Optimal Path Selection in Ad Hoc (MANET) by Using Genetic Fuzzy Petri Net

Abstract

A mobile Ad-hoc network (MANET) is a dynamic multi hop infrastructure less wireless network established by a group of mobility nodes in which there is no central administration. Due to the absence of central infrastructure and mobility of nodes that led to dynamic network topology, therefore, the routing is one of the most important challenges in ad-hoc networks. This challenge led to the development of many techniques which deal with it, one of the most important techniques it's a fuzzy logic. This paper focuses on finding the optimal path which using to transfer data, so two techniques ware used in this work. The first one is: The network simulator version 2 (NS-2) was used to build network system for study and evaluate the behavior of two routing protocols (AODV and AOMDV), in order to prove that path selected from this routing protocol is not optimal path, the performance metrics that used to evaluate the routing protocols are: the throughput, the dropped packets, average end-to-end delay, normalizes routing load, packet delivery fraction, and jitter. Therefore, we proposed model depended on second techniques Fuzzy Petri Net model, this model was proposed in order to mimic the route discovery mechanism in routing protocols with using a new parameters like (Number of hops, Local Battery level, Received Signal Strength Indicator) as the input to fuzzy inference system. Genetic algorithm has active role in the generating of fuzzy rules, where it used in this work to get the best number of generated rules, depended on training stage to generate fuzzy rules in offline, this fuzzy rules is very important in the reasoning processes to making a final decision about what is path most select. [Hussein Attya Lafta, Fadhil Mohammad Salman. Optimal Path Selection in Ad Hoc (MANET) by Using Genetic Fuzzy Petri Net. Researcher 2014;6(8):31-44]. (ISSN: 1553-9865). http://www.sciencepub.net/researcher. 5

Full text

Researcher 2014;6(8) http://www.sciencepub.net/researcher
31
Optimal Path Selection in Ad Hoc (MANET) by Using Genetic Fuzzy Petri Net
Hussein Attya Lafta 1, Fadhil Mohammad Salman 2
1. Assistant Professor Doctor, En, Iraq, University of Babylon, Iraq 2014.
2. Researcher, University of Babylon-College of Sciences, Iraq 2014.
hzazmk@yahoo.com, fadhilmohammad2000@yahoo.com
Abstract: A mobile Ad-hoc network (MANET) is a dynamic multi hop infrastructure less wireless network
established by a group of mobility nodes in which there is no central administration. Due to the absence of central
infrastructure and mobility of nodes that led to dynamic network topology, therefore, the routing is one of the most
important challenges in ad-hoc networks. This challenge led to the development of many techniques which deal with
it, one of the most important techniques it’s a fuzzy logic. This paper focuses on finding the optimal path which
using to transfer data, so two techniques ware used in this work. The first one is: The network simulator version 2
(NS-2) was used to build network system for study and evaluate the behavior of two routing protocols (AODV and
AOMDV), in order to prove that path selected from this routing protocol is not optimal path, the performance
metrics that used to evaluate the routing protocols are: the throughput, the dropped packets, average end-to-end
delay, normalizes routing load, packet delivery fraction, and jitter. Therefore, we proposed model depended on
second techniques Fuzzy Petri Net model, this model was proposed in order to mimic the route discover y
mechanism in routing protocols with using a new parameters like (Number of hops, Local Battery level, Received
Signal Strength Indicator) as the input to fuzzy inference system. Genetic algorithm has active role in the generating
of fuzzy rules, where it used in this work to get the best number of generated rules, depended on training stage to
generate fuzzy rules in offline, this fuzzy rules is very important in the reasoning processes to making a final
decision about what is path most select.
[Hussein Attya Lafta, Fadhil Mohammad Salman. Optimal Path Selection in Ad Hoc (MANET) by Using
Genetic Fuzzy Petri Net. Researcher 2014;6(8):31-44]. (ISSN: 1553-9865). http://www.sciencepub.net/researcher.
5
Key words: MANET, AODV, AOMDV, NS-2, GA, FL, FIS, PN, FPNs
1. Introduction
Wireless networks can be defined as a network in
which the nodes are interconnected by wireless
channels, due to increasing development of wireless
network, there are two architectures existing for these
networks, the first one is known as infrastructure
networks and the second one is known as
infrastructure less networks. An ad hoc network is a
collection of wireless mobile hosts forming a
temporary network without needing of any established
infrastructure or centralized administration. In such an
environment, it may be necessary for one mobile host
(node) to enlist the aid of other hosts (nodes) in
forwarding a packet from a source node to its
destination, Due to the limited range of each mobile
host’s wireless transmissions. So, MANET's nodes are
act as a both host and router by receiving and
forwarding data packets [12]. If these nodes change
their positions dynamically, it is called a mobile ad
hoc network (MANET). The MANET is
self-organizing and self-configuring multi hop
wireless networks where the structure of the network
changes dynamically because of mobility of nodes, the
nodes in a MANET are free to move and organize
themselves in the arbiter fashion. Wireless mobile ad
hoc network (MANET) technology is designed for the
establishment of a network anywhere and anytime,
without any fixed infrastructure to support the
mobility of the users in the network. A MANET can
be a standalone network or it can be connected to
external networks (Internet). The main two
characteristics of MANET are nodes mobility and
multi hop, hence multi hop operation requires a
routing mechanism designed for mobile nodes. In a
MANET networks where there is no infrastructure
support and since a destination node might be out of
range for source node that transmitting packets,
therefore, efficient route is needs to exchange
information between different nodes, it's done by
using different routing protocols. So, efficient routing
procedures is always needed to find a path between
nodes, in order to obtain appropriate transfer of the
packets between the source node and the destination
node [2]. Therefore, a central challenge in the design
of ad hoc networks is the development of dynamic
routing protocols that can efficiently find and
maintains the routes between two communicating
nodes (source & destination). This paper focuses on
the routing process in order to selected the optimal
path between any two nodes, that path is reduce some
challenges that faced the MANET such as (end-to-end
delay, loss packets, energy consuming, etc) and

Researcher 2014;6(8) http://www.sciencepub.net/researcher
32
increasing the performance of network such as
(throughput, packets delivery, quality of service, etc).
The nodes often have a finite capacity path cache, it
may not be possible to store all paths, therefore,
always needed to find the optimal path
communication between sender nodes to destination
node for mobile ad hoc networks. Organization of the
paper: Section I provides introduction to wireless
ad-hc network. Section II provides overview about
some
related work. Section III provides introduction to
NS-2. In section IV provides an overview of different
theory's that used this work. Section V provides a
discussion on the simulation which used NS-2 and the
proposed FPN model finally section VI provides the
conclusion on the proposed scheme, VII Future
Works.
II. Related Works
 Tzu-Chiang Chiang and et al. at 2009
proposed a knowledge-based inference approach to
the new path discovery for multicasting. A fuzzy Petri
net agent, which is a special expert system was
introduced at each node to learn and to adjust itself to
fit the dynamic conditions in a multicast ad-hoc
network. the simulation results show that the proposed
approach is up to 67.17% more efficient in the packet
delivery ratio as compared with a bandwidth effective
multicast routing protocol [17].
 A novel reliable routing algorithm in
mobile ad-hoc networks using fuzzy Petri net with its
reasoning mechanism was proposed by Zhi-gang Hu,
et al. at 2005 to increase the reliability during the
routing selection process. which has a fuzzy reasoning
mechanism for finding the routing sprouting tree from
the source to the destination node in the mobile ad hoc
environment, by compared the degree of reliability in
the routing sprouting tree, so, the most reliable route
can be computed. The results show that the routing
reliability was increased by more than 80% then
applying the proposed algorithm to the ad hoc on
demand distance vector routing protocol [21].
 The performance of hybrid routing protocol
for the MANET was analyzed by Subramanyam, P.V,
et al. at 2006, when they used Fuzzy Petri nets model
which calculate the level of firing transitions among
the nodes and comparing it with a threshold value for
the nodes of the network. The results indicated
enhanced performance with dynamically viable node
movement [14].
 Tzu-Chiang Chiang and et al. at 2004
described the multicast routing representation using
fuzzy Petri net model with the concept of immediately
reachable set in wireless ad hoc (MANET) networks
which all nodes equipped with GPS unit. A heuristic
algorithm is used to compute the multicast tree based
on the local network topology with a multicast source.
The results shows that the percentage of the
improvement is more than 15% when compared the
IRS method with the original method [16].
III. Introduction to NS-2
NS-2 simulator is an open source simulator; it is
very useful and important for founding and
investigating variety of protocols. It use a very large
numbers of applications, protocols, network types,
network elements, mobility models and traffic models
to investigation realistic simulation. At the simulation
layer NS-2 uses OTcl (Object oriented Tool
Command Language) programming language to
interpret user simulation scripts. OTcl language is in
the fact an object oriented extension of the Tcl
Language. The Tcl language is fully compatible with
the C++ programming language. At the top layer,
NS-2 is an interpreter of Tcl scripts of the users, they
work together with C++ codes [6]. An OTcl script
written by a user is interpreted by NS. While OTcl
script is being interpreted, NS creates two main
analysis reports simultaneously as a files. One of them
was called the NAM (Network Animator) object that
shows the visual animation of the simulation. The
other was called the trace object that consists of the
behavior of all objects and all the event occur in the
simulation, both of them are created as a file by NS-2.
Former is .nam file used by NAM software that comes
along with NS. Latter is a “.tr” file that includes all
simulation traces in the text format. NS-2 project is
normally distributed along with various packages (ns,
nam, tcl, otcl , etc.) named as “all-in-one package”,
but they can also be found and downloaded separately
[11]. That means After simulation NS-2 output are
Trace file and Nam trace file as a simulation results.
To interpret these results graphically and interactively,
tools such as Network Animator (NAM) are used.
NAM is an animation tool for viewing network
simulation traces and packet traces, and XGraph is
used as a data plotter to analyze a particular behavior
of the network.
IV. Theory of The Work
A- Genetic Algorithms (GAs)
Genetic Algorithms (GAs) are general-purpose
search algorithms that represent evolutionary
optimization approach, proposed by John Holland in
1970. They are particularly applicable to problems
which are large, non-linear & possibly discrete in
nature. GA try to work on principle of natural
selection, as in natural selection over the time
individuals with “good” genes survive whereas “bad”
ones are rejected [1]. As the natural evaluation has the
following feature [13]:-
 The individual characteristics are encoded
as a gene on the chromosome.
 Compute the fitness function value to each
chromosome according to the environment in which it

Researcher 2014;6(8) http://www.sciencepub.net/researcher
33
exists.
 Individual chromosomes has strong
behaviors are able to survive and produce next
 generations of strong individual
chromosomes.
 genetic operators can be generate a new
variants during reproduction The individuals by its
operation called (mutation).
In Genetic Algorithm the solution of the problem
is encoded on a string of bits that represented as a
gene comparable with the chromosome of the
biological system analogy. The Genetic Algorithm
keeps a population of randomly selected chromosomes
to combine by mutation or crossover techniques and
produce the offspring having new characteristics,
which in turn replaces the low fitness old
chromosomes. This process is repeated until we find a
chromosome with best fitness and repeated
characteristics for the successive generations of the
population. Which finally represent the optimal
solution for the problem [10].
B- Fuzzy Petri Nets
The classic Petri nets was used to modeling
system that accurately describe is unsuitable in the
systems of uncertain (imprecise) and ambiguous
information, While fuzzy logic is dealing with
uncertain and ambiguous data. According to the
uncertain information is to be display with fuzzy logic,
therefore to be useful integration theory of fuzzy in
petri nets to increase the power of modeling. In 1988,
was performed this work by Looney and several
authors of Petri Net, where they are combine the
collection of artificial intelligence due to its adequacy
to represent the reasoning process as a dynamic
discrete event system with petri net and design types
of petri nets that are more or less compatible with the
theory of Petri Net but more powerful. The techniques
that are combine the Petri Nets and fuzzy sets, called
Fuzzy Petri Nets (FPNs), that use for knowledge
representation and as a control in the fuzzy
management systems [19]. The formalism of Petri nets
can be used to model fuzzy-rule based systems by
simply identifying some elements (places and
transitions) and features (marking function) of
Petri-net’s formalism with the basic elements of a
fuzzy-rule based such as knowledge base (KB),
propositions (rules), degree of truth of the rules and
implication relationships, where transitions serve as
rules, places serve as propositions, and markings are
assigned fuzzy values between 0 and 1. [8]. FPNs is
an application specific Petri Nets based approach
developed to represent uncertain operations and
approximate conditions in areas such as robotics,
traffic control, communication, medical diagnosis,
flexible manufacturing systems and fuzzy controllers.
FPN model was introduced for the specification of
rule-based reasoning using propositional logic. Places
are interpreted as conditions having fuzzy truth values
(tokens), while transitions represent the fuzzy decision
values of rules, The relationships from places to
transitions and vice versa are represented by directed
arcs. Reasoning in the fuzzy petri net can be
performed by iteratively maxing (generalized OR) and
mining (generalized AND) transitions and fuzzy truth
values of tokens, respectively [4]. Because normal PN
cannot deal with vague or fuzzy information such as
“very high” and “good”, several Fuzzy Petri Nets
(FPN) have been introduced. As a model of
knowledge-based systems, fuzzy Petri net model for
expert systems is called Adaptive Fuzzy Petri Nets
(AFPN), This model has both the benefits of a fuzzy
Petri net and the learning ability of a neural network to
have the ability of learning like neural networks,
where each transition serves as neurons, therefore it
can be learned as well as it can be train to adapt the
change situation. Accordingly an AFPN model can be
used for dynamic knowledge representation and
inference [18]. The firing rule of transition's in the
fuzzy petri net , don't remove the token from the input
place after transition firing, therefore, FPNs
unbounded places because token will be unlimited in
places [7]. The FPNs was expanded from a petri net is
a bipartite graph that has place and transition nodes
like the petri nets. But, in FPNs a token incorporated
with a place is associated with a real value between 0
and 1, so, the transition is associated with a certain
factor (CF) real value between 0 and 1. FPNs is a
promising modeling tool for expert systems and it is
suitable for fuzzy knowledge representation and
reasoning. A generalized fuzzy petri nets is defined as
a 10-tuple, table (1) [20]:
Table (1): Format Definition of Fuzzy Petri Nets.
FPN = (P, T, I, O, D, W, μ, f, α, β)
Symbols Description
P {p1, p2, . . , pn} denotes a finite nonempty set of places.
T {t1, t2, . . , tm}denotes a finite nonempty set of
transitions.
I P → T was the input function, a mapping from places to
transitions, input incidence matrix.
O T → P was the output function, a mapping from
transitions to bags of places.
D {d1, d2, . . , dn}denotes a finite set of propositions, that
interprets fuzzy linguistic, P  T  P, |P| = |D|.
W WI U WO, is a finite set of input and output weights of
nets.
μ T → [0, 1] denotes the certainty factors of fuzzy rules.
F T → [0, 1] denotes the threshold of a transition firing.
α P → [0, 1] is an association function which maps from
places to real values between zero and one .
β P → D is also an association function mapping from
places to propositions.
That means, each place in FPNs may or may not
contain token associated with a truth value between 0

Researcher 2014;6(8) http://www.sciencepub.net/researcher
34
and 1, and the token in the places represent the degree
of belong to membership of this real value. The
transition has a certainty factor associated between 0
and 1 and it represents the inference engine.
V. A- Simulation Using NS-2
In this paper, the NS-2 with two routing
protocols (Ad-hoc On Demand Distance Vector
(AODV) and Ad-hoc On Demand Multipath Distance
Vector (AOMDV)) are used in the simulation, in the
different scenarios, the main reasons to select the
AODV routing protocol is that it was represented as a
single path routing protocol, and selected the
AOMDV routing protocol because it was represented
as a multipath routing protocol. Figure (1) explains the
main stages of simulation used in this work, that
implemented by using NS-2, where the code was
written in a Tcl language and the outputs of the NS-2
are trace file and NAM file. The trace that is used in
this work is old trace file because it contains all the
fields that are required in the computation of the
performance metrics. This performance metrics was
computed according to several steps that programmed
using AWK programming language:
Figure (1): Flowchart for Stages of Simulation.
Simulation Algorithm
Step1:- Start.
Step2:- Set S=0 (S represent the number of
scenario
file (movement file)).
Step3:- Build the traffic generation file "CBR
file" that
is generated by "cbrgen.tcl" file that supported
by NS-2. This script is found in (ns-allinone-
2.34/ns-2.34/ind_util/cmu _ scen_gen/ ).
Step4:- Set I=0 (I represents the number of
routing
protocols that are used in this work).
Step5:-Build MANET's scenario (movement file)
using support of NS-2 by the "setdest" script.
Step6:- Build "tcl" script that represents
simulation
environment of MANET with mobility model
for one routing protocols.
Step7:- Select suitable parameters that input to
this
"tcl" file in the NS-2 in order to perform the
simulation, and the outputs are "NAM" file to
display and trace file contain all the simulation
event to analysis.
Step8:- Analyze the trace file and compute the
performance metrics for the network
(throughput, drop packet, end to end delay,
jitter, packets delivery and normalize routing
load).
Step9:- Increment I by 1.
Step10:- If (I< 2) then go to step6 ( to
implemented
another routing Protocol) and save the new
results in metrics file, Otherwise, S=S+1.
Step11:- If (S<10) then go to step5 (S is the
number of
MANET scenarios). Otherwise, go to the
step12.
Step12:- Split the resulted file in to two files
(each one
contains the results of one routing protocol).
Step13:- Compute the average of the
performance
criterion for each routing protocols file and
put it in final file.
Step14:- possibility draw the results with the
suitable
parameter by using Xgraph this script is
supported by NS-2 or draw it as a
histograms.
Step15:- End.
Simulation Environment
A discrete event Network Simulator NS-2
version 2.34 was used in this work to simulate the
mobile ad-hoc network, depending on the "Tcl"

Researcher 2014;6(8) http://www.sciencepub.net/researcher
35
programming language that is used to build the
simulation environment of the mobile ad-hoc network.
There are several performance metrics that can be
used to analysis the performance of mobile ad-hoc
network or the protocols in order to understand it
performance as following:-
1. The Throughput
It is the amount of digital data transmitted per
unit time from the source node to the destination node,
it is usually measured in bits per sec [9].
2. Packet Delivery Fraction (PDF)
It is the ratio between the number of packets
originated by the “application layer” CBR sources and
the number of packets received by the CBR sink at the
final destination [9].
3. Dropped Packets
It is the number of packets that sent by the source
node and unsuccessful to reach to the destination node
[3].
4. Normalize Routing Load (NRL)
Is the total number of control packets (include
RREQ, RREP, RERR and REP_ACK packets)
divided by number of transmitted data packet in the
network [15].
5. Average End-To-End Delay
Is the average time taken by data packets when
released by sources until reach to their destinations
[15].
6. Average Jitter
It is the absolute value of the difference between
the end-to-end delays of two sequential packets, The
average jitter is obtained by summing the jitter of all
received packets divided by the total number of the
received packets [5].
Table (2): Network Parameters Used During The
Simulation.
Parameter
Value
The simulator
NS
-
2 version 2.34
MAC
802.11
Propagation model
Two ray ground
Routing protocols
AODV & AOMDV
Simulation time
75s
Traffic generator
CBR
Antenna
Omni Antenna
Packets size
512 bytes/packet
Transition rate
2.0 packets/second
Mobility model
Random wayp
oint model
Pause time type
Uniform
Speed type
Uniform
Table (3) Shows the suggested parameters used to
build simulation scenario that input to "Tcl" script, as
follows:-
Table (3): Parameters Used During Create Scenario.
Parameter
Value
Node number
3
0
Pause time
8.00s
Max node speed
20.00 m/s
Area
1000m*1000m
The number of results obtained after applying
this collection of parameter according to suggested
steps explained previously, figure (2) describes the
results of simulation with AODV routing protocol for
10 scenarios as histograms:-
a . Histogram of throughput.
b . Histogram of drop packet
c . Histogram of end to end delay.

Researcher 2014;6(8) http://www.sciencepub.net/researcher
36
d . Histogram of packets delivery fraction.
e . Histogram of normalize routing load.
f . Histogram of jitter.
Figure (2): Histograms of AODV Results.
The following figure (3) describes the results of
simulation with AOMDV routing protocol for 10
scenarios as histograms:-
a . Histogram of throughput.
b . Histogram of drop packet.
c . Histogram of end to end delay.
d . Histogram of packets delivery fraction.

Researcher 2014;6(8) http://www.sciencepub.net/researcher
37
e . Histogram of normalize routing load.
f . Histogram of jitter.
Figure (3): Histograms of AOMDV Results.
The following figures describes the simulation
results for AODV & AOMDV routing protocol with
(10,20,30) nodes, depended on Xgraph script which
supported by NS-2 simulator:-
Figure(4): Comparison Throughput for AODV &
AOMDV with versus number of nodes (10 , 20 , 30).
Figure (5): Comparison Dropped Packets for AODV
&AOMDV with versus number of nodes (10, 20, 30).
Figure (6): Comparison Average End-To-End Delay
for AODV&AOMDV with versus number of nodes
(10, 20, 30).
Figure (7): Comparison Packet Delivery Fraction
(PDF) for AODV&AOMDV with versus number of
nodes (10, 20, 30).

Researcher 2014;6(8) http://www.sciencepub.net/researcher
38
Figure (8): Comparison Normalize Routing Load
(NRL) for AODV&AOMDV with versus number of
nodes (10, 20, 30).
Figure (9): Comparison Average Jitter for AODV &
AOMDV with versus number of nodes (10, 20, 30).
The obvious differences in previous results with
(AODV & AOMDV) routing protocol for 10
scenarios, that led to conclusion, that path used from
this routing protocols to transfer data is not optimal
path because it was selected with regardless to some
of important issues that influence on the behaviors of
network such as (limited battery power for nodes,
congestion in the network, number of hops, signal
power, …, etc). In order to select the best nodes to be
part of the routes, a Genetic Fuzzy Petri Net is
proposed to mimic the route discovery process, where,
depending on this route metrics, the decision of path
selection is done.
V. B- The Proposed GFPN System.
In this work the GFPN model is developed to
simulate the route discover y process in ad-hoc routing
protocols depending on the inference process from
antecedent to the consequent propositions, which use
reasoning process of the two techniques (Fuzzy set
and Petri nets).
Figure (10) explains the main stages of proposed
FPN system:-
Figure (10): Flowchart for Modeling of The Proposed
System.
V. B-1. The Data Base
The data base used in this work is organized in
the form of data-tree having a depth of three levels.
The root of tree be in the first level (represents starting
pointer), predicates be in the second level and fuzzy
beliefs that must correspond to each predicate (second
level) are in the third level. This organization helps in
efficient searching of the data base. The fuzzy beliefs
must be collect from sources with good authentication
level. Each belief must have a certainty factor. For
more explain show the figure (11).
Figure (11): Tree of The Data Base.

Researcher 2014;6(8) http://www.sciencepub.net/researcher
39
The following algorithm is used for create
data-tree, figure (12) shows this algorithm.
Figure (12): Algorithm for Creating Data Tree.
V. B-2. Production Rules
The rules for decision making are created by
using the genetic algorithm. GA generate the rules by
using the Michigan approach, where each individual
in population represents as a rule with its inputs and
outputs. So, all the rules are evaluated off-line before
being applied in the proposed system. The evaluation
algorithm assigns strength to every classifier with a
non-zero degree of activation. Figure (13) shows the
training steps of genetic algorithm as following:-
Figure (13): Flowchart for Production Rules
Algorithm.
Production Rules Algorithm
Step1:- Start.
Step2:- Create random initial population, where
eachindividual in this population represents rulewith its
inputs and outputs and initial strength0.5. Every input
and output represents gene,hence, the length of
chromosome is 4 gene(according to the fuzzy input
parameters:Number of hops, Local Battery
level,Received Signal Strength Indicator) with
theoutput.
Step3:- The individuals (rules) in the population
areevaluation by getting membership degree ofever y
gene in the chromosome and then hasstrength of rule to
represent fitness of eachrule. Where the active rule
according to itsfitness is used in crossover to get a
newindividual, this new individual is replace byworst
individual.
Step4:- The binary tournament selection is used
forselecting the best individual from thepopulation and
uniform crossover is used toproduce new individual
from the parent (activerules). So, the soft mutation
modifies one of thecharacteristic points (a, b, c) of the
trianglefuzzy set.
Step5:- In this stage every new individual is
evaluatedby checking if it is active to get the strength
ofrule.
Step6:- The strength of rules is update with every
newgeneration, depending on the system'sperformance
and rule adaption rate, in order touse it to compute the
fitness of each individualto every new generation.

Researcher 2014;6(8) http://www.sciencepub.net/researcher
40
Step7:- The suitable rules according to the
specificmeasures have got go to the step8,
Otherwise,go to the step4.
Step8:- End.
The following table (5) shows the set of rules
that result from training phase of GA with use of the
following abbreviation, as in table (4).
Table (4): Abbreviations That Used in The Production
Rules Process.
Symbol
Description
N
Number of hops
B
Local battery level
S
Received signal strength indicator
O
Optimal
path
L
Low
M
Medium
H
High
Table (5): List of Generated Rules.
NO
Rule
1
"if n is h and
b is l and
s is h then
o is m
"
2
"if n is l and
b is m and
s is m then
o is m
"
3
"if n is h and
b is m and
s is h then
o is m
"
4
"if n is m and
b is l and
s is l then
o is l
"
5
"if n is m and
b is m and
s is l then
o is l
"
6
"if n is m and
b is l and
s is h then
o is m
"
7
"if n is h and
b is h and
s is l then
o is l
"
8
"if n is m and
b is h and
s is m then
o is m
"
9
"if n is l and
b is l and
s is l
then
o is l
"
10
"if n is h and
b is h and
s is l then
o is l
"
11
"if n is h and
b is m and
s is l then
o is l
"
12
"if n is l and
b is m and
s is l then
o is m
"
13
"if n is l and
b is l and
s is m then
o is l
"
14
"if n is l and
b is l and
s is h
then
o is m
"
15
"if n is m and
b is h and
s is l then o is l
"
16
"if n is m and
b is m and
s is m then
o is m
"
17
"if n is l and
b is m and
s is h then o is h
"
18
"if n is l and
b is h and
s is l then
o is m
"
19
"if n is h and
b is h and
s is h t
hen
o is m
"
20
"if n is m and
b is l and
s is m then
o is l
"
21
"if n is m and
b is h and
s is h then
o is h
"
22
"if n is h and
b is h and
s is m then
o is m
"
23
"if n is h and
b is m and
s is m then
o is l
"
24
"if n is m and
b is m and
s is h th
en
o is m
"
25
"if n is l and
b is h and
s is m then
o is h
"
26
"if n is h and
b is l and
s is m then
o is l
"
27
"if n is l and
b is h and
s is h then
o is h
"
V. B-3. Fuzzy Petri Net Model Building
The FPNs model is created depending on
arranged fuzzy rules and data base, after the last was
represented as a tree. The algorithm which used to
build the FPN model is explained in figure (14) as
follows:-
Figure (14): Algorithm for Building a FPN Model.
The following interface helps the user to create
the FPN model with aid the given database and
production rules. This interface consists of several
interactive sub interfaces like (Create FPN, Draw
fuzzy Petri net, Input fuzzy initial values, and
Decision making with explanation). Figure (15) shows
these interface.

Researcher 2014;6(8) http://www.sciencepub.net/researcher
41
Figure (15): The User Interface for Creating FPN
Model.
Figure (16) shows the FPN model of proposed system:-
Figure (16): The Fuzzy Petri Net Model.
V. B-4. FPN Inputs
The input to the proposed system is three
parameters (number of hops, Local Battery level,
Received Signal Strength Indicator), that Fuzzified by
using triangular membership function, where its input
values is converted to fuzzy values according to
membership function that related to each input
parameters as follows:-
A. Number of Hops (N):- This is the length of the
path. This input variable is divided to three fuzzy sets
as shown in table (6), and the membership function is
shown in figure (17).
Table (6): Fuzzy Sets for The Number of Hops (N).
Input Range to The
Number of Hops (N)
Fuzzy Set
0
–
10
Low (L)
10
–
30
Medium (M)
20
-
40
High (H)
Figure (17): Membership Function for Number of Hops
B. Local Battery Level (B):- This represented the
battery level of the node, this input variable is divided
to three fuzzy set as shown in table (7), and the
membership function is shown in figure (18).
Table (7): Fuzzy Sets of Local Battery Level (B).
Input Range of Local
Battery Level (B)
Fuzzy Set
0
–
50
Low (K)
25
–
75
Medium (R)
50
–
100 %
High (T)
Figure (18): Membership Function of Local Battery
Level (B).

Researcher 2014;6(8) http://www.sciencepub.net/researcher
42
C. Received Signal Strength Indicator (S):- This
represented the strength of the received signal is an
indicator of the quality of communications between
two nodes. This input variable is divided to three fuzzy
set as shown in table (8), and the membership function
is shown in figure (19).
Table (8): Fuzzy Sets of Received Signal Strength
Indicator (S).
Input Range of Received
Signal Strength Indicator (S)
Fuzzy Set
0
–
50
Low (U)
25
–
75
Medium (V)
50
–
100 %
High (C)
Figure (19): Membership Function of Signal Strength
Indicator.
D. Optimal Path:- This is the output of fuzzy
system that represents the suitability of a node to be
considered for inclusion in the route. This output is
divided to three fuzzy sets as shown in table (9), and the
membership function is shown in figure (20).
Table (9): Fuzzy Sets for The Optimal Path.
Input Range for The Optimal
Path
Fuzzy Set
0
–
0.5
Low (LL)
0.25
–
0.75
Medium (MM)
0.5
–
1
High (HH)
The approach of fuzzification is used to obtain
the fuzzy membership degrees for each input
arguments (number of hops, Local Battery level,
Received Signal Strength Indicator), as mentioned in
chapter four, by using triangular membership function.
Defuzzification is the final phase in any fuzzy system,
as explained in chapter four. According to this phase,
the decision making takes place about which path is
selected to be output of proposed system (optimal
path), with some conclusions such as (Path number,
Path nodes, Path gain value), after input random fuzzy
initial values. In summary, the path with the higher
gain is selected and the information on this path is
then used for providing an explanation to the user, as
shown in the following figure (21).
Figure (20): Membership Function of Optimal Path.
Figure (21): Decision Making and Explanation.
VI. Conclusions
1. The routing is one of the most important
challenges in ad-hoc networks due to absence of
central administration and mobility of nodes, therefore,
many techniques ware development to deal with it like
fuzzy logic.
2. The network simulator version 2 (NS-2) was
used to build a system for studying and evaluating the
behavior of two routing protocols (AODV and

Researcher 2014;6(8) http://www.sciencepub.net/researcher
43
AOMDV). From the difference in results obtained
after running the simulation for 10 iterations, we
conclude that the path selected from this routing
protocol is not the optimal path.
3. The lack of an efficient metric to evaluate
node conditions in routing protocols, has been solved
by the definition of a new efficient metric based on the
combination of different node and new network
parameters by using a Genetic Fuzzy Petri Net system.
4. The Genetic Fuzzy Petri Net system is used to
select the optimal path, depending on three fuzzy
input parameters (number of hops, Local Battery level,
Received Signal Strength Indicator) and a set of fuzzy
rules which are created from training phase of genetic
algorithm. This model is used in order to mimic the
route discovery mechanism in ad-hoc routing
protocols to select the optimal path.
5. Genetic Fuzzy Petri Net system combines the
advantage of both fuzzy logic and petri nets, to create
useful system that deals with uncertain information
and it can provide a visible modeling.
6. Genetic algorithm has an effective role for
generating set of fuzzy rules by training stage and
selecting the best between them.
7. When the data base is represented as a tree
structure, it make the searching operation more
powerful and faster.
8. Using the fuzzy logic as a metric in network
routing improves the performance of real networks.
9. Using the fuzzy logic as a metric in network
routing improves the performance of the intelligent
dense monitoring of the network, thus, led to
increasing quality of service.
10. The fuzzy logic is a useful and powerful
approach that has demonstrated to be effective when
combined with other disciplines such as routing
approaches for wireless ad-hoc (MANET) networks.
VII. Future Works
This work can be extended in different directions,
in the following some suggested ideas are given:-
1. Future research can be done by the addition
of new parameters as input to the fuzzy inference
system and studying the performance achieved by
these new parameters, such as the bandwidth, the load
of the link, the traffic load, the power consumption,
the total vector cost, the time to life for the path and
the node density.
2. In the future research one can use the fuzzy
logic in other ad-hoc layers, such as the MAC layer,
that will help to provide a priority in the contention
period for candidates nodes with better conditions.
3. Depending on the proposed Genetic Fuzzy
Petri Net model that is explained in this thesis, the
optimal path can be selected. Therefore, a future
research can programming the approach of this model
in C++ language, in order to combine the result code
with routing protocol code in the part that allocated
for route discovery process to obtain an efficient
routing protocol.
References
1. Anjum Mohammed, Gihan Nagib, “Optimal
Routing In Ad-Hoc Network Using Genetic
Algorithm”, Int. J. Advanced Networking and
Applications, Volume: 03, Issue: 05, Pages:
1323-1328, 2012.
2. Akshatha. P. S, Namita Khurana, Anju Rathi,
“OPTIMAL PATH FOR MOBILE AD-HOC
NETWORKS USING REACTIVE ROUTING
PROTOCOL”, International Journal of Advances
in Engineering & Technology, Sept, 2011.
3. Aliff Umair Salleh, Zulkifli Ishak, Norashidah
Md. Din, and Md Zaini Jamaludin, “Trace
Analyzer for NS-2”, IEEE, pp. 29-32, Student
Conference on Research and Development
(SCORED), Malaysia, 2006.
4. Vasilis Gerogiannis, D.Achilles Kameas,
E.Panayotis Pintelas, “Comparative study and
categorization of high-level petri nets”, The
Journal of Systems and Software (43) 133±160,
1998.
5. Er. Deepinder Singh Wadhwa and Er. Tripatjot
Singh Panag, “Performance Comparison of
Single and Multipath Routing Protocols in Ad
hoc Networks”, Int. J. Comp. Tech. Appl., ISSN:
2229-6093, Vol 2 (5), pp.1486-1496, 2011.
6. Jae Chung & Mark Claypool “NS by Example”,
worcester polytechnig institute, computer science,
2002. Available in the page:
http://nile.wpi.edu/NS/ [08/01/2002 9.09.25].
7. Liu, J.HU and J.LI, “Diagnostic knowledge
representation and reasoning based on fuzzy fault
petri nets”, IEEE, 2011.
8. M. Georgi Dimirovski, “FUZZY-PETRI-NET
REASONING SUPERVISORY CONTROLLER
AND ESTIMATING STATES OF MARKOV
CHAIN MODELS”, University of Istanbul,
Computing and Informatics, 563–576, Vol. 24,
2005.
9. Mamoun Hussein Mamoun, “A New Reliable
Routing Algorithm for MANET”, International
Journal of Research and Reviews in Computer
Science (IJRRCS), Vol. 2, No. 3, June, 2011.
10. Oscar Cordon, Francisco Herrera, Frank
Hoffmann & Luis Magdalena, “GENETIC
FUZZY SYSTEMS: EVOLUTIONARY
TUNING AND LEARNING OF FUZZY
KNOWLEDGE BASES”, Book, Copyright by
World Scientific Publishing Co. Pte. Ltd. Vol. 19,
2001.
11. SEM_H DOKURER, “SIMULATION OF

Researcher 2014;6(8) http://www.sciencepub.net/researcher
44
BLACK HOLE ATTACK IN WIRELESS
AD-HOC NETWORKS”, Master thesis in
Computer Engineering submitted to Atılım
University, 2006.
12. Siddesh. G.K, K.N. Muralidhara and Manjula.
Harihar, “Routing in Ad Hoc Wireless Networks
using Soft Computing techniques and
performance evaluation using Hypernet
simulator”, International Journal of Soft
Computing and Engineering (IJSCE) ISSN:
2231-2307, Volume-1, Issue-3, July, 2011.
13. Sonam Jain and Sandeep Sahu, “Geometric
Routing Protocol based on Genetic Algorithm for
Minimized Delay in MANETs”, International
Journal of Electronics Communication and
Computer Technology (IJECCT) Volume 2 Issue
3 , May, 2012.
14. Subramanyam, P.V & Chauhan, A, Singh,
“Modelling of a Hybrid Protocol for a MANET
using Fuzzy Petri Nets”, India Conference, 2006
Annual IEEE, Print ISBN:1-4244-0369-3, Sept,
2006.
15. Taqwa Odey Fahad & Prof. Abduladhim A. Ali,
“improvement of AODV routing on MANETs
using fuzzy systems”, University of Basra, Iraq J.
Electrical and Electronic Engineering, Vol.7
No.2, 2011.
16. Tzu-Chiang Chiang and Yueh-Min Huang,
“Multicast routing representation in ad hoc
networks using fuzzy Petri nets”, Advanced
Information Networking and Applications, 2004.
AINA, International Conference on (Volume:2 ),
2004.
17. Tzu-Chiang Chiang, Cheng-Feng Tai and
Ting-Wei Hou, “A knowledge-based inference
multicast protocol using adaptive fuzzy Petri
nets”, International Journal of Expert Systems
with Applications, Vol 36, Issue 4, Pages
8115–8123, May, 2009.
18. X. Li, F. Lara-Rosano, “Adaptive fuzzy petri nets
for dynamic knowledge representation and
inference”, Centre for Instrumentation Research,
National University of Mexico (UNAM), Expert
Systems with Applications (19) 235–241, 2000.
19. Zahra Masoudi Nejhad, Ali Harounabadi and
Mohammad Teshnehlab, “ Mapping fuzzy
formal model (FPN) to fuzzy extended markup
language (FXML), (FPNML)”, Organized by
Khavaran Institute of Higher Education,
Bandar-Abbas, Iran, 2013.
20. Zhenhua Yu, Xiao Fu, Yuanli Cai and Mehmet C.
Vuran, “A Reliable Energy-Efficient Multi-Level
Routing Algorithm for Wireless Sensor
Networks Using Fuzzy Petri Nets”, journal
sensors ISSN 1424-8220, 11, 2011.
21. Zhi-gang Hu, Hao Ma, Guo-jun Wang, Lin Liao,
“A reliable routing algorithm based on fuzzy
Petri net in mobile ad hoc networks”, Journal of
Central South University of Technology, Volume
12, Issue 6, pp 714-719, December, 2005.
8/3/2014