CLBNSRM - CONFIDENCE LEVEL BASED UNBLEND NEIGHBOR SELECTION &BLEND NODE REPORT BASED OPTIMIZED ROUTE FORMATION IN MANET

International Journal of Computer Networks & Communications (IJCNC) Vol.12, No.2, March 2020
DOI: 10.5121/ijcnc.2020.12206 109
CLBNSRM - CONFIDENCE LEVEL BASED
UNBLEND NEIGHBOR SELECTION &BLEND NODE
REPORT BASED OPTIMIZED ROUTE FORMATION IN
MANET
S. J. Patil1
,L.S. Admuthe2
and M. R. Patil3
1
Research Scholer,Department of Electronics and Communication,
VTU, RRC, Belgavi, Karnataka, India.
2
Department of Electronics Engineering, DKTE’S Textile & Engineering Institute,
Ichalkaranji. Maharashtra, India.
3
J.A.G.M. Institute of Technology, Jamkhandhi, Karnataka, India.
ABSTRACT
A mobile Ad-hoc network (MANET) is an impulsive network that can be recognized with no predetermined
infrastructure. To achieve safe path selection cryptographic key exchange was implemented mostly in turn
of huge computational cost. Confidence based coordination in MANET focuses on routing challenges
created by selfish nodes, as energy utilization & time factor are key issues in this aspect. The present
protocol is focused on fuzzy optimization-based node confidence estimation and path selection with
minimum energy utilization. The node with maximum confidence value will give high priority to include in
the path for transmission. In the implemented protocol to build a novel confidence-based model
multidimensional factors like confidence value, link cost, degree of node and node energy are included as
decision-making factors. The proposed protocol CLBNSRM estimates confidence level in four steps to
decide a trustworthiness of neighboring node. To estimate the efficiency of the present confidence model
various protocols are compared by using attributes like the number of nodes, node speed, malicious node
variation, etc. Moreover, different parameters like Packet delivery ratio, Throughput, Residual energy, and
Packet dropped are considered with these attribute variations. Experimental results indicate that PDR and
Throughput increase although in presence of malicious nodes, along with the utilization of minimal energy.
Statistical analysis is carried out for mathematical modeling. This analysis shows that a linear model of an
implemented protocol is better than compared protocol with all the aspects.
KEYWORDS
MANET, Path Selection, Confidence, Priority, Fuzzy Optimization.
1. INTRODUCTION
A mobile Ad-hoc network (MANET) is a self-constructed system with many wireless mobile
nodes. The nodes in the network do not only act as destination node but also forward packets to
neighbor nodes as a router. Hence, its popularity has been increased due to its vast application in
the field of military action, emergency rescue operations, law enforcement and security prone
environments. Rapid expansion in mobile devices and interest in mobile communication, mobile
Ad-hoc networks have been receiving a lot of attention in the recent past [1].
The mobile Ad-hoc network (MANET) is susceptible to various attacks due to its features like
dynamic topology, open environment, limited physical security, limited bandwidth, energy
exhaust, etc. The various attacks are imposed in MANET due to its easy compromisation [2].

110
Since the attacks like a black hole, gray hole, DoS are interrupting node communication lead to
losing integrity, loss of authenticity and exhaust of battery power consequently collapse of the
network. Hence, standard Ad-Hoc communication needs data validation against misbehaved
nodes. Trust estimation based structural design is the recent trend in the mobile network. In the
past, key sharing security systems are engaged to attain defence however, they might have invited
further computational cost and raises the overhead complexity. The cryptographic key sharing
consists of encryption and decryption methods with a reliable third-party device placement that
controls the key sharing, signatures, hash functions and node privacy [3]. But these judgments fail
to handle the malfunction of node position, where the node drops the packets. In case, that data
gets modified by any node is also needed to be checked and eliminated from the network.
Confidence based defence mechanism is considered to compute the belief cost of each node based
on the sequence of contact records passed out by that node. The character-based confidence
values are computed based on unblendUB(one hop) and blend BN(two hop) node reports. If the
contact is unbeaten, then the communicated node is measured as a confident device. The
unblend(UB) contact may be bogus due to, instant link loss and blockage. Although confidence is
computed based on character record, for that correct validation is required to carry further
communication. If the volume of the contract record increases, the confidence level get weakens.
The confidence level based optimized path selection preserves confidence accounts of each
device and uses effortless statistical functions to work out the confidence in its nearby nodes. The
CL based path establishment using fuzzy logics not only watching the character of the node but
also assists to pick up the output result. In the environment, the node accumulates a packet of the
previous node and retransmit it to the next hop up to the ultimate destination. The broadcast
among the nodes measured as uniform qualities and it uses 802.11 wireless network MAC
standards. These wireless channels are susceptible to some form of harassment imposed by
egotistic or miss behaved nodes.
In this proposed protocol, opinion-based confidence assessment and route formation based on the
fuzzy decision is implemented. The discrimination of the reliable node and a malicious node is
identified by computing confidence level CL for all nodes based on the UB knowledge and advice
of the node from its nearby nodes.
The rest of the article is organized as follows: Section 2 gives a short overview of trust
management and routing protocols. The points and steps made to implement CLBNSRM are
described in Section 3. In section 4 simulation model and parameters considered are discussed. In
section 5 results are discussed and statistical analysis is carried out. InSection 6, the conclusion of
the present study and scope for future work are discussed.
2. RELATED WORKS
In recent years, a lot of research work has been undertaken on trust management models, attacks
on MANET and trust-based protocols. Although the application of trust management models in
mobile agent systems has received much attention, still overall efficient models including the
energy system have to be received. In this section related research work regarding trust
management models, attacks and trust - based protocols with fuzzification are discussed in brief.
Generally routing protocols in Ad-hoc networks are classified as proactive & reactive. Proactive
routing protocols are table-driven protocols and need more computational cost. This may result in
to high utilization of bandwidth and energy which are generally known to be limited resources.
However, reactive routing protocols are on-demand routing protocols which find the path to the
destination whenever necessary, hence limited resource like bandwidth and energy are conserved.
Therefore, later one has driven the attention of researchers in MANETs. Perkins et.al.[4] have put
forwarded single path routing protocol AODV which is the combination of DSR and DSDV. In
this protocol, authors consider that every node is co-operative and honest.

111
In the existence of security protocol reduction of attacks can be achieved. The mobile hosts
dynamically found paths among one another for communication. Consequently, the
accomplishment of MANET communication highly relies on the association of the involved
mobile nodes [5]. MANET, networks are more susceptible to attacks than infrastructure networks.
So, safety is a vital issue in MANET to afford secure communication among mobile nodes [6].
During the previous decade, broad studies have been conducted on routing in mobile ad-hoc
networks, and have resulted in a number of established routing protocols [7]. Ad-hoc networks
frequently suffer from malicious attacks because of its features like dynamic topology, lack of
central monitoring and management, open medium, no clear defence mechanism. These factors
have distorted the combat field circumstances for the MANET against the security pressure [8].
The ultimate goal of the security solutions for MANETs is to provide security evinces, such as
authentication, confidentiality, integrity, anonymity, and openness, to mobile users [9]. The
design behind hybrid routing protocols is to employ both routing protocols, proactive mechanisms
in some areas of the network at convinced times and reactive routing for the rest of the network.
The proactive operations are restricted to a small domain in order to diminish the control delays
and overheads [10].
Trust-based reactive routing protocols like trusted AODV, DSR & TORA were analyzed through
their performance with respect to variation in the number of malicious nodes along with another
experimental setup Pirzada et.al [11]. According to Guo. et. al. [12] Trust-DSR facilitates five
route selection strategies that are dependent on trust evaluation of transmission links. Since the
selection of route is limited on route gained from standard DSR. However, the ultimately selected
route might not be necessary the most trusted. Xia. et.al [13] have discussed the trust
administration model for mobile Ad-hoc networks based on the systematic hierarchy process and
fuzzy theory. In this research trusted routing algorithm, reactive routing protocol based on
standard dynamic source protocol and fuzzy trusted dynamic source routing protocol is proposed.
In this work, authors have focused on direct trust, recommended trust, incentive function, and the
active degree to calculate overall trust. Furthermore, in recommended trust direct
recommendation experience and indirect recommended experience have been considered as
analytical hierarchy process and fuzzy theory.
A fuzzy-based Ad-Hoc on-demand distance vector routing protocol was put-forwarded by
Manicakam [14] and fuzzy logic is used for evaluating trust. The Threshold trust value was set so
as to verify the trust on the neighbor’s node. Here, in this protocol only modified attacks were
identified, that only at the route discovery stage. Junhai et.al [15] implemented a trust model
based on the fuzzy recommendations for mobile Ad-hoc networks. In this model, authors have
included five types of fuzzy trust recommendation relationships based on fuzzy relation theory
and a mathematical description for MANET’s. Furthermore, authors have considered fuzzy direct
trust model, fuzzy indirect trust model and the fuzzy global trust model for calculating overall
trust on the neighbor node. In-depth, the fuzzy recommendation trust model, fuzzy transitivity
recommendation trust model and fuzzy consensus recommendation trust model were considered
for calculating the fuzzy indirect trust. This protocol has considered average energy consumed as
a matrix, but still, energy cost for computing trust value is not considered. Furthermore, this
model neither considered average energy consumed nor residual energy in comparison with other
trust models.
Shuaishuai et. al. [16] have proposed a novel trust management system, so as to secure the data
plane of Ad-hoc networks. In this management system fuzzy logic is used to calculate the path
trust while graph theory is adopted to calculate node trust value. In fuzzy-OLSR, authors have
[17] considered only first handed information and purposefully avoided second-handed
information to overcome the overhead problem in the network. But in this model along with first-
hand information, second and third hand information are gathered so as to build not only trust but

112
also confidence on a neighboring node. Although in fuzzy-OLSR importance given to overhead,
the confidence on the neighboring nodes is more important hence the present study is focused on
secured path without running the confidentiality. However, energy -efficient module is also
included so as to improve the performance of the protocol.
3. FUZZY BASED CLBNSRM IMPLEMENTATION
In the present confidence model estimation of confidence, assessment of confidence, estimation
of UB confidence of nodes, confidence of blend nodes, and overall, the confidence level was
considered.
3.1 Confidence Estimation:
Confidence is a conviction of a node on a new device for a definite mission. A confidence
network is measured as a bound for subjective chart C={D, N, t},where D=(d1, d2, ….dn)describes
the device counts in the region and N = {n12, n23 …nij}, 𝑖 ≠ 𝑗, is the confidence association
between neighbors inD, nij is the UB link between two neighbors di and dj, that bringst. The
confidence assessment between 0 to 1, signified by,CL (nij) ɛ [0,1], and |N| is the count of UBlinks
in the environment.
 All nodes observe and record the character of its neighbors in the random deployment
environment.
 Each node will also collect the character of its next neighbor from blend nodes (BN) and
record it for additional validation.
 Every node has the knowledge to compute the confidence of its neighbors based on the
transactions handled between them.
Usually, trust-based protocols do not give precise results with respect to stratification in depth.
Here, fuzzy logic is a significant tool through which one can precisely stratify the confidence
level. Hence, fuzzy logic is used to compute CL based routing for proper decisions.
3.1.1 Confidence Assessment:
Confidence assessment (CA) is computed for each node to recognize its character and consistency
for efficient contact. To achieve the confidence level (CL) from the network the subsequent
assumptions are prepared
 The communicating nodes forever having confidence in themselves.
 Almost all nodes should behave & co-operate with each other.
 A Minimum number of nodes might misbehave as selfish in the network.
 The CL of the route is computed based on the honest transaction between nodes.
Nodes need to check the packet forward count (FC) which is the ratio of the number of packets
forwarded correctly to the number of packets supposed to be forwarded. Packet lifetime (PLT) is
the TTL value from the packet header. HopCount(HC) is total number of hops. Packet Loss Count
(PLC) is the total number of packets dropped in the network and Resend (RS) is the timer, when it
becomes zero, source has to send packet once again. Source packets can be retransmitted based
on the complete confidence in the route nodes and the strength of the channel in the path. A
confidence assessment needs to record honest co-operation between the UB nodes which is shown
in Figure.1 and their third neighbor confidence report. In this way selfish behavior is measured in
this network.

113
Figure1.Unblend Links
Algorithm 1: Confidence Assessment AC
Algorithm 1 (pseudo code) shows how confidence assessment is done. Initially, packet
forwarding count and packet loss count are checked with CL. If it is less than CL, then two more
conditions are checked such as hop count and packet life count. If these are greater than zero the
packet will be sent to next node j, otherwise, packet will be deleted.
3.1.2 UB Confidence Estimation between Nodes (Direct):
The UB confidence is computed by estimating the characteristics of its face-to-face next UB
neighbor reportsCR. To compute the UB node confidence at every interval, just make- believe that
the confidence implications of each node must be equal to 1 at the initial stage of the
communication to all nodes. Algorithm 2 gives an idea about how the UB trust estimation between
two adjacent nodes. The received data is checked for hop count, end of the queue and validation
of packet expiry count, if it satisfies, then packet forward count will be updated otherwise PLC
will be updated. If it doesn’t satisfy the above condition, then UB trust level will be updated by
using the formula stated in equation 1.
A B C D
Confidence on A by B
Confidence on B by A
UB LINKS
Initiate LC Verification(i,j)
{
if ( BU (i,j)node CF = LC && 𝑃𝐿𝑐< 𝐶𝐿 then
{
if ( CH >0) then
{
Check LTP > 0
Forward Data to j
} else
{
Drop packet
}}}

114
Algorithm 2: UB Trust Estimation between Nodes
UB𝐶𝐿( 𝑖𝑗) =
𝑠
s+(𝛾∗𝑓)
(1)
Where,
s = packet send number.
f = Received Hello count
γ = Conversion factor.
Which is calculated as follows:
γ= 1 +
log10
(1+𝑓)
(𝑠+𝑓)
log10 2
(2)
3.1.3 BLEND Nodes Confidence Report (Indirect):
The BN confidence report is collected based on the suggestions from the other neighbors. In this
case nodeihas zth
adjacent nodes in the surroundings where,ρ ={n1,n2….nz} that is sum of nodes in
the network. Imagine that node ilikes to maintain a confidence-based dealing with its second BN
inside ith
area that is HC>1.
The BN confidence is gained from the nearest nodes reports as shown in Figure 2. Initially, node
verify the CL of all its nearby UBneighbors and make a decision to select a collection of nodes
whose confidence (CL>L) than the given limitations L as shown in algorithm 3. Also, within the
node icoverage rangeRC, it broadcast theCL report through the message of confidence level report
announcement CLRA to all of its nearby nodes j. The CLRA message reached up to the BN node. Let
j, be the neighbor node and neighbor of neighbor, which is located within the UBand BN neighbor
of ith
node limitation. Later, the nodes of j will send a reply to the node ienclosed with CLreport as
CLRA-replaymessage. So, theBN report CL about node i can be computed and tag with reply message
as
𝐵 𝑁𝑖𝑗 = 𝑈 𝐵𝑖𝑗 × 𝑈 𝐵𝑗𝑧 (3)
Once receiving the data
if ((hc !=0) || (queue != deadline) {
if (𝑅𝑆 < 𝑃𝐿𝐶) {
Validate packet expiry time, before PLC, if yes
𝐹𝐶 = 𝐹𝐶 + 1 Update forward count
} else
{
𝑃𝐿𝐶 = 𝑃𝐿𝐶 + 1 Declare Attack Presence
}
}
j drops data after collecting, then validate
else if (ℎ𝑐 == 0) || (Queue == Deadline) {
𝑃𝐿𝐶 = 𝑃𝐿𝐶 + 1 Selfish Node
}
Update the UBCL.
return TL

115
Figure 2. Blend Node Confidence
Algorithm 3: Blend node confidence check
The BN nodeCL report obtained from the nearby second hop neighbor nodes may beCL< L, then the
observing confidence report obtained from the CR of a node may produce the terrible suggestions
about ith
node. So, the witnessing nodes of neighbors notices and removing the selfish nodes
whose CL is less than the L and thus the network is protected by the unkind remarking nodes. The
CL of all nodes will be computed and updated in a CL report as shown in table 1. Here the value of
CL sorts off least value 0 which signifies as dishonest node to maximum value 1 indicates the
complete CL of a node.
Table 1: Node CL Report
CL CL Linguistic
3.00  Absolute Selfish
5.03.0  n Suspicious
7.05.0  n Least CL
8.07.0  n CL
18.0  n High CL
NB Checks LC (i,j) in second hop {
Fori = 0 to LRAC Message time {
Obtain thenearby nodes LC Report
node[i]=id(i);Check Node address as identification
}
if LC ((node[i] 𝑈 𝐵) > L ) then {
LC = LC +1
LC [i] =
𝑈 𝐵𝑖𝑗
∑ 𝑈 𝐵𝑖𝑗
𝑧−𝑠
𝑗=1
,
}}
}
A B
D
C
Unblend Link
Blend Link
C – States the Confidence about B to A

116
The algorithm 4 shows that the device i can construct a CL path to j through z - s routs, in case the
BNCL> L of j. If so, node i will select one of its nearby devices with the greatest CL, to resend the
data to j. The fuzzy CL path selection supports the linguistic form (LF) as confidence scores of a
node on the other nodes.
Algorithm 4: Blend nodeBN confidence CLupdate
3.1.4 Overall Confidence Estimation
Equation 4 depicts that the overall confidence is computed based on the suggestions observed
from the nearby nodes, as well, based on direct confidence. Each UB and BNwill compute the CL
of its nearby nodes based on the truthful packet relocation within the expiry period. At initial CL
and the link, abilities are assigned as 1.
𝑂𝐶 = 𝛽 ∗ 𝑈𝐵𝐶𝐿 + (1 − 𝛽) 𝐵𝑁𝐶𝐿 (4)
3.2. Fuzzy Logic
Fuzzy logic is a multi-valued logic in which the values of variables may be any real number
between true (1) and false (0). With fuzzy logic, an input can be mapped into an output space.
Figure 3. Fuzzy logic architecture
Fuzzy logic is composed of if-then rules. The if-part of the rule is called the originator while
thenpart is called the consequential. A fuzzy logic system maps crisps inputs to crisp outputs.
There are four mechanisms in a fuzzy logic system, namely rules, fuzzifier, inference mechanism,
and defuzzifier as shown in Figure 3. Rules are fundamentally if-then rules which must be
evaluated during an input/output process. The output of the system depends on these rules.
Fuzzifier is responsible to take crisp numbers as input and give fuzzy sets as output. The
beginning of rules is dependent on the output of the fuzzifier. The Inference mechanism in the
fuzzy logic system is the decision making part. Defuzzifier maps the fuzzy output of the inference
For i =1 to d
{
Update CL [j]
}
ith
node updates the LC from zth
node by
CL [i] =
𝑈 𝐵𝑖𝑗
∑ 𝑈 𝐵𝑖𝑗
𝑧−𝑠
𝑗=1
, else
}
Update CL
Rule Base
Fuzzifier
Inference
DefuzzifierCrisp Input Crisp Output

117
mechanism into crisp numbers to make it function for further processing by the system [18]. A
triangular fuzzy number is chosen for better results and can be defined by a triplet (a1, a2, a3).
The membership function is shown in equation 5.
𝜇𝐴( 𝑥) =
{
0, 𝑥 < 𝑎1
𝑥 − 𝑎1
𝑎2 − 𝑎
, 𝑎1 ≤ 𝑥 ≤ 𝑎2
𝑎3 − 𝑥
𝑎3 − 𝑎2
, 𝑎2 ≤ 𝑥 ≤ 𝑎3
0, 𝑥 > 𝑎3
(5)
3.2.1 Fuzzy Inputs:
Total four inputs (Linguistic variables) used as crisp set of fuzzy. These are discussed in this
section
1. Node CL: Node CL above 0.8 in the path is considered. Here node CL is calculated based on
blend confidence and suggestion of nearby nodes. The mean value of the unblend LC of the
nearby nodes that are having a superior suggestion is only considered.
𝐶𝐿 = 𝛽 ∗ 𝑈𝐵𝐶𝐿 + (1 − 𝛽) 𝐵𝑁𝐶𝐿 (6)
2. Channel Bandwidth: It is considered to know the capacity of the link. It is a ratio of
bandwidth to node MAC bandwidth.
𝐶 =
𝐵𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ
𝑁𝑜𝑑𝑒𝑀𝐴𝐶𝑏𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ
(7)
3. Hop count: It gives the total number of intermediary nodes through which data must go by a
source to destination. It is considered to get the shortest path from source to destination. Energy
obligation is directly proportional to the hop count so the hop count plays a major role in saving
the node energy.
4. Node Energy: The node energy is calculated by using the following formula and used as one
of the inputs of a fuzzy crisp set.
𝐸𝑔𝑦 = ∑
𝐸𝑔𝑦
𝐼𝐸
(8)
Where, IE is the initial energy
3.2.2 Fuzzification membership functions:
The set of memberships for the given inputs are the triangular function and the different range of
the input as mentioned in table 2.
'

118
Table 2: Parameters Chart
Parameters Rules
CL Superior Standard Low
Hops Superior Standard Low
Channel
Bandwidth
Superior Standard Low
Node
Energy
Superior Standard Low
In general, using fuzzy inference decision-based path selection corresponds to CL specified as
linguistic forms (LF). The process of Fuzzification links the input to the equivalent LF as low,
standard and superior as shown in figure 4.
Figure 4. Membership Functions
3.2.3 Output of fuzzification:
The final output of the present fuzzification model obtained in five levels as shown in figure 5.
Figure 5. Defuzzyfied results for CL
In the present model, four inputs were given to the fuzzy system and five levels of confidence
were obtained varying from 0 to 1. Here 0 indicates poor CL while 1 indicates superior high CL. In
between poor and superior high another three levels of CL were obtained that is a low standard
and good. The collection of LF sends the CL of the route is given by
CL-PATH = Superior high / Good / Standard / Low / Poor as end result.

119
3.3 Statistical Analysis:
The statistical analysis was carried out by feeding the results of protocols in the software R-3.4.2.
The correlation analysis was done for attributes and parameters regarding to all three models. To
check significance level p-values were calculated. Furthermore, the generalized linear regression
model was fit for protocols to obtain future values of particular attributes for various parameters.
4. PERFORMANCE EVALUATION
4.1. Performance Evaluation
To observe the performance of CLBNSRM, AOTDV and AOMDV NS-2.34allinone simulator is
used. The application layer protocol used was Constant Bit Rate (CBR) it generates the data in the
network. The Transport agent used was User Datagram Protocol (UDP) which configures the
transport layer. Two Ray Ground model was used as the propagation model. This model is
appropriate for long-distance communication. As shown in table 3 the simulation model considers
a square network area of 1000m x 1000m to transmit data packets of 512 bytes. The node
variation considered is 50 to 150 with a step size of 25 nodes. The network is simulated for 100
seconds and repeated for various attributes. The performance of the proposed CLBNSRM is
evaluated and compared with the existing AOTDV and AOMDV. To get better results from
various attributes like number of node variation, node speed variation and the number of
malicious nodes variation is considered. The simulation parameters are summarized as shown in
Table 3.
Table 3: Simulation parameters considered
Network Simulator NS-2.34allinone
Network area 1000 X 1000
Number of nodes 50, 75, 100, 125, 150.
Speed of the nodes 0, 5, 10, 15, 20, 25, 30m/s.
Malicious nodes 1, 2, 3, 4, 5.
Traffic load CBR
Packet Size 512 bytes
MAC protocol IEEE802.11b
Simulation time 100s. and repeated for various attributes
4.2 Performance metrics considered
4.2.1Packet delivery ratio: The ratio of the data packets delivered to the destination nodes to
those sent by the source nodes.
4.2.2Throughput: It is the average rate of successful message delivery over a communication
channel.
4.2.3Residual Energy: The remaining energy of the nodes in the network is called as residual
energy.
4.2.4 Packets Dropped: It is the total number of packets dropped in the network.

120
5. RESULT AND ANALYSIS
In the present study number of performance, matrices is analyzed for AOMDV, AOTDV, and
CLBNSRM and represented in graphs. Performance matrices like Packet delivery ratio, Number
of packets dropped, Throughput and Residual energy is considered. These performance matrices
also corresponded with various attributes like the number of nodes, node speed and number of
malicious node variation.
In node variation, the number of nodes was varied between 50 and 150 with a step size of 25
nodes, node speed 25m/s, pause time 20s, area considered 1000m x1000m and simulation time
100s. In the node speed variation, the nodes start with a low velocity of 0 m/s and then the node
velocity increases up to 30 m/s. The data rate is kept constant and the number of nodes and pause
time was fixed at 50 and 25 respectively.The third scenario is considered by varying malicious
nodes from 1 to 5. In this setting the total number of malicious nodes in the network was inserted
purposefully to observe its effect on different parameters like PDR, throughput, packet dropped,
residual energy, etc.
5.1 Packet delivery ratio:
Figure 6. Comparison of nodes Vs packet delivery ratio.
The Packet delivery ratio has been evaluated by corresponding node variation which is presented
in figure 6. Results show that PDR was more in CLBNSRM as compared to AOTDV and
AOMDV in every variation with respect to a number of nodes. Among the three protocols
proposed CLBNSRM shows the best output than others. Due to the selection of neighbors based
on confidence the path was constructed properly. Ultimately minimizes the packet misuse and
dropdown hence increases the PDR.

121
Figure 7. Comparison of node speed Vs packet delivery ratio.
The Packet delivery ratio was corresponded with node speed and is shown in figure 7. The packet
delivery ratio with reference to AOMDV and AOTDV shows dropping initially up to node speed
15 m/s and a gradual increase was noted up to 30m/s and thereafter again it was dropping.
However, PDR of CLBNSRM has a quietly different scenario. The PDR is comparatively higher
and there was no effect observed on PDR with the change in node speed. Precisely, one can say
that PDR is significantly better and undisturbed with variation in the node speed.
Figure 8. Comparison of malicious nodes Vs packet delivery ratio.
The packet delivery ratio is also corresponded with variation in a number of malicious nodes and
exhibited in figure 8. The results for AOMDV and AOTDV show that PDR was lower and further
the PDR of AOMDV was lowest than even AOTDV. In comparison with CLBNSRM, it was far
better than the other two protocols.

122
5.2 Throughput:
Figure 9. Comparison of nodes Vs throughput.
Evaluation for throughput with respect to AOMDV, AOTDV, and CLBNSRM is given in figure
9. The results of throughput for AOMDV was noticed declining as the number of nodes get
inclined up to 100 nodes and slightly inclined at 125 nodes and again declined. However, the
results of AOTDV were quite good with reference to the throughput but not satisfactory. In
CLBNSRM, throughput level was noticed much higher than the other two protocols. The
Throughput of CLBNSRM was declined with an incline number of nodes, but the variation is not
significant as compared to AOMDV and AOTDV.
Figure 10. Comparison of nodes speed Vs throughput.
Throughput has corresponded with node speed which was shown in figure 10. The throughput of
the present protocol has not much affected by node speed. However, node speed significantly
affected throughput with reference to AOMDV and AOTDV. In AOMDV and AOTDV,
throughput got declining as the speed of nodes increased up to the certain extent and again gets
increased and the sudden drop was noted. The result of CLBNSRM for node speed variation was
higher as well as constant.

123
Figure 11. Comparison of malicious nodes Vs throughput.
Throughput has also corresponded with variation in the number of malicious nodes and results are
exhibited in figure 11. The Throughput of AOTDV and CLBNSRM was the same when the
numbers of malicious nodes were less. But as the number of malicious nodes gets increased in the
network, the throughput of AOTDV gets declined while in the case of CLBNSRM it was constant
throughout the variation in the number of malicious nodes. On the other hand, the throughput of
AOMDV was declined initially and later on it was increased.
5.3 Residual Energy:
Addressing the energy issues in MANET is a need of the present day. There are several protocols
in which energy as a parameter was not at all considered. While there are some protocols that
have considered energy as parameters. But the energy efficient model was not put forward by any
researchers. Here in CLBNSRM protocol, an emphasis is given to energy efficiency. The results
of residual energy with respect to node variation for all three protocols are depicted in figure 12.
Consumed energy was deducted from total energy to obtain residual energy, so, the residual
energy is inversely proportional to consumed energy. In the present study, residual energy for
CLBNSRM is higher throughout all variations. However, residual energy for the other two
protocols was gradually declining up to 125 nodes and thereafter slight increment was observed.
Figure 12. Comparison of nodes Vs residual energy.

124
Figure 13. Comparison of node speed Vs residual energy.
Present study is also carried out to evaluate residual energy about node speed (figure 13) and
found that CLBNSRM have better result than other two protocols. The study depicts that along
with increased node speed variation the residual energy is not at all fluctuate and also saved more
energy. It might be due to constant number of nodes which ultimately reduces the control
overhead in the network. Hence, in comparison with AOMDV and AOTDV, CLBNSRM is far
better to residual energy after varying node speed.
Residual energy was corresponded with introduction of malicious nodes in the network. The
results for all three protocols are given in figure 14. The study indicates that CLBNSRM was
notably higher as compared to AOMDV and AOTDV. The present protocol efficiently eliminates
malicious nodes from the network. Hence, unwanted energy consumption was get reduced and it
was conserved as energy.
Figure 14. Comparison of malicious nodes Vs residual energy.

125
5.4 Packet Dropped:
Figure 15. Comparison of nodes Vs packet dropped.
In figure 15. Packet dropped was corresponded to node variation for three protocols. The results
indicate that AOMDV and AOTDV have high packet drop ratio. However, in CLBNSRM packet
drop ratio is lower up to 100 nodes and there after slight increase was noted. Still in comparison
with the AOMDV and AOTDV, CLBNSRM was having minimum packet drops.
Figure 16. Comparison of node speed Vs packet dropped.
Packet dropped ratio was also corresponded to node speed (Figure 16). The results of packet
dropped verses node speed indicated that AOMDV as well as AOTDV have maximum packet
dropped at 15 m/s of node speed. In the case of both protocols, throughout all the node speed
variation packet dropped was noticed. However, In the case of CLBNSRM packet dropout ratio
was noticed minimum.

126
Figure 17. Comparison of malicious nodes vs packet dropped.
Packet drop to explain the network weakness and improper path selection to exchange the packets
between sources to a destination. The above figure 17 shows the minimum packet loss in
CLBNSRM than the other available protocols. This shows the strong path selection based on each
node confidence level using proposed protocol.
5.5 Statistical Analysis
5.5.1 Correlation Analysis:
In the implemented protocol three different attributes were corresponded with the parameters
Packet delivery ratio (PDR), Throughput (TP), Residual energy (RE) and Packet dropped (PD)
for AOMDV, AOTDV and CLBNSRM. Since to check Statistical relation between attributes and
parameters and also significance level was carried out.
Table 4: Correlation analysis of protocols.

127
Firstly, in the node variation there was a significant correlation corresponded with PDR, TP, RE
and PD as in all the parameters p-values are 0.00751, 0.00669, 0.1159 and 0.04241 respectively
as shown in table 4. Based on p-values it indicates that CLBNSRM is effective protocol than
AOMDV and AOTDV. Although p-value is not significant for malicious nodes and node speed
variations corresponded with various parameters in comparison with remaining protocols this
protocol shows quite significant.
5.5.2 Model Comparison:
In this section CLBNSRM is individually compared with AOMDV and AOTDV by estimating
paired t-test for various attributes corresponded with respective parameters and respective p-
values are noted.
Table 5: Comparison of CLBNSRM with other protocols by considering p-values.
P-values for PDR, TP, RE and PD about to nodes variation, malicious node variation and node
speed variation of CLBNSR with AOMDV and AOTDV is represented in table 5. The results
show, that CLBNSRM is highly significant protocol in comparison with AOMDV and AOTDV
based on of p-value.
5.5.3 Generalized linear model:
To estimate the mathematical model of implemented protocol generalized linear model is
considered. The model gives equation with y-intercept and coefficient of respective attributes as
shown in table 6. From these equations, one can estimate the future values of particular attributes.
Table 6: Linear Model between CLBNSRM with parameters and different attributes
6. CONCLUSION AND FUTURE SCOPE
In the present work, ideal confidence-based model has been proposed. Here unblended and blend
confidence were estimated and incorporated into fuzzy logic along with bandwidth, hop count
and node energy. This model creates a trustworthy path to the destination by excluding all
malicious nodes in the limelight of the fuzzification process. It also facilitates a reliable way to

128
deliver packets. The present protocol discovers multiple trustworthy paths from source to the
destination for proper communication. To estimate the efficiency of the CLBNSRM various
attributes and parameters were considered and compared with other reactive routing protocols
such as AOMDV and AOTDV. Furthermore, the energy module was included and it has been
found that the present model is energy efficient than the other considered protocols. In the
evaluation of various parameters like PDR, Throughput, Residual energy and Packet dropped by
varying attributes like the number of nodes, node speed and number of malicious nodes in the
network. Finally, based on a mathematical model it can be concluded that the present CLBNSRM
protocol gives a better result than the compared models in all the aspects. For future work, model
will be evaluated for various types of attacks. This work can be extended with the neuro-fuzzy
technique for confidence calculations.
REFERENCES
[1] S Buchegger, Le Boudec, (2005) Self-policing mobile ad-hoc networks by reputation.IEEECommun.
Mag. 43 (7):1-7.
[2] Sun YL, Han Z, Yu W, Ray LKJ (2006) Attacks on trust valuation in distributed networks. Proc. 40th
Annual Conf. on Information Sciences and Systems, :1461–1466.
[3] Wu B, Wu J, Fernandez EB, Ilyas M, Magliveras S (2007) Secure and efficient key management in
mobile ad hoc networks.J. Netw. Comput. Appl., 30, (3):937–954.
[4] C. E. Perkins and E. M. Royer, (1999) Ad-hoc on-demand distance vector routing. Proceedings
WMCSA'99. Second IEEE Workshop on Mobile Computing Systems and Applications, New
Orleans, LA, USA, pp. 90-100.
[5] Sachin Lalar, (2014) Security in MANET: Vulnerabilities, Attacks & Solutions. International Journal
of Multidisciplinary and Current Research, l (2):62-68.
[6] Mousumi Sardar and Koushik Majumder, (2013) A Survey on Trust Based Secure Routing in
MANET. Computer Science & Information Technology (CS & IT), 243-253.
[7] CH V Raghavendran, G Naga Satish, P Suresh Varma, (2013) Security Challenges and Attacks in
Mobile Ad-Hoc Networks. I.J. Information Engineering and Electronic Business, 3:49-58.
[8] Swaijit Kaushal, Reena Aggarwal, (2015) A study of different types of attacks in MANET and
performance analysis of AODV protocol against wormhole attack. International Journal of Advanced
Research in Computer Engineering & Technology (IJARCET), 4 (2):301-305.
[9] B Harikrishna, N Anusha, G Murali, (2015) Security in Mobile Ad-hoc Networks: Challenges and
Solutions. International Journal of Mechanical Engineering and Information Technology, 3 (1):968-
971.
[10] P Narendra Reddy, C H Vishnuvardhan, V Ramesh, (2013) Routing Attacks in Mobile Ad-hoc
Networks. International Journal of Computer Science and Mobile Computing 2 (5):360-367.
[11] Pirzada A. A, McDonald C, Datta A, (2006) Performance comparison of trust-based reactive routing
protocols.IEEE Trans. Mobile Comput., 5 (6):695–710.
[12] Guo W, Xiong ZW, Li ZT, (2005) Dynamic trust evaluation based routing model for ad hoc
networks. Proc. Wireless Communications, Networking and Mobile Computing, 2:727–730.
[13] H Xia, Z Jia, L Ju, Y Zhu, (2011) Trust management model for mobile ad-hoc network based on
analytic hierarchy process and fuzzy theory. IET Wireless sensor systems., 1(4):248-266.
[14] Martin J, Manickam L, Shanmugavel, S, (2007) Fuzzy based trusted ad hoc on-demand distance
vector routing protocol for MANET. Adv. Comput. Commun. (ADCOM 2007):414–421.

129
[15] Junhai Luo, Xue Liu, Mingyu Fan, (2009) A trust based on fuzzy recommendation for mobile ad-hoc
networks. Computer Networks 53:2396-2407.
[16] Shuaishuai Tan, Xiaoping Li, and Qingkuan Dong, (2016) A trust management system for securing
data plane of ad-hoc networks. IEEE transactions on vehicular technology, 65(9):7579-7591.
[17] Zia Liiah, M Khan, I Ahmed, N Javaid, and M. I. Khan. (2016) Fuzzy-based trust model for detection
of selfish nodes in MANETs. IEEE 30th International Conference on Advanced Information
Networking and Applications. :965-972.
[18] Jerry M Mendel, (1995) Fuzzy logic systems for engineering: a tutorial. Proceedings of the IEEE,
83(3):345–377.
AUTHORS
S. J. Patil - Received the BE degree in Electronics from DKTE’s TEI, Ichalkaranji,
affiliated to Shivaji University, Kolhapur, and an ME degree from the same University,
currently working as an Assistant Professor in the Department of Electronics, DKTE’s
TEI, Ichalkaranji, Shivaji University, Kolhapur. He has more than 12 years of teaching
experience. He is a Ph.D. research scholar of VTU, RRC,Belgavi in the field of
Wireless Networks.
Lalita S. Admuthe - (M’15) member of IEEE, Computer Society. The Author has
received an M.E. and Ph.D. Degree in electronics engineering both from Shivaji
University Kolhapur, India in 1994 and 2013 respectively. Author’s research
interestincludes Neural Networks, Wireless Networks, Fuzzy Logic and Optimization
Problems.
Since 2013 she has been a Professor in Electronics Engineering at DKTE’s Textile and
Engineering Institute Ichalkaranji. Currently, she is working as Dy-Director and Head
of Electronics Engineering Department in the same institute. Her teaching experience
includes the topics of Artificial Neural Networks, Radom Signal Processing, Computer Architecture
and,Parallel Processing.
Meenakshi R. Patil (M-07, SM-17) became member IEEE in 2007 and a senior
member of IEEE in 2017. The Author is graduated in Electronics and
Communication Engineering from PVPIT Budhagaon in 1994 and received post-
graduate degree in Electronics Engineering from WCE Sangli in 2002. The Author
has completed her Ph.D. degree from Shivaji University Kolhapur in 2011. Authors
research interest includes digital watermarking, Digital Image processing,
Communication and, network security.
From 1999 to 2007, she was the head of department BCE Shravanbelgola. Since
2011, she has been a Professor with the AGM Group of institutions Hubbali, Karanataka, India. She is
currently working with JAGMIT Jamkhandi, Karnataka, India. Presently five research students are working
on various fields in Signal processing and communication under her supervision.

CLBNSRM - CONFIDENCE LEVEL BASED UNBLEND NEIGHBOR SELECTION &BLEND NODE REPORT BASED OPTIMIZED ROUTE FORMATION IN MANET

More Related Content

What's hot (17)

Similar to CLBNSRM - CONFIDENCE LEVEL BASED UNBLEND NEIGHBOR SELECTION &BLEND NODE REPORT BASED OPTIMIZED ROUTE FORMATION IN MANET (20)

More from IJCNCJournal (20)

Recently uploaded (20)

CLBNSRM - CONFIDENCE LEVEL BASED UNBLEND NEIGHBOR SELECTION &BLEND NODE REPORT BASED OPTIMIZED ROUTE FORMATION IN MANET