic effect on message dissemination if there
is an inhibition rule. In that case, vehicles in the (k + 1)-
th nominal interval receive two or more copies of the same
message from the nodes in the k-th interval and are forced to
cancel message forwarding, thus stopping dissemination. This
impairment can be worked around by means of TBN rules,
thanks to the variables k
A
and k
S
(see Section II).
The average value of the winning timer is given by
E[] =
ae
a
1 e
a
n

i=1
if(i)e
aG(i+1)
=
a/n
e
a
1
n1

j=0
(n j)

e
a/n

j
=
a/n
e
a
1
(e
a
1)e
a/n
n(e
a/n
1)
(e
a/n
1)
2
n

1
a

1
e
a
1

the last relationship holding asymptotically as n . Hence,


for large values of n, the expected delay induced by the
relaying timer of the FL is of the order of T
max
/a. This
analysis provides equations to dimension the value of T
max
,
once the MAC quantization time is given and the vehicle
spatial density is known.
We note that the concurrent forwarding problem is also
inherent in the other timer-based protocols and thus also in
the DDT protocol, described in Section I. Since this issue
is solved with TBN, we also introduce a hop count rule for
DDT, working like k
S
and k
A
in Section II, in particular
the hop count starts from zero and it is incremented at
every hop: if a node receives two packets with the same
hop count value, it assumes the packets are coming from
concurrent forwarders and thus the inhibition rule does not
apply. That implementation is here called DBF
hop
(Distance
Based Forwarding Hop Count).
V. PERFORMANCE ANALYSIS
We have done simulations in a highway environment by
using the ns-2 tool. An RSU at the center of the considered
road span broadcasts a message ow; messages are forwarded
on the two sides of the RSU with a dissemination protocol.
Tested protocols are: i) the proposal of this work, TBN; ii)
the DBF, derived from the DDT; iii) its improved version
DBF
hop
; iv) the VBN. All protocols use inhibition rule to
sop forwarding of duplicate messages.
With DBF, a node A receiving a message from a node S
sets its timer to T
max
(1 d
AS
/R
max
), where d
AS
is the
distance between A and S and R
max
is an upper bound of
the radio range. The improved version DBF
hop
adds a eld
in each message that is incremented by each relay node. It
serves the purpose to avoid a catastrophic dead end of message
propagation in case two or more close-by nodes send out
copies of the same message due to their close values of timers.
The VBN is introduced as a reference ideal case. Under
VBN the vehicular network along a road is partitioned in
intervals of length D. Each interval is served by a single relay
node, elected to be part of the vehicular backbone network
and chosen to be the node closest to the center of the relevant
interval. All nodes along the road collaborate with one another
to elect the backbone nodes off line (i.e., using a control
channel). The elected node maintains its role in spite of vehicle
movement, provided it keeps within a given maximum distance
from the nominal relay position at the center of the interval.
In the simulations we have set this distance equal to 20% of
the overall interval length D. Once this threshold is overcome,
a new election takes place.
MAC and PHY parameters are set equal to those specied
by the IEEE 802.11p standard. Results have been derived by
repeating each simulation 20 times, with simulations lasting
16 s. Table 1 summarizes the values of the main parameters.
We assume a road with two lanes going opposite directions.
Vehicles are placed according to a truncated Poisson distri-
bution with parameter with a minimum distance between
vehicles set equal to g = 25 m. During the simulation, each
vehicle moves along its own lane with a speed value uniformly
distributed between 0 and v
max
= 108 km/h. The vehicle
speed value is changed for each vehicle with a time sampling

t
= 3 s; vehicle move according to the car following model:
in each lane, the follower vehicle advances by the minimum
between the distance d(v
f
) it would cover at its target speed
v
f
during sampling time and the free space between the
follower and the leading vehicle, namely g + d(v
l
), where v
l
denotes the target speed of the leading vehicle.
The following performance metrics are evaluated.
Packet Delivery ratio, PDR: the average fraction of mes-
sages received by a vehicle; it is computed as
PDR =
1
J

J
j=0
r
j
E[N(t)]
where J is the number of transmitted packets, N(t) the
number of nodes in the scenario at time t and r
j
the number
of nodes that receive message j.
Forwarding Nodes, FN: the average fraction of vehicles
acting as message forwarders, computed as
FN =
1
J

J
j=0
v
j
E[N(t)]
where v
j
is the number of nodes that forward message j.
MAC Collisions: the average fraction of discarded MAC
frames due to collisions out of the overall number of received
MAC frames.
End to end delay, E2E: this metric provides the average
time needed to deliver a message from the RSU to the farthest
node able to receive it. We compute E2E as:
E2E =
1
J
J

j=1
W
j
(i

), i

= arg max
1iN
Y
j
(i)
where W
j
(i) is the delay of the j-th message from the RSU
up to the i-th reached node and Y
j
(i) is the distance of the
i-th node from the RSU when it receives the j-th message.
Fig. 1 plots the average fraction of nodes receiving RSU
messages for TBN with T
max
= 200 ms, in comparison with
a simpler TBN version that does not use the variables k
A
and k
S
(see Section II). We can observe how the check on
the origin interval in the TBN impacts the fraction of reached
nodes with differences in the order of 20% more than the TBN
version without such functionality.
All the remaining gures plot the performance metrics
dened above.
The curves in Fig. 2 show that VBN offers PDR = 100%;
also TBN delivers a high percentage of messages (around 80%
in the worst case) especially in the case of low vehicle density,
while the PDR in case of DBF and DBF
hop
is 5% 35%
lower. In particular, we can see how the hop count positively
impacts on delivered messages for DBF
hop
with respect to
the standard DBF, thus conrming the analysis in Section
IV, with almost always a 10% more delivered messages. The
two DBF schemes also introduce larger delays, whereas TBN
keeps close to the ideal VBN. The percentage of forwarding
nodes FN is about the same for all protocols, except of TBN,
which for high veicular density uses a relatively large number
of relay nodes; this is due to the requirement that there be
a relay per interval of length D. The MAC collision fraction
(a) Scenario and vehicular trafc parameters
Parameters Values
Road length (km) 10
Number of lanes 2
(vehic/km) 20, 30, 40, 50, 60
D (m) 500
RSU data rate (kbit/s) 80, 160, 320, 640
Link Rate (Mbit/s) 6
Vehicle max speed (vmax km/h) 108
Max forwarding delay (Tmax ms) 100
(b) MAC and PHY parameters
Parameters Values
Propagation Model Two ray ground
MAC, PHY parameters IEEE 802.11p
CS Threshold (dBm) 100
Transmission Power (mW) 500
Frequency (GHz) 5.9
Noise Floor (dBm) 104
Broadcast packet payload (bytes) 1000
Max range (Rmax m) 1400
TABLE I
SIMULATION PARAMETER VALUES.
20 25 30 35 40 45 50 55 60
0
10
20
30
40
50
60
70
80
90
100
(vehic/km)
R
e
a
c
h
e
d

N
o
d
e
s

%


TBN with Ka and Ks
TBN
Fig. 1. Benets of introduction of the hop count rule in TBN in term of
percentage of reached nodes
shows that the high value of relay nodes in TBN produces
more collisions than the other two schemes, even if this does
not prevent messages to be disseminated. On the contrary, the
PDR is always bigger than for the other protocols, also at
high vehicle density ( = 60 veh/km).
In Fig. 3, we set the vehicular density to an intermediate
value of = 30 veh/km and we plot the performance metrics
as a function of the bit rate of the message ow sent through
the RSU. We observe that DBF cannot handle a high rate,
as pointed out by the big increase of the E2E delay curve in
the region above 200 kbit/s together with a strong fall in the
PDR. TBN can support all the considered data rates with a
performance that is very close to the ideal VBN.
In Fig. 4, 5 and 6, we report the behavior of all protocols
with different data rates when vehicle density increases. We
notice in Fig. 5 that DBF performance strongly depend on the
ow rate and that in general, we get an average 70% 80%
of PDR only for lower rates (when also delay is low). It
is possible to observe the same behavior in Fig. 4, with the
only important difference that DBF
hop
sistematically gets 20%
more PDR with good delay and collisions percentages. It
is also important to notice that this hop count version can
reach 100% of PDR for lower rates, while DBF is never able
to. Finally, TBN in Fig. 6 further improves these results by
achieving PDR = 100% in almost all situations. In particular,
only high densities at the highest data rate cause a fall in the
PDR (together with collisions, forwarding nodes and delay),
but still TBN obtains better results with respect to the other
protocols in similar conditions.
20 30 40 50 60
0
20
40
60
80
100
P
D
R

(
%
)
(vehic/km)
20 30 40 50 60
0
1
2
3
4
5
E
2
E

d
e
l
a
y

(
s
)
(vehic/km)
20 30 40 50 60
0
5
10
15
20
F
o
r
w
a
r
d
e
r

N
o
d
e
s

(
%
)
(vehic/km)
20 30 40 50 60
0
20
40
60
80
100
C
o
l
l
i
s
i
o
n

(
%
)
(vehic/km)


DBF VBN
DBF
hop
TBN
Fig. 2. Performance behavior of different dissemination protocols as a
function of trafc density , for D = 500 m and R = 640 kbit/s
200 400 600
0
20
40
60
80
100
P
D
R

(
%
)
Data Rate (kbps)
200 400 600
0
1
2
3
4
5
E
2
E

d
e
l
a
y

(
s
)
Data Rate (kbps)
200 400 600
0
5
10
15
20
F
o
r
w
a
r
d
e
r

N
o
d
e
s

(
%
)
Data Rate (kbps)
200 400 600
0
20
40
60
80
100
C
o
l
l
i
s
i
o
n

(
%
)
Data Rate (kbps)


DBF VBN
DBF
hop
TBN
Fig. 3. Performance behavior of different dissemination protocols as a
function of RSU data rate, for D = 500 m and for = 30 veh/km
20 30 40 50 60
0
20
40
60
80
100
P
D
R

(
%
)
(vehic/km)
20 30 40 50 60
0
0.5
1
1.5
2
2.5
E
2
E

d
e
l
a
y

(
s
)
(vehic/km)
20 30 40 50 60
0
5
10
15
20
F
o
r
w
a
r
d
e
r

N
o
d
e
s

(
%
)
(vehic/km)
20 30 40 50 60
0
20
40
60
80
100
C
o
l
l
i
s
i
o
n

(
%
)
(vehic/km)


80kbps 160kbps 320kbps 640kbps
Fig. 4. Performance behavior of DBF
hop
as a function of trafc density ;
for D = 500 m and different RSU Rates
20 30 40 50 60
0
20
40
60
80
100
P
D
R

(
%
)
(vehic/km)
20 30 40 50 60
0
0.5
1
1.5
2
2.5
E
2
E

d
e
l
a
y

(
s
)
(vehic/km)
20 30 40 50 60
0
5
10
15
20
F
o
r
w
a
r
d
e
r

N
o
d
e
s

(
%
)
(vehic/km)
20 30 40 50 60
0
20
40
60
80
100
C
o
l
l
i
s
i
o
n

(
%
)
(vehic/km)


80kbps 160kbps 320kbps 640kbps
Fig. 5. Performance behavior of DBF as a function of trafc density ; for
D = 500 m and different RSU Rates
20 30 40 50 60
0
20
40
60
80
100
P
D
R

(
%
)
(vehic/km)
20 30 40 50 60
0
0.5
1
1.5
2
2.5
E
2
E

d
e
l
a
y

(
s
)
(vehic/km)
20 30 40 50 60
0
5
10
15
20
F
o
r
w
a
r
d
e
r

N
o
d
e
s

(
%
)
(vehic/km)
20 30 40 50 60
0
20
40
60
80
100
C
o
l
l
i
s
i
o
n

(
%
)
(vehic/km)


80kbps 160kbps 320kbps 640kbps
Fig. 6. Performance behavior of TBN as a function of trafc density ; for
D = 500 m and different RSU Rates
VI. CONCLUSIONS
To efciently overcome the broadcast storm problem, that
can arise during the data dissemination in a VANET, timer-
based solutions have been proposed. These solutions rely on
timers that are used by nodes as a distributed way to elect
a relay node for each message. In this paper we consider a
timer-base vehicular backbone network protocol, where timer
values are chosen so as to distribute relay nodes evenly along
the road. We have analyzed TBN performance by taking
into account the effect of a nite capacity underlying MAC
protocol, thus highlighting a general issue with timer-based
dissemination protocols, namely spurious forwarding due to
nodes selecting timer values that differ by less than the MAC
service time (MAC time quantization). We have showed that
such quantization brings erroneous breaks in the forwarding
scheme and proposed a solution inside our new protocol,
but also implemented it in an old protocol (DBF
hop
). We
then showed, via simulations, the good performance of our
proposed TBN (and of its centralized version VBN) with
respect to DBF and to its improved version DBF
hop
.
ACKNOWLEDGMENTS
Authors are grateful to Mirco Magrini for the work done in
the simulations set up.
REFERENCES
[1] N. da Fonseca, Notes on DSRC & WAVE Standards Suite: Its Architec-
ture, Design, and Characteristics, Communications Surveys & Tutorials,
IEEE, vol. 12, no. 4, pp. 439440, 2010.
[2] S.-Y. Ni, Y.-C. Tseng, Y.-S. Chen, and J.-P. Sheu, The broadcast
storm problem in a mobile ad hoc network, in Proceedings of the 5th
annual ACM/IEEE international conference on Mobile computing and
networking, ser. MobiCom 99, 1999, pp. 151162.
[3] M.-T. Sun, W.-C. Feng, T.-H. Lai, K. Yamada, H. Okada, and K. Fu-
jimura, GPS-based message broadcast for adaptive inter-vehicle com-
munications, in Vehicular Technology Conference, 2000. IEEE VTS-Fall
VTC 2000. 52nd, vol. 6, 2000, pp. 2685 2692 vol.6.
[4] M. O. Cherif, S.-M. Senouci, and B. Ducourthial, Efcient data dissem-
ination in cooperative vehicular networks, Wireless Communications
and Mobile Computing, pp. n/an/a, 2011.
[5] G. Korkmaz, E. Ekici, F.

Ozg uner, and

U.

Ozg uner, Urban multi-
hop broadcast protocol for inter-vehicle communication systems, in
Proceedings of the 1st ACM international workshop on Vehicular ad
hoc networks. ACM, 2004, pp. 7685.
[6] P. Salvo, F. Cuomo, A. Baiocchi, and A. Bragagnini, Road side
unit coverage extension for data dissemination in vanets, in Wireless
On-demand Network Systems and Services (WONS), 2012 9th Annual
Conference on. IEEE, 2012, pp. 4750.
[7] B. Blaszczyszyn, A. Laouiti, P. Muhlethaler, and Y. Toor, Opportunistic
broadcast in vanets (ob-van) using active signaling for relays selection,
in ITS Telecommunications, 2008. ITST 2008. 8th International Confer-
ence on. IEEE, 2008, pp. 384389.
[8] J. Jeong, S. Guo, Y. Gu, T. He, and D. Du, Trajectory-based statistical
forwarding for multihop infrastructure-to-vehicle data delivery, Mobile
Computing, IEEE Transactions on, vol. 11, no. 10, pp. 15231537, 2012.
[9] C. Shea, B. Hassanabadi, and S. Valaee, Mobility-based clustering
in vanets using afnity propagation, in Global Telecommunications
Conference, 2009. GLOBECOM 2009. IEEE. IEEE, 2009, pp. 16.
[10] I. Rubin, A. Baiocchi, F. Cuomo, and P. Salvo, Message relaying in
linear road VANET, in Internal Technical report, NelLab at DIET,
University of Roma Sapienza, October, 2012.

You might also like