Effects of Free Flow state Highway Trafc

Parameters on the Performance of Amplify and

Forward Cooperative Multi-hop Vehicular Networks
Prasad Perera, Dhammika Jayalath
School of Electrical Engineering and Computer Science
Faculty of Science and Engineering
Queensland University of Technology, Brisbane QLD, Australia
Email: ovitigalage.perera@qut.edu.au, dhammika.jayalath@qut.edu.au
AbstractIn this paper we analyse the effects of highway
trafc ow parameters like vehicle arrival rate and density on the
performance of Amplify and Forward (AF) cooperative vehicular
networks along a multi-lane highway under free ow state. We
derive analytical expressions for connectivity performance and
verify them with Monte-Carlo simulations. When AF cooperative
relaying is employed together with Maximum Ratio Combining
(MRC) at the receivers the average route error rate shows ten
fold reduction. A threefold increase in maximum number of
traversable hops can also be observed at different vehicle densities
when AF cooperative communication is used to strengthen
communication routes.
I. INTRODUCTION
Vehicle to vehicle connectivity is the basic building block
in Vehicular Ad-hoc Networks (VANETs). Vehicles equipped
with radio communication devices initiate and maintain com-
munication links to full data communication requirements
in such networks. End to end connectivity of these networks
depend on variety of parameters. Some of these parameters
are, radio conditions, vehicle speed, transmit power at each
node, data transmission rates and road trafc conditions etc.
Substantial amount of work on VANET connectivity can be
found in open literature. Early studies have assumed linear
one dimensional scenarios for connectivity. Some analytical
models assume VANETs with infrastructure support [1] while
majority assume fully ad-hoc networks [2], [3]. More often the
work on connectivity consider a direct communication model
between vehicles. Often, VANET environment is more dy-
namic and Vehicle to Vehicle (V2V) links become unreliable.
This directly affects connectivity of the network. Effects of
radio range and transmission power of vehicles on connectivity
is analysed in [4]. However the impact of cooperative relaying
on end to end vehicular connectivity not studied widely.
Employing cooperative relaying to increase the performance
in physical layer of VANETs were rst presented in [5], [6].
However signicant work could not be found which study the
effect of vehicular ow parameters (arrival rate, density) on the
connectivity of cooperative vehicular networks. We attempt
to address the above research gap in this paper. Work on
[7],[8] present comprehensive analytical models for vehicular
connectivity, yet pays relatively lesser attention on the V2V
communication model.
The rest of the paper is organised as follows. In Section II
the related work are discussed. The system model is presented
in Section III. Analytical expressions are derived in Section
IV. While simulations and results are presented in Section V,
Section VI concludes the paper.
II. RELATED WORK
Inter vehicle spacing distribution for vehicles in multiple
lanes are derived in [3]. They have studied the implications
of vehicle speed on the connectivity and derived performance
bounds based on an innite server queue model. This work
provides useful guidelines to approach connectivity problem
on uninterrupted highways under free ow condition. However
the authors have limited their discussion to a direct commu-
nication model which can be a signicant shortcoming on
sparsely populated highways.
Inspired by the work of [3], authors of [2] have investigated
the the same problem with a Physical layer perspective. They
adopt a communication model similar to [3]. The instantaneous
Signal to Noise Ratio (SNR) on a V2V link is expressed in
terms of the inter vehicle spacing and used as a fundamental
building block throughout the analysis. The authors have
derived analytical expressions for average route error rate, the
maximum number of possible hops (subjected to a target bit
error rate) and the minimum power requirement for connectiv-
ity in a highway under free ow state. The contribution of this
analysis is signicant in relating the vehicle ow parameters
to connectivity. However it can be extended with a cooperative
communication model, which has provided a key motivation
for our analysis.
Work of [5] is one of the pioneering analysis on cooperative
relay based communications in VANETs. This analysis is
limited to a single Source Relay Destination triplet
and not considered for multi-hop scenarios. Work in [9]
can be identied as one of the rst attempts to address the
benets of cooperative relaying on multi-hop connectivity in
a vehicular network. They have demonstrated the capacity
improvements achievable with a cooperative relaying model

i
=
2
i
/

2
i
erf
_
v
max
i

2
_
erf
_
v
min
i

2
_
_
v
max
i
v
min
i
1
v
exp
_
(v
i

i
)
2
2
2
i
_
dv (1)
for multi-hop connectivity but hardly relates the vehicular ow
parameters to the performance gain. This gap is the main
contribution in our work. Recently Khlass et.al [10] proposed
to combine cooperative relaying and analog network coding
to improve connectivity in vehicular networks. Their results
show that Bit Error Rates (BER) and throughput of multi-hop
vehicular communication paths can be signicantly improved
with the selected approach. However the road network used
for simulation is considerably diferrent from an actual setup.
Similar to [9] vehicular ow parameters are not considered as
critical parameters for end-to-end connectivity.
III. SYSTEM MODEL
The model used for our analysis is inspired by the work in
[2], [3]. An uninterrupted highway segment under free ow
state is considered for the analysis. A multi-lane highway is
assumed with M(> 1) number of lanes. Each lane i( M)
is dened by a Poisson vehicle arrival process with arrival
rate
i
. Hence the inter vehicle distance in each lane is an
exponentially distributed random variable with rate
i
. Here

i
is the vehicle density on lane i.
In free ow state, the vehicle density is very low and vehicle
speed and trafc ow are independent. The vehicle speed
is modeled as a truncated Gaussian random variable. V
i
is
the random variable representing vehicle speeds on lane i.
The average vehicle density along lane i in free ow state is
expressed by (1) [3] where,
i
= average speed,
i
= standard
deviation of speeds on lane i and v
max
i
=
i
+ 3
i
; v
min
i
=

i
3
i
. The average number of vehicles N
i
in lane i over a
nite highway length L can be given by N
i
=
i
L. In (1) the
error function erf() is dened as, erf(x) =
2

_
x
0
e
t
2
dt.
A xed transmission range R is assumed for all vehicles
in the network. At the same time all V2V links are assumed
to be Rayleigh fading. Therefore, the average SNR of a link
separated by a distance Y
i
will be
i
= P
T
/Y

i
P
noise
. Here
is a constant associated with path loss model. P
noise
is the
thermal noise power at the receiver and is the path loss
exponent. With these assumptions the average route bit error
rate is dened as E[P
link
(Y )] =
_
R
0
P
link
f
Y
(y)dy, where
P
link
is the link bit error rate at distance Y and f
Y
(y) is the
PDF of inter-vehicle spacing distribution. f
Y
(y) [2] is dened
as,
f
Y
(y) =
_
e
(y)
1e
(R)
0 y R
0 otherwise
(2)
A. Cooperative Communications Framework
In the light of cited work above we extend the analysis
with a cooperative relaying model. As shown in Figure 1, we
assume a multi-lane highway with lanes assigned for different

v1, 1
v2, 2
v3, 3
vi, i
Yi
w
R2
w
Relay Selection zone for
A-B vehicles
A(S) B(D)
R1
YSR
YRD
Figure 1. Improving connectivity in VANETs with Cooperative relay based
forwarding at each hop
speed ranges. Whenever a multi-hop communication route is
established along a highway lane, AF Cooperative relaying
is utilized at each possible intermediate hop to strengthen
the connectivity. This process depends on the availability of
vehicles to act as AF relays in the adjacent lane. In this work
we do not consider an optimum relay selection protocol which
is a separate branch of research.
IV. CONNECTIVITY ANALYSIS
Using an AF cooperative relaying model we derive relation-
ships between trafc ow parameters and following metrics.
(i) Probability of relay availability, (ii) Average bit error rate
of a route with AF cooperative relaying along a highway lane
segment of xed length L, (iii) Maximum number of hops
possible meeting certain power and threshold average bit error
rate requirements.
A. Probability of relay availability
Whenever a V2V link is formed, the initiator will search
for a potential relaying vehicle from the adjacent lane. Figure
1 shows the zone in the adjacent lane selected for the search.
This zone can be dened in alternative ways, however the
complexity of the search will vary accordingly. The relay
path with the maximum instantaneous SNR is selected for
AF relaying. For a selection, a relay should be available in
the zone at the time of searching. The availability of a relay
depends on the vehicle arrival rates on the adjacent and source
node lanes, {
i+1
,
i
}. The truncated CDF of V
i+1
explained
in [11] is used to derive the probability of relay availability
P
rel
on the adjacent lane when both lanes are on free ow
state. After substituting appropriate parameters CDF of V
i+1
,
F
V
T
(v) can be expressed as below,
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
vehicle arrival rate
i+1
(vehi/s)
P
r
o
b
a
b
i
l
i
t
y

o
f

r
e
l
a
y

a
v
a
i
l
a
b
i
l
i
t
y


P
r
e
l


Int. Vehicle Spacing =250 m
Int. Vehicle Spacing =150 m
Int. Vehicle Spacing =100 m
Int. Vehicle Spacing =50 m
Figure 2. Probability of relay availability for a given inter vehicle spacing
and different arrival rates for a lane with mean velocity = 70km/h and
= 21km/h
F
V
T
(v) =
_

_
0 v v
min
i+1
c
1
(F
V
(v) F
V
(v
min
i+1
)) v
min
i+1
< v < v
max
i+1
1 v v
max
i+1
(3)
where,
c = Q
_
v
max
i+1

i+1

2
i+1
_
Q
_
v
min
i+1

i+1

2
i+1
_
(4)
F
V
(v) =
1
2
_
1 + Q
_
v
i+1

2
i+1
__
(5)
here Q() is the complementary error function. The proba-
bility of availability of a relay P
rel
can be dened as,
P
rel
= Pr (No. of vehicles in spacing Y
i
on lane i+1 1)
The number of vehicles in lane i + 1 within a spacing
equal to Y
i
can be expressed as,
i+1
Y
i
=

i+1
v
i+1
Y
i
. Therefore,
it can be written as P
rel
= Pr
_

i+1
v
i+1
Y
i
1
_
. By changing
parameters we get,
Pr (v
i+1

i+1
Y
i
) = F
V
T
(
i+1
Y
i
) (6)
Therefore, P
rel
can be expressed as below,
P
rel
=
_

_
0
i+1
a
c
1
(F
V
(
i+1
Y
i
) F
V
(v
min
i+1
)) a <
i+1
< b
1
i+1
b
(7)
where, a =
v
min
i+1
Y
i
; b =
v
max
i+1
Y
i
Figure 2 illustrates the probability of availability of a relay
vehicle in the adjacent lane for a given inter vehicle spacing
Y
i
at different arrival rates of
i+1
. Innite number of curves
can be plotted for different spacing, however we show four of
them here. It can be seen that for a certain inter vehicle spacing
on lane i the probability of having at least one potential relay
node in the adjacent lane, i + 1 increases with the vehicle
arrival rate
i+1
. Once
i+1
is above a certain threshold value
it is certain to nd a potential relay, i.e. P
rel
is 1. Similarly
for a given setup there is critical value of
i+1
which P
rel
is 0. When large inter vehicle spacings are considered P
rel
reaches higher probability values at lower arrival rates due to
the widening of the relay searching region. This interpretation
of P
rel
will be different when the relay searching region is
dened in an alternative way. However this will complicate
the searching process further.
B. Average bit error rate of a route with AF cooperative
relaying along a highway lane segment of xed length L
Analytical expressions are derived for the average bit error
rate for a single hop between two vehicles travelling in the
same direction. We assume AF at the relaying node together
with symmetrical Rayleigh fading in all V2V links (i.e SR,
SD, RD). At destination receiving vehicles employ MRC
for direct and relayed signals. Consider vehicles A and B in
lane i separated by Y
i
(Figure 1). The average SNR in the link
between A and B can be expressed as
SD
= P
T
/Y

i
P
noise
.
Similarly, SNR for the SR and RD links will be,
SR
=
P
T
/Y

SR
P
noise
and
RD
= P
T
/Y

RD
P
noise
. The equiva-
lent SNR for the relay path
eq
can be expressed by (8) [12],

eq
=

SR

RD
1 +
SR
+
RD
(8)
The maximum equivalent SNR
max
(
eq
) is used to
represent the relay path here after. Using the Moment Gener-
ating Functions (MGF) to derive the average symbol error rate
assuming BPSK data transmission, The instantaneous symbol
error rate per link (P
link
) is calculated using (9) [13].
P
link
=
_
/2
0
_
sin
2

sin
2
+
SD
/2
__
sin
2

sin
2
+
max
/2
_
d
(9)
by substituting
SD
= P
T
/Y

i
P
noise
and solving the inte-
gration in (9) we get (10),
Were, K =
P
T
P
noise
. The average link bit error rate is found
using equation (11),
P
link
=
_
R
0
P
link
f
Y
(y)dy = 1 +I
2
I
3
(11)
where, I
2
and I
3
are complex integrals of y found using nu-
merical methods. Equations (12) and (13) shows the integrals
after expansion.
I
2
and I
3
can be found using numeric integration and hence
P
link
. Earlier we derived the P
rel
that for different vehicle
arrival rates. In a cooperative transmission frame work this
effects the P
link
.
Since we assume that a relay node will be utilized from the
adjacent lane depending on its availability the nal average
link bit error rate P
CT
link
,
P
link
=
1
2
_
1 +
_
_
KY

i
_
2
KY

i

max
_
1
KY

i
_
1 +KY

i
_
_

2
max
KY

i

max
_

max
(1 +
max
)
_
(10)
I
2
=
_
R
0
__

i
e
(
i
y)
1 e
(
i
R)
_
_
(Ky

)
2
Ky

max
_
1
Ky

(1 +Ky

)
_
dy (12)
I
3
=
_
R
0
__

i
e
(
i
y)
1 e
(
i
R)
__

2
max
Ky

max
_

max
(1 +
max
)
_
dy (13)
F =
2/

2
i+1
erf
_
v
max
i+1

i+1

i+1

2
_
erf
_
v
min
i+1

i+1

i+1

2
_
_
v
max
i+1
v
min
i+1
1
v
exp
_
(v
i+1

i+1
)
2
2
2
i+1
_
dv (14)
P
CT
link
=
_

_
P
DT
link

i+1

v
min
Y
i
P
rel
P
link
+ (1 P
rel
)P
DT
link
v
min
Y
i
<
i+1
<
v
max
Y
i
P
link

i+1

v
max
Y
i
(15)
where Y
i
= E[Y
i
]. P
DT
link
is the average bit error rate for a
link with BPSK transmission for direct communication. AF
cooperative relaying happens whenever relaying vehicles are
available, otherwise direct communication will happen in V2V
links. P
DT
link
=
P
noise
(+1,
i
R)
4P
T

i
(1e

i
R
)
as cited in Section 2. For
comparison purpose we use a similar approach described in
[2] to derive average bit error rate in (15) for n hops for a
BPSK transmission.
P
route
= 1
_
1 P
CT
link
_
n
(16)
C. Maximum number of hops possible meeting a threshold
average bit error rate
The maximum number of hops achievable hops constrained
by a threshold target bit error rate for the route P
target
can
be calculated using (17) [2]. To study the effects of trafc
ow parameters on this metric in a AF cooperative vehicular
network we use similar denitions used in [2], [14]. We use
previously derived values of P
CT
link
here.
P
route
P
target
(17)
n
max

ln(1 P
target
)
ln
_
1 P
CT
link
_
(18)
V. SIMULATIONS AND RESULTS
A. Average route error rate
The simulation results are compared with the analytical
results in this Section. P
route
is simulated for two differ-
ent ranges of
i+1
on lane i + 1. The density ranges are,
Fv
min
i+1
Y
i

i+1

Fv
max
i+1
Y
i
and
i+1

Fv
max
i+1
Y
i
where F
is given by (14).
The simulation uses two lanes only. Inter vehicle spacings
are generated from an exponential random process for lanes
i, i + 1 with rates
i
,
i+1
. P
rel
is calculated within the relay
selection zone 1 at each simulated hop. BPSK modulated
symbols are transmitted and the bit error is estimated at the
receiving end. This process was iterated 10000 times to nd
the average bit error rate for a V2V AF cooperative link. The
process is repeated for different values of
i
and
i+1
for all
density ranges. A highway segment of length L = 10000m
is considered and number of hops n =
i
L 1 is used to
nd P
route
. Figure 3 compares the analytical and simulation
results. We also compare the results with direct transmission
scenario in [2]. The results shows signicant improvements
in connectivity. Due to the fact that P
rel
changes with the
vehicle spacing distribution on lane i, the average route error
rate changes non-linearly with
i
for a given value of
i+1
.
There is different curve for each value of
i+1
. The AF
cooperative communication based system operates according
to the respective curve when the densities change dynamically.
We leave such a dynamic model for future investigations.
B. Maximum number of hops
Under this section we simulate the maximum number of
reachable hops n
max
constrained by a threshold route bit
error rate P
target
assuming AF cooperative relaying at in-
termediate hops. All vehicles transmit with an equal power
P
T
= 33dBm. When relay nodes are available the path with
maximum SNR is chosen. BPSK modulation is assumed in
this simulation as well. The route bit error rate is estimated
until the received power is equal or less than the P
target
. Each
iteration is counted as one hop and the maximum number
of hops are found. The process is repeated for different
vehicle densities. This simulation is performed using the
vehicle densities in the range
i+1

Fv
max
i+1
Y
i
on lane i + 1.
Figure 4 compares the results with the direct communication
0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
10
-4
10
-3
10
-2
vehicle density on lane
i

i
(veh/m)
A
v
e
r
a
g
e

r
o
u
t
e

e
r
r
o
r

r
a
t
e

[
P
r
o
u
t
e
]


Coop Tx
i+1
= 0.02(< Fv
maxi+1
/Y
i
)-Analytical
Direct Tx - Analytical
Coop Tx
i+1
= 0.08(> Fv
maxi+1
/Y
i
)-Analytical
Direct Tx - Sim
Coop Tx
i+1
= 0.08(> Fv
maxi+1
/Y
i
)-Sim
Coop Tx
i+1
= 0.02(< Fv
maxi+1
/Y
i
)-Sim
Figure 3. Average route error rate for Cooperative and Direct Transmission
(P
T
= 33dBm, = 2, r
b
= 6Mbps) for different vehicle densities (
i
) in
lane i
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
0
200
400
600
800
1000
1200
1400
vehicle density on lane i
i
(veh/m)
M
a
x
i
m
u
m

n
u
m
b
e
r

o
f

a
c
h
i
e
v
a
b
l
e

h
o
p
s

n
m
a
x


Direct Tx - Analytical
Coop Tx - Analytical
Direct Tx - Simulation
Coop Tx - Simulation
Figure 4. Maximum number of traversable hops vs vehicle density with
Cooperative and Direct Transmission (P
T
= 33dBm, = 2, r
b
= 6Mbps)
for P
target
=10

3 and
i+1
= 0.08

Fv
max
i+1
Y
i

scenario. AF cooperative communication and direct models,

n
max
shows an exponential relationship to vehicle density.
The network with AF cooperative communication shows a
threefold increase in the maximum number of achievable hops
when compared with a network with direct communication.
There is a predictability in this result since AF cooperative
relaying provides retransmission when possible. However the
result can be used to dimension cooperative protocols for
VANETs.
VI. CONCLUSION
We have derived analytical expressions relating highway
trafc ow parameters such as arrival rate and density with
connectivity performance in a multi-hop VANETs employing
an AF cooperative relaying model in contrast to a direct com-
munication model. The analytical results were veried with
Monte-Carlo simulations. Even though previous research shed
light on possible improvements with cooperative relaying on
VANETs we have tried to quantify the benets to some extent
using vehicle ow parameters as well. It was noted that when
the vehicle density on the adjacent lane increased, connectivity
performance matrices improved. A threefold increase in the
maximum number of achievable hops was observed with AF
cooperative relaying compared to a the network with direct
communication model while the improvement for average
route error rates were ten fold. We believe the analytical
relationships which we have presented for a AF cooperative
vehicular network will be useful for the researchers to ne
tune their cooperative protocols further for VANETs.
REFERENCES
[1] S. C. Ng, W. Zhang, Y. Yang, and G. Mao, Analysis of access
and connectivity probabilities in infrastructure-based vehicular relay
networks, in Wireless Communications and Networking Conference
(WCNC), 2010 IEEE. IEEE, 2010, pp. 16.
[2] P. Neelakantan and A. Babu, Connectivity analysis of vehicular
ad hoc networks from a physical layer perspective, Wireless
Personal Communications, pp. 126, 2012. [Online]. Available:
http://dx.doi.org/10.1007/s11277-012-0795-z
[3] S. Youse, E. Altman, R. El-Azouzi, and M. Fathy, Analytical model
for connectivity in vehicular ad hoc networks, Vehicular Technology,
IEEE Transactions on, vol. 57, no. 6, pp. 33413356, 2008.
[4] M. Torrent-Moreno, P. Santi, and H. Hartenstein, Distributed fair
transmit power adjustment for vehicular ad hoc networks, in Sensor
and Ad Hoc Communications and Networks, 2006. SECON06. 2006
3rd Annual IEEE Communications Society on, vol. 2. IEEE, 2006, pp.
479488.
[5] H. Ilhan, M. Uysal, and I. Altunbas, Cooperative Diversity for Intervehi.
Comm.: Performance Analysis and Optimization, Vehicular Technology,
IEEE Transactions on, vol. 58, no. 7, pp. 3301 3310, Sept. 2009.
[6] H. Ilhan, I. Altunbas, and M. Uysal, Cooperative diversity for relay-
assisted inter-vehicular communication, in Vehicular Technology Con-
ference, 2008. VTC Spring 2008. IEEE. IEEE, 2008, pp. 605609.
[7] S. Ukkusuri and L. Du, Geometric connectivity of vehicular ad hoc
networks: Analytical characterization, Transportation Research Part C:
Emerging Technologies, vol. 16, no. 5, pp. 615 634, 2008.
[8] W.-L. Jin and W. Recker, An analytical model of multihop connectivity
of inter-vehicle communication systems, Wireless Communications,
IEEE Transactions on, vol. 9, no. 1, pp. 106112, 2010.
[9] T. Koike, M. Tanaka, and S. Yoshida, Capacity improvement of
multihop inter-vehicle communication networks by stbc cooperative
relaying, IEICE transactions on communications, vol. 88, no. 9, pp.
35463553, 2005.
[10] A. Khlass, Y. Ghamri-Doudane, and H. Gacanin, Combining coopera-
tive relaying and analog network coding to improve network connectivity
and capacity in vehicular networks, in Global Telecommunications
Conference (GLOBECOM 2011), 2011 IEEE, 2011, pp. 15.
[11] W. R. McShane and R. P. Roess, Trafc engineering, 1990.
[12] S. Ikki and M. H. Ahmed, Performance analysis of cooperative diversity
wireless networks over nakagami-m fading channel, Communications
Letters, IEEE, vol. 11, no. 4, pp. 334336, 2007.
[13] M. K. Simon and M.-S. Alouini, Digital communication over fading
channels. Wiley. com, 2005, vol. 95.
[14] A. Akki and F. Haber, A st1986statisticalatistical model of mobile-
to-mobile land communication channel, Vehicular Technology, IEEE
Transactions on, vol. 35, no. 1, pp. 27, 1986.