Content uploaded by Subhash Sagar
Author content
All content in this area was uploaded by Subhash Sagar on Jun 08, 2020
Content may be subject to copyright.
arXiv:1212.4244v1 [cs.NI] 18 Dec 2012
1
On Probability of Link Availability in
Original and Modified AODV, FSR and OLSR
Using 802.11 and 802.11p
S. Sagar, N. Javaid, J. Saqib, Z. A. Khan$, U. Qasim‡, M. A. Khan
COMSATS Institute of Information Technology, Islamabad, Pakistan.
$Faculty of Engineering, Dalhousie University, Halifax, Canada.
‡University of Alberta, Alberta, Canada.
Abstract—Mobile Ad-hoc NETworks (MANETs) comprise on
wireless mobile nodes that are communicating with each other
without any infrastructure. Vehicular Ad-hoc NETwork (VANET)
is a special type of MANETs in which vehicles with high mobility
need to communicate with each other. In this paper, we present
a novel framework for link availability of paths for static as well
as dynamic networks. Moreover, we evaluate our frame work for
routing protocols performance with different number of nodes
in MANETs and in VANETs. We select three routing protocols
namely Ad-hoc On-demand Distance Vector (AODV), Fish-eye
State Routing (FSR) and Optimized Link State Routing (OLSR).
Furthermore, we have also modified default parameters of
selected protocols to check their efficiencies. Performance of these
protocols is analyzed using three performance metrics; Packet
Delivery Ratio (PDR), Normalized Routing Overhead (NRO) and
End-to-End Delay (E2ED) against varying scalabilities of nodes.
We perform these simulations with NS-2 using TwoRayGround
propagation model. The SUMO simulator is used to generate
a random mobility pattern for VANETs. From the extensive
simulations, we observe that AODV outperforms among all three
protocols.
Index Terms—AODV, FSR, OLSR, packet delivery ratio, end-
to-end delay, normalized routing load, MANETs, VANETs
I. INTRODUCTION
MANETs comprise of wireless mobile nodes that are com-
municating with each other without any centralized control.
MANETs are self-starting network and consist of collection
of mobile users that communicate over wireless links with
reasonably constrained bandwidth independently. In MANETs,
each node acts as a specialized router, thus, it is capable
of forwarding packets to other nodes. Topologies of these
networks are random and are changed frequently.
VANET is a special type of MANET in which nodes
(Vehicles) with high mobility can communicate with each
other. Due to expensive employment in real world their sim-
ulations are required comprehensively. There should be broad
level study so that the movement patterns of vehicles can be
modeled accurately. VANETs are distributed, self-organizing
communication networks built up by moving vehicles. These
nodes are highly mobile and have limited degrees of freedom
in the mobility patterns. In VANETs, routing protocols and
other techniques must be adapted to vehicular-specific capa-
bilities and requirements. A range of many useful applications
has been brought up by this new idea of VANETs. Some of the
application areas are traffic management, routing in VANETs,
handover, etc.
For calculating efficient routes in wireless networks special
routing protocols are used because traditional routing protocols
for wired network like link state and distance vector algorithms
cannot work efficiently. These protocols are divided into two
main categories with respect to their routing behavior; reactive
and proactive. Reactive routing protocols calculate routes for
destination in the network when demands for data is arrived,
therfore also known as On-demand routing protocols, whereas
in proactive routing routes are calculated periodically. As
proactive protocols are based on periodic exchange of control
messages and maintaining routing tables, that is why these
are known as table-driven routing protocols and each node
maintains complete information about the network topology
locally. It usually takes more time to find a route for reactive
protocol compared to a proactive protocol. For our analysis,
we select one reactive routing protocol, AODV [1] and two
proactive routing protocols FSR[2] and OLSR [3]. Moreover,
we also change default parameters of these protocols to
obtain efficient performance. We perform simulation in NS-2
through taking different scalabilities using RandomWay Point
propagation model.
Rest of the paper is arranged as follows: In section II
Related Work, in section III Motivation, and in section IV,
Performance Evaluation Metric are discussed. Section V shows
the Simulation results. Performance trade-offs are given in
section VI. Finally section VII concludes the paper.
II. RELATED WORK
A number of studies have been presented using different
mobility models or performance metrics in which performance
of different routing protocol is compared. One of the compre-
hensive studies is done by Monarch Project [4].
AODV, Destination-Sequenced Distance Vector (DSDV),
DSR and Temporally-Ordered Routing Algorithm (TORA) are
compared in this study using some of the performance metrics.
Clausen et al. [5] evaluate AODV, Dynamic Source Routing
(DSR) and OLSR in varying network conditions (node mobil-
ity, net-work density) and with varying traffic conditions with
2
Fig. 1: Link connectivity model for two nodes
Transport Control Protocol (TCP) and User Datagram Proto-
col (UDP). They show that different from previous studies,
OLSR performs like reactive protocols. After this scenarios-
based testing of protocols started, performance is altered after
changing the scenarios.
Fleetnet project [6] performed most detailed studies and
provided the platform for inter vehicular communication.
In the study [7], AODV, DSR, FSR and TORA on highway
scenarios are compared.
The study [8] have compared AODV, DSR, FSR and TORA
in city traffic scenarios. The authors of this study found,
for example, that AODV and FSR are the two best suited
protocols, and that TORA and DSR are completely unsuitable
for VANETs.
DYnamic MANET On-Demand (DYMO) is a reactive
routing protocol and the main candidate for the upcoming
reactive MANET routing protocols. It is based on the work and
experience from previous reactive routing protocols, especially
AODV and DSR [9].
III. MOTIVATION
Four protocols AODV, DSDV, DYMO and DSR are com-
pared in [10] for MANETs, in which throughput, E2ED, PDR,
Packet Drop Fraction, and NRO are taken versus number of
nodes, pause times and node speeds.
In [11] authors compared DYMO, AODV, AOMDV and
DSDV in VANETs in which performance is evaluated on
the basis of average E2ED, throughput, and overhead versus
number of nodes, speeds and number of packets.
AODV and OLSR are compared with respect to E2ED, PDR
and NRO against varying scalabilities of nodes [12]. But this
evaluation is performed only in VANETs. Authors in this paper
also modify some default parameter.
paper [13] analyze the performance of AODV, DSR and
DSDV protocols for TCP traffic pattern on the basis of Packet
Delivery Ratio, Throughput and Jitter.
In [14], authors present a frame work for Link Availability
probability in wireless network by using distance information
instead of complete information.
The studies from [4] to [13], as mentioned above, compare
the performance of routing protocols either in MANETs or in
VANETs. In this paper, we compare a reactive; AODV and two
proactive protocols; FSR and OLSR both in MANETs and in
VANETs with varying number of nodes and we also modify
the default parameters like [12]. We present our framework
with some different cases and find Link Availability time and
probability like [14].
IV. LINK AVAILABLE TIME
In [14], author find the available link time, link availability
probability and availability of a path between the originator
and the destination by using only the distance information
instead of complete neighboring information.
Suppose we have two nodes A and B. First, authors find
the available link time by using two nodes in which node B
is stationary and node Ais moving with velocity v, as shown
in Fig.1.
Amoves with distance zi, with time t=t1, t2and angle is
ϑ= Π −v1+v2+φ. Using cosines law:
d2
1=z2
1+d2
0+ 2z1d0cosϑ (1)
d2
2=z2
2+d2
0+ 2z2d0cosϑ (2)
Using z1=vzt1and z2=vzt2:
t1
t2
=v2
zt2
1+d2
0−d2
1
v2
zt2
2+d2
0−d2
2
(3)
Using eq.3,
vz=s(t2−t1)d2
0+t1d2
0−t2d2
1
t1t2(t2−t1)(4)
Case-1: Like [14], we consider two nodes Xand Yin which
both nodes are moving, as shown in Fig.2, where, Xand Y
move to X′and Y′at time t1and distance between them is
3
Fig. 2: Link connectivity model for two nodes
a0. Again X′and Y′move to X′′ and Y′′ at time t2and the
distance between them is a1. So,
a2
0=z2
1+d2
0−2z1d0cosϕ1(5)
a2
1=z2
2+d2
1−2z2d1cosϕ2(6)
Using z1=vzt1and z2=vzt2, we obtain,
t1a0cosϕ1
t2a1cosϕ2
=v2
zt2
1+d2
0−a2
0
v2
zt2
2+d2
1−a2
1
(7)
Fro eq.7, we get,
vz=s(z2
2+d2
1−a2
1)z1
(z2
1+d2
0−a2
0)z2×t1(d2
0−a2
2) + t2(d2
1−a2
1)
t1t2(t2−t1)(8)
Let Tbe the time in which one of the node is at the
boundary of other with a distance aand z=vzT, then similar
to Eq.5,
a2=z2+d2−2zdcosφ (9)
v2
zT2−z2
1+d2
0−a2
0
z1
vzT+d2−a2= 0 (10)
By simplifying eq. (10), we get;
T2−bT +c= 0 (11)
Whereas, by solving eq. (11), we get;
T=b−√b2−4c
2(12)
Where, b=z2
1+d2
0−a2
0
z1,c=d2−a2
v2
z
and z1=vzt1. We can
say from eq. (12) that time Tdepends on the distances and
time instead of any speed and other factors.
Case-2:Xmoves to X′for the first time and then to X′′
for second time but Yis stationary, as depicted in Fig.2. In
this type of case, authors in [14] puts the same angles for both
distances but we think these angles can not be same, contrary
to this, we put different angles for both distances.
d2
0=z2
1+d2
4−2z1d4cosθ1(13)
z2
3=d2
4+d2
2−2d4d2cosθ2(14)
Here, d2=vzt2. From eq.13 and eq.14, we get,
vz=s(e2+f2−d2
2)z1
(z2
1+e2−d2
0)f×t1d2
2+t2d2
0−e2(t1−t2)
t1t2(t2−t1)(15)
Case-3: We assume that both nodes are static, so, the link
availablity time is infinite because the distance between the
nodes remains constant.
Let L(t)is defined as the probability that two nodes are
connected directly at any time t. In case-1, we have assumed
that both nodes are moving and distance between them is Z. If
node Xmoves and reaches at the boundary of Ywith distance
Dand same as for case-2, let the distance between Xand Y
is d.
Z=pD2−(dcosθ)2−dsinθ (16)
From the above equation we can find the angle which is,
θ=cos−1z2+d2−D2
2Zd (17)
Hence, the probability that a link is available at any time t,
is,
L(t) =
1Z≤D−d
cos−1z2+d2−D2
2Zd D−d < Z ≤D+d
0Z > D +d
(18)
4
In the case when both nodes are static then there is a
maximum probability that a link is available which is 1.
V. SIMULATIONS AND DISCUSSIONS
In this section, we provide the details for the simulation
conducted for this study.
TABLE I: Simulation Parameters for MANETs and VANETs
PARAMETERS VALUES
NS-2 Version 2.34
OLSR Implementation UM-OLSR [15]
FSR Implementation FSR [16]
Number of nodes 10, 20, 30,., 70
Speed Uniform 40 kph
Data Type CBR
Simulation Time 900 seconds
Data Packet Size 1000 bytes
PHY Standard 802.11/802.11p
Radio Propagation Model TwoRayGround
SUMO Version 0.13
We have modified selected routing protocols and these
modifications are discussed below.
DEF-AODV uses TTL VALUES for
TTL I NC REM EN T ,T T L T H RES HOLD and
N ET DI AME T ER as mwntioned in [1]. In code provided
by AODV in [7] uses N E T DI AM ET ER of 30. Whereas,
in MOD-AODV, these TTL VALUES are modified with 4,
9 and 10 hops. As both FSR and OLSR use high exchange
intervals, therefore we shorten updates intervals in these
protocols; in MOD-FSR we change IntraS cope I nterval
from 5s to 1s and InterS cope Interval from 15s to 3s, and
in MOD-OLSR HELLO message and TC message intervals
are change from 2s and 5s to 1s and 3s.
Fig.3 and Fig.6 show the percentage of PDR, Fig.4 and Fig.7
show E2ED and Fig.5 and Fig.8 show NRO against varying
scalabilities.
A. PDR
The Medium Access Control protocol in IEEE 802.11p uses
the Enhanced Distributed Channel Access (EDCA) mechanism
originally provided by IEEE 802.11e. Therefore, successful
packet delivery rate of all protocols is better in VANETs as
compared to MANETs as shown in Fig.3. Moreover, reactive
protocols due to their on-demand nature are more suitable
for dynamic networks due to quick convergence feature.
Consequently, PDR of AODV is more as compared to FSR
and OLSR in VANETs. As shown in Fig.3.a,c, PDR of all
protocols is more in medium scalabilities and less in higher
scalabilities in MANETs, because congested networks suffer
more interferences which augment drop rates. Among reactive
protocols performance of AODV (DEF-AODV as well as
MOD-AODV)is high as compared to FSR (DEF-FSR and
MOD-FSR) and OLSR (DEF-OLSR and MOD-OLSR) both
in MANETs and in VANETs as depicted in Fig.3. Local link
repair during route maintenance provides more convergence
to AODV. Performance of AODV is better because it uses
gratuitous route reply (grat. RREP) and Expanded ring search
algorithm (ERS) which make AODV to perform better in all
scalabilities but with some cost. In DEF-OLSR, PDR for low
density is better but as the number of nodes are increased,
PDR decreases a little bit and is almost constant, beacause
it is proactive protocol and it cannot alter the route or link
on failure quickly. Thus MPRs continuously send packets to
destination which are lost due to absence of link but in MOD-
OLSR, PDR is high as compared to DEF-OLSR, this is due to
short interval of Topology control (TC) and Hello messages.
PDR of FSR is almost better in all scalabilities as compared
to OLSR but decreases when number of nodes increase due to
scopes techniques used in it. In MOD-FSR, due to shortening
the intervals of periodic exchange updates, PDR becomes
more as compared to DEF-FSR in VANETs (Fig.3) but by
shortening the interval of both FSR and OLSR, PDR becomes
high. Moreover, PDR of MOD-AODV increases a little bit as
compared to DEF-AODV in VANETs.
10 20 30 40 50 60 70
0
20
40
60
80
100
Number of Nodes
PDR (%)
DEF−AODV
DEF−OLSR
DEF−FSR
(a) AE2ED of Orig. Prot.s MANETs
10 20 30 40 50 60 70
0
20
40
60
80
100
Number of Nodes
PDR (%)
(b) AE2ED of Orig. Prot.s VANETs
10 20 30 40 50 60 70
0
20
40
60
80
100
Number of Nodes
PDR (%)
MOD−AODV
MOD−OLSR
MOD−FSR
(c) AE2ED of Mod. Prot.s MANETs
10 20 30 40 50 60 70
0
20
40
60
80
100
Number of Nodes
PDR (%)
(d) AE2ED of Mod. Prot.s VANETs
AODV OLSR FSR
0
10
20
30
40
50
60
70
80
90
100
PDR(%)
Default
Modified
(e) AE2ED of Prot.s MANETs
AODV OLSR FSR
0
10
20
30
40
50
60
70
80
90
100
PDR(%)
Default
Modified
(f) AE2ED of Prot.s VANETs
Fig. 3: End-to-end delay produced by reactive and proactive
protocols
B. E2ED
In general, E2ED of proactive protocols OLSR (DEF-
OLSR and MOD-OLSR) and FSR (DEF-FSR and MOD-
FSR) is lower as compared to AODV both in MANETs and
in VANETs because of pre-computation of routes decreases
the delay. Generally, AODV possesses the highest routing
delay both in VANETs and in MANETs. In AODV, local
link repair some time augments routing latencies, Moreover,
MOD-AODV due to incrementing initial search diameters
(TTL INCREM E N T = 4, T T L T H RES HOL D =
9and N E T W ORK DI AM ET ER = 10 hops reduce the
expansion rate and thus delay is decreased. Another reason
for OLSR to generate low delay is due to Multi-point Re-
lays (MPRs) provide efficient flooding mechanism instead of
broadcasting, control packets are exchanged with neighbours
5
only these MPRs are calculated by TC and Hello messges.
FSR and OLSR maintains a route for every node even before
starting the transmission of data packet. That is why increase
in intermediate nodes does not highly affect the E2ED as
shown in Fig. 4. DEF-FSR produces the more delay in high
scalabilities of 40, 50, 60 and 70 nodes as compared to MOD-
OLSR as shown in Fig. 4,c. Whereas DEF-OLSR in Fig.
4,d.produces same delay as that of FSR in Fig. 4,a,b. E2ED of
AODV in VANETs is increasing, whereas, for MANETs, in
high and low scalabilities routing latency is less as compared
to medium scalabilities.
10 20 30 40 50 60 70
0
0.05
0.1
0.15
0.2
0.25
Number of Nodes
AE2ED (s)
DEF−AODV
DEF−OLSR
DEF−FSR
(a) AE2ED of Orig. Prot.s MANETs
10 20 30 40 50 60 70
0
0.2
0.4
Number of Nodes
AE2ED (s)
(b) AE2ED of Orig. Prot.s VANETs
10 20 30 40 50 60 70
0
0.05
0.1
0.15
0.2
0.25
Number of Nodes
AE2ED (s)
MOD−AODV
MOD−OLSR
MOD−FSR
(c) AE2ED of Mod. Prot.s MANETs
10 20 30 40 50 60 70
0
0.2
0.4
Number of Nodes
AE2ED (s)
(d) AE2ED of Mod. Prot.s VANETs
AODV OLSR FSR
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
E2ED (s)
Default
Modified
(e) AE2ED of Prot.s MANETs
AODV OLSR FSR
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
E2ED (s)
Default
Modified
(f) AE2ED of Prot.s VANETs
Fig. 4: End-to-end delay produced by reactive and proactive
protocols
C. NRO
Among proactive protocols, OLSR attains the highest rout-
ing load as depicted in Fig.5 as compared to FSR. In OLSR,
any change in MPRs cause flooding to update routing table
entries, Moreover, checking the link status on routing layer
OLSRs sends HELLO probes along with periodic nature
of this protocol. Increase in TTL V ALU ES of MOD-
AODV augments routing load as compared to DEF-AODV in
MANETs as depicted in Fig.5. Whereas, Because of shorten-
ing the routing messages exchange intervals in both MOD-FSR
and MOD-OLSR (as shown in Fig.5) as compared to DEF-
FSR and DEF-OLSR (as can be seen from Fig.5) generates
more routing control messages for updating route table entries.
AODV possesses high NRO in higher scalability due to Hello
messages for link sensing, grat. RREP and LLR technique
increase control packets. NRO for DEF-OLSR and MOD-
OLSR is increasing because distribution of control packets
in the entire network is controlled by MPRs. Calculation of
these MPRs through TC and HELLO messages increases the
routing overhead as shown in Fig. 5,c,d.
10 20 30 40 50 60 70
0
5
10
15
20
25
30
Number of Nodes
NRO
DEF−AODV
DEF−OLSR
DEF−FSR
(a) NRO of Orig. Prot.s MANETs
10 20 30 40 50 60 70
0
10
20
30
Number of Nodes
NRO
(b) NRO of Orig. Prot.s VANETs
10 20 30 40 50 60 70
0
10
20
30
Number of Nodes
NRO
MOD−AODV
MOD−OLSR
MOD−FSR
(c) NRO of Mod. Prot.s MANETs
10 20 30 40 50 60 70
0
10
20
30
Number of Nodes
NRO
(d) NRO of Mod. Prot.s VANETs
AODV OLSR FSR
0
2
4
6
8
10
12
14
16
18
20
NRO
Default
Modified
(e) NRO of Prot.s MANETs
AODV OLSR FSR
0
2
4
6
8
10
12
14
16
18
20
NRO
Default
Modified
(f) NRO of Prot.s VANETs
Fig. 5: Routing overhead faced by protocols
VI. TRADE-OFFS MADE BY ROUTING PROTOCOLS TO
ACHIEVE PERFORMANCE
AODV: DEF-AODV achieves high PDR at the cost of high
delay because of local link repair which augments E2ED and
diminishes route re-discovery thus NRO is reduced as shown
in Fig. 4.e as compared to Fig. 5.e. In MOD-AODV, NRO is
reduced due to increase T T L IN C REM EN T AL value as
depicted in Fig. 5.e. Whereas, this increase lessens E2ED, as
can be seen from Fig.4.e.
FSR: Graded-frequency technique is more useful to reduce
routing latencies, but outer-scope causes more NRO (as de-
picted from Fig. 5.a.b and Fig. 5.c.d. Moreover, shortening
scope intervals in MOD-FSR in VANETs results high PDR in
as shown in Fig. 3.f at the cost of highest NRO values which
can be depicted from Fig. 5.f.
OLSR: Checking the topological connectivity of neighbors
on routing layer and triggered TC messages for topological
information in OLSR (DEF-OLSR and MOD-OLSR) increase
NRO as shown in Fig. 5.e,f, but due to transmission of
TC messages only through MPRs reduces routing latency
(Fig.4.e,f).
VII. CONCLUSION
In this paper we evaluate the performance of one reactive
protocol; AODV and two proactive protocols; FSR and OLSR
in both MANETs and VANETs using NS-2 simulator and
TwoRayGround radio propagation model and also we have
modeled link avalability time and the link avalability probabil-
ity. The SUMO simulator is used to generate a mobility pattern
for VANET to evaluate the performance of selected routing
protocols for three performance parameters, E2ED, NRO and
PDR. Our simulation results show that AODV performs better
at the cost of delay in MANETs and in VANETs.
6
During this study, we observe that routing link metrics is
an important component of a routing protocol. Because, a link
metric provides all available end-to-end paths and the best path
information to the respective protocol. So, in future, we are
interested to develop a new link metric, like [17] and [18].
REFERENCES
[1] C. Perkins, et al. IETF RFC3561, “Ad hoc On-Demand Distance
Vector (AODV) Routing”. Available: http://www.ietf.org/rfc/rfc3561.txt
Comput. Commun. Rev., 1994.
[2] M. Gerla, et al., IETF Draft-01, “Fisheye State Routing Protocol (FSR)
for Ad Hoc Networks”, 2000.
[3] T. Clausen,P. Jacquet, “Optimized Link State Protocol(OLSR)”, Project
Hipercom, INRIA,October 2003.
[4] Amir Qayyum, “Analysis and Evaluation of Channel Access Schemes
and Routing Protocols for Wireless LANs”, PhD thesis, University of
Paris-sud, Orasy, France 2000.
[5] Thomas Heide Clausen, PhilippeJacquet, LaurentViennot, “Comparative
Study of Routing Protocols for Mobile Ad-hoc Networks”, In 1st IFIP
MedHocNet Conference,2002.
[6] The FleetNet Project, http://www.et2.tu-harburg.de/fleetnet/english/vision.html
[7] Sven Jaap, Marc Bechler, and Lars Wolf, “Evaluation of Routing
Protocols for Vehicular Ad-hoc Networks in Typical Road Traffic
Scenerios”, in Proc of the 11th EUNICE Open European Summer School
on Networked Applications, Colmenarejo, Spain, July, 2005.
[8] Sven Jaap, Marc Bechler, and Lars Wolf, “Evaluation of Routing
Protocols for Vehicular Ad-hoc Networks in city Traffic Scenerios”, in
Proc of the 5th international Conference of Intelligent Transportaion
Systems Telecommunications (ITST), Brest, France, June 2005.
[9] Rolf Ehrenreich Thorup, “Implementing and Evaluating the DYMO
Routing Protocol”, Department of Computer Science, university of
AARHUS, Denmark, February, 2007.
[10] NeerajKr.Shakya, Gopal Singh Kushwah, Sandeep Singh Sen-
gar,“Efficiency Evaluation of Routing Protocol in MANET”,IJCST Vol.
2, Issue 2, June 2011.
[11] Mohammad Azouqa, Ali Jamoos, and SamerBali, “Investigation of
Vehicle-To-Vehicle Communication In A Road Intersection”, Depart-
ment of Electronic Engineering, Al-Quds university, Jerusalem, Pales-
tine.
[12] Imran Khan and Amir Qayyum, “Performance evaluation of AODV and
OLSR in highly fading vehicular Ad hoc Network Environment”, Center
of Research in Networks in Telecom.
[13] Ajay Kumar and Ashwani Kumar Singla, “Performance evaluation of
MANET routing protocols on the basis of TCP traffic pattern”, Inter-
national Journal of Information Technology Convergence and Services
(IJITCS) Vol.1, No.5, October 2011.
[14] Ming Yu et al., “Link Availability Modeling for Routing Algorithms
to Reduce the Link Break Time in MANETs”, 2011 International
Conference on Networking, Sensing and Control, Netherlands, 2011.
[15] ftp.heanet.ie/disk1/disk1/sourceforge/f/.../fs/fsr.../fsr-patch
[16] http://sourceforge.net/search/?q=UM-OLSR
[17] Nadeem Javaid, Ayesha Bibi, Karim Djouani, “Interference and band-
width adjusted ETX in wireless multi-hop networks”, USA, Globe-
com2010.
[18] Nadeem Javaid, Muti Ullah, Karim Djouani, “Identifying Design Re-
quirements for Wireless Routing Link Metrics”, USA, Globecom2011.
