Trust aware cooperative routing method for WANETs

Abstract

Wireless Ad-hoc NETworks are prone to various internal security attacks such as black hole, gray hole, packet modification, etc., because of decentralized and open network operations. Developing soft security systems using trust as the metric has gained significant attention in the security research. These models work in conjunction with routing protocols for mitigating internal security attacks. To this end, this paper presents a behavior-based trust model for cooperative routing in Wireless Ad-hoc NETworks. The proposed trust model has been used with ad-hoc on-demand distance vector (AODV) protocol (hereinafter Behavior based Trust-aware Adhoc On-Demand Distance Vector Routing (BT-AODV)) to identify and isolate malicious nodes from the routing process. The performance of BT-AODV is evaluated against recently proposed trust and energy based-AODV protocol through ns-2 simulations. The simulation results show that BT-AODV is robust in detecting malicious nodes, and the network performance metrics such as packet delivery ratio, routing load, end-to-end delay, and energy consumption have been significantly improved as compared with trust and energy based-AODV protocol. Copyright

Full text

SECURITY AND COMMUNICATION NETWORKS
Security Comm. Networks
(2017)
Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/sec.1765
RESEARCH ARTICLE
Trust aware cooperative routing method for WANETs
P. Raghu Vamsi* and Krishna Kant
Department of Computer Science and Engineering, Jaypee Institute of Information Technology, Noida, India
ABSTRACT
Wireless Ad-hoc NETworks are prone to various internal security attacks such as black hole, gray hole, packet modification,
etc., because of decentralized and open network operations. Developing soft security systems using trust as the metric has
gained significant attention in the security research. These models work in conjunction with routing protocols for mitigating
internal security attacks. To this end, this paper presents a behavior-based trust model for cooperative routing in Wireless
Ad-hoc NETworks. The proposed trust model has been used with ad-hoc on-demand distance vector (AODV) protocol
(hereinafter Behavior based Trust-aware Adhoc On-Demand Distance Vector Routing (BT-AODV)) to identify and isolate
malicious nodes from the routing process. The performance of BT-AODV is evaluated against recently proposed trust and
energy based-AODV protocol through ns-2 simulations. The simulation results show that BT-AODV is robust in detecting
malicious nodes, and the network performance metrics such as packet delivery ratio, routing load, end-to-end delay, and
energy consumption have been significantly improved as compared with trust and energy based-AODV protocol. Copyright
© 2017 John Wiley & Sons, Ltd.
KEYWORDS
Ad-hoc networks; AODV; AOTDV; adaptive weighing; routing security; security attacks; trust models; TE-AODV; WANETs
*Correspondence
P. Raghu Vamsi, Department of Computer Science and Engineering, Jaypee Institute of Information Technology, Noida, India.
E-mail: prvonline@yahoo.co.in
1. INTRODUCTION
Wireless Ad-hoc Networks (WANETs) are the special
class of wireless networks popularly used to establish
ad-hoc communication in the applications like disaster
management, battlefield, etc. [1]. These networks are com-
posed of wireless mobile nodes having limited transmis-
sion range and energy. The topology of the network is
dynamic because of node mobility. Limited transmission
range of nodes restricts them to follow multi-hop com-
munication to route the packets from the source to the
destination. It means that the intermediate nodes between
the source and destination need to be cooperative to
accomplish the routing decisions. However, the limitations
of WANETs such as decentralized network operations,
dynamic topology, openness, and remote deployment can
raise various security vulnerabilities and prone to internal
security attacks such as Blackhole, Gray-hole, packet mod-
ification, etc. [2]. Conventional cryptography methods are
proven efficient in mitigating security attacks posed from
outside of the networks [3–6]. Moreover, using cryptogra-
phy methods, it is very difficult to identify if a valid and
legitimate node misbehavior during the routing process.
In recent years, developing soft security systems using
a human behavior pattern called trust has received con-
siderable attention in the security research to mitigate the
internal attacks. Trust is used to define the degree of belief
about the behavior of an entity [7]. In WANETs, a node
assesses the trust of its neighboring nodes on the basis of
cooperative behavior shown in packet forwarding. When a
node forwards the packet to its neighbor, it observes the
packet forwarding behavior of neighboring node using the
promiscuous use of the network interface. The trust toward
nodes is calculated using these observations. Trust value
of a node increases with respect to the positive behavior
(such as successful packet forwards) and decreases with
respect to negative behavior (such as packet drops or tam-
pering packet integrity). Finally, the calculated trust values
are used in association with routing protocols to bypass the
malicious nodes from routing path establishment. To this
end, the current study presents behavior based trust-aware
cooperative routing method for WANETs. It is an inte-
grated trust model that calculates the Consolidated Trust
Value (CTV) using direct and indirect observations. The
proposed trust model has been incorporated into well-
known ad-hoc on-demand distance vector (AODV) routing
protocol [8]. The features of the trust model are as follows:
Copyright © 2017 John Wiley & Sons, Ltd.

Trust aware cooperative routing method for WANETs P. R. VAMSI AND K. KANT
Adaptive weight assessment to trust metrics.
Calculating direct trust value using node behavior.
Reporting indirect trust values without communica-
tion overhead.
Calculating the indirect trust value by discarding false
recommendation.
CTV calculation using direct and indirect trust values.
Identifying energy efficient and trusted path between
source and destination using the CTV.
The remainder of the paper is organized as follows.
Section 2 presents the related work. Section 3 describes the
network and the adversary model. The behavior-based trust
model for cooperative routing in WANETs is presented in
Section 4. The performance of the proposed trust model
with AODV protocol is evaluated using simulation study in
Section 5. Finally, Section 6 concludes the paper with the
future scope.
2. RELATED WORK
There are various trust models proposed in the literature
for use in association with routing protocols. The concept
of trust is not a new topic; however, it has been used in
various fields such as psychology, sociology, anthropol-
ogy, economics, political science, and computer science
related fields such as e-commerce, social networks, etc.
[9–11] . Trust concepts have received considerable atten-
tion in communication networks to ensure the security in
the network operations such as routing, data aggregation,
and others. Broadly, trust models can be classified into
two categories such as direct and indirect trust models.
In direct trust models, the trust value of a node is cal-
culated solely based on direct observations. That means
trust value is calculated with the subjective assessment of
the node behavior. However, along with the packet for-
warding, a node has to perform several other operations
such as localization, route maintenance, cluster formation,
etc. based on the protocol under use. Hence, a node may
skip some observations while performing intended opera-
tions. In such cases, collecting indirect trust values help in
strengthening the opinion toward a node. For this, a node
can obtain trust information indirectly by collecting trust
opinions of neighboring nodes in a distributed fashion or
by receiving recommendations from trusted third parties in
a centralized or hierarchical fashion. The goal of either of
the models is to calculate consolidated trust values to miti-
gate potential risks such as malicious, dead, or ambiguous
paths. The trust value can be useful to circulate a warning
or alarm message among friend nodes. In case, if the trust
value is very low then the node will be isolated from the
network operations [12,13].
Co-operation of Nodes Fairness In Dynamic Ad-hoc
Networks (CONFIDANT) [14] and A Collaborative Rep-
utation Mechanism (CORE) [15] are the primary works
toward developing trust models for secure ad-hoc rout-
ing. These models are composed of several components
such as watchdog, path rater, trust manager, and reputa-
tion manager for trust derivation, trust computation, and
trust application. In trust derivation, each node collects evi-
dence related to the network activities carried out by its
neighboring nodes. Trust ratings are computed with the
collected evidence. Computed trust ratings are applied in
the routing process to improve the capability of suspect-
ing malicious nodes. The effectiveness of these models has
been evaluated by incorporating them in Dynamic Source
Routing (DSR) Protocol [16]. DSR protocol is composed
of two mechanisms: route discovery and route mainte-
nance. The route from the source to the destination is
identified when it is required. The identified paths may not
be consistent because of the mobility of nodes. Therefore,
route maintenance mechanism is responsible for maintain-
ing and identifying new routes in case of old routes fail.
In this context, the calculated trust value is used in estab-
lishing and maintaining trusted paths. Pirzada et al. [17]
proposed a direct trust model and evaluated its perfor-
mance with reactive protocols such as AODV, DSR, and
Temporally Ordered Routing Algorithm (TORA) [18]. It
is observed from the simulation results that trusted TORA
protocol is robust in routing packets by bypassing the
malicious nodes. Venkatraman et al. [19] proposed vec-
tor autoregression-based trust model. It is a direct trust
model, and it calculates the trust value using time series
analysis. This method addresses multiple attacks such
as packet forwarding, content modification, rushing, and
flooding attacks. This method has been used with AODV
and Optimized Link State Routing (OLSR) [20] protocols.
In [21], the AODV protocol has modified to mitigate the
Black-hole attacks. In this method, each node sets the timer
to buffer route reply messages and analyze the messages
once the timer expires. Node filters the false reply mes-
sages and blocks the nodes from which such messages are
received. It has observed that the packet delivery ratio has
been improved in the presence of Blackhole attacks; how-
ever, the end-to-end delay has been increased because of
the buffering mechanism.
Lie et al. [22] proposed ad-hoc on-demand trusted
path distance vector (AOTDV) protocol. It is a direct-trust
model that calculates the trust values using two trust met-
rics such as Control packets Forwarding Ratio (CFR) and
Data packets Forwarding Ratio (DFR). Each node running
AOTDV keeps track of these two trust metrics. Heuris-
tic weight values are assigned to CFR and DFR. For each
trust update interval, the total trust value is calculated as
the sum of the products of trust metrics (CFR and DFR)
and the corresponding weight values. The best trusted path
is selected as the highest product of the trust values of
nodes between source and destination. This way of path
selection identifies multiple trusted paths between source
and destination. Hence, the AOTDV protocol follows mul-
tipath strategy to route the packets. Recently, Venkanna
et al. [23] proposed trust and energy based ad-hoc on-
demand distance vector (TE-AODV) protocol to improve
AOTDV protocol. In this model, each node calculates Final
Trust Value using direct and indirect trust values. The
trusted path between the source and destination is selected
using Final Trust Value and hop count. Initially, the trust
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

P. R. VAMSI AND K. KANT Trust aware cooperative routing method for WANETs
Ta b l e I . Security features of AOTDV [22],TE-AODV[23],and
BT-AODV(proposed).
Security
featu re
Trust model
AOTDV TE-AODV BT-AODV
Black-hole attack Yes Yes Yes
Gray-hole attack Yes Yes Yes
Modification attack Yes No Yes
Bad-mouth attack No No Yes
Lifetime consideration No Yes Yes
False-energy attack No No Yes
Selfish behavior No Yes Yes
On-Off attacks No No Yes
AOTDV, ad-hoc on-demand trusted path distance vector; TE-AODV, trust
and energy based ad-hoc on-demand distance vector.
value of each node is set to 0.5, and the value will be
updated for every fixed trust update interval. Trust value is
scaled in [0,1]. The direct trust value is calculated based on
packet forwarding behavior of nodes. Indirect trust opin-
ions are requested when a nodes’ trust value is less than
0.5. This model considers two potential security attacks
such as Black-hole and Gray-hole attacks. However, broad-
casting the indirect trust request (TRREQ messages) and
receiving the recommendations from neighboring nodes
(TRRES message) increases the communication overhead.
Further, a bad-mouth attacker may provide false recom-
mendations for benign nodes to pollute the indirect trust
values.
A Heuristic Approach based Trust Worthy Architec-
ture for Wireless Sensor Networks (WSN) is proposed in
[24]. Heuristic Approach based Trust Worthy Architecture
considers the challenges of the trust system and focuses
on the collaborative mechanism for trust evaluation and
maintenance. This architecture is capable of fulfilling secu-
rity, reliability, mobility, and performance requirements
for reliable communication while being readily adaptable
to different applications. Further, trust models are also
developed to guard geographic routing protocols [25–28].
It is observed from the literature that each trust model
has its own advantages and limitations. It means the trust
model that addresses multiple attacks is limited. To this
end, the current study presents an integrated trust model to
address multiple attacks. The trust model has been incor-
porated in AODV protocol. It is named as BT-AODV
protocol. The security features of AOTDV [22], TE-AODV
[23] and the proposed BT-AODV protocols are given in
Table I.
3. NETWORK MODEL, ADVERSARY
MODEL, AND ASSUMPTIONS
3.1. Network model and assumptions
An ad-hoc network is considered in which the network
consists of mobile nodes with fixed transmission range.
Each node communicates among them only if they are in
communication range of each other. Each node periodi-
cally broadcast a hello message consisting of node identity,
remaining energy value, and the sequence number of hello
packets. The information provided in the hello packets are
used to maintain neighbor table. Each node keeps track
of the fulfillment of routing activities by their neighboring
nodes via the promiscuous use of the network interface.
Further, each node calculates the consolidated trust value
using direct and indirect observations. The trust value of
each node ranges in [0,1]. Initially, each node is assigned
the trust value of 0.5. Nodes having trust value greater than
or equal to 0.5 are regarded as benign nodes and less than
0.5 are considered as malicious nodes.
3.2. Adversary model
It is assumed that there are no malicious nodes during the
initial stages of the network operations. However, adver-
sary activities start as the network operations progress. The
most fundamental and active security attacks on routing
are considered for the study. It is because ignoring them
can lead to very powerful attacks such as wormhole attacks
and show an adverse impact on the network performance
metrics. Security attacks are [2]
Black-hole attack : A malicious node creates an
impression as the next node to forward packets. When
it receives the packet, it drops.
Packet-modification attack: A malicious node modi-
fies the packet integrity by tampering its unique code
or hash code so that a receiving node discards the
packet as invalid.
Gray-hole attack: It is a variant of Black-hole attack in
which a malicious node selectively drops the packets
and/or tampers the packet integrity.
Selfish-behavior attack: It is a kind of non-cooperative
behavior. A selfish node does not show interest to par-
ticipate in routing process to save the resources such
as energy and bandwidth.
Bad-mouth attack: It is a severe threat to the repu-
tation system. A bad-mouth attacker provides false
recommendations to damage well-behaving node’s
reputation by continuously advertising poor trust
value.
False-energy attack: A malicious node reports false
energy information to mislead a benign node from
choosing an energy efficient path.
On-Off attack: A malicious entity behaves good and
bad alternatively to remain undetected while causing
damage to the network.
4. TRUST-AWARE COOPERATIVE
ROUTING METHOD
4.1. Trust metrics selection
The trust metric is a parameter with which security attacks
are identified. The cooperative behavior of nodes is a
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

Trust aware cooperative routing method for WANETs P. R. VAMSI AND K. KANT
vital factor during the routing process. This behavior is
assessed by careful observation of routing activities ful-
filled by nodes. In order to detect routing attacks mentioned
in Section 3.2, appropriate trust metrics has to be defined.
As described, Black-hole, Selfish, and On-off attacks can
be detected with the metrics such as sincerity in packet for-
warding and providing network acknowledgment. It means
with these metrics, dropping of packets can be identified.
In addition to packet dropping, gray-hole, and packet mod-
ification attacks can be identified by observing the trust
metric such as sincerity in maintaining packet integrity.
Further, energy information is an important parameter
when the trust model considers the lifetime of nodes. Peri-
odic validation of energy information provided by nodes
helps in avoiding dead paths, energy holes, and Selfish
nodes. When a trust model receives recommendations (or
indirect trust values) from its neighbors, it is very impor-
tant to check the validity of recommended trust values.
Utilizing the indirect trust values by discarding false rec-
ommendations helps in identifying bad-mouth attacks. The
energy information verification and recommendation trust
verification has been presented in the Sections 4.2 and 4.3,
respectively. To summarize, the list of trust metrics con-
sidered for the study are sincerity in packet forwarding
(m1), maintaining packet integrity (m2), network acknowl-
edgments (m3), energy information (m4) and recommenda-
tions (m5). The values of these trust metrics are initialized
to 1. Direct trust calculation using these trust metrics is
provided in the next section.
4.2. Direct trust calculation
Consider a routing scenario shown in Figure 1. Let S, A,
B, C, D, E, and F be the mobile nodes in which S is the
source and D is the destination. The thick edge between
nodes represents the trusted path and the dashed line rep-
resents a malicious path. When S has to send a packet to
D, it chooses C as the next node to forward because next
hop to C is D. When a route is available, S stores the copy
of the forwarding packet, its corresponding sequence num-
ber, and time-stamp in its packet buffer. When A receives
the packet from S, it first checks for packet integrity. If
the integrity check is successful then node forwards further
otherwise, it drops the packet as invalid. When A forwards
the packet further, S passively listens the packet (because
S is in the transmission range of A) and updates the ful-
fillment of trust metrics m1,m2and m3. To do this, each
node maintains success (Scount) and failure (Fcount ) coun-
ters for each trust metric. When a node fulfilled a trust
metric, then its corresponding Scount is incremented, oth-
erwise corresponding Fcount is incremented. Along with
these metrics, each node periodically verifies the valid-
ity of energy information received in the hello packets.
Let ETx,ETr,E0,Ec, and Pcbe the energy consumed for
packet transmission, energy consumed for packet recep-
tion, initial energy, energy consumption information, and
packet count (including control and data packets), respec-
tively, then the inconsistencies in the energy information is
identified as follows.
|E0–Ec|<Pc*(ETx+ETr) (1)
Node records energy information reported by a neigh-
boring node as valid when the Equation (1) is satisfied
and then increments m4corresponding Scount, otherwise
its Fcount will be incremented. Similarly, each node checks
Figure 1. Routing scenario. RREP, route reply; RREQ, route request.
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

P. R. VAMSI AND K. KANT Trust aware cooperative routing method for WANETs
for validity of recommendation trust values (presented in
Section 4.3) and accordingly corresponding counter values
of m5are incremented. In this way, each node updates the
counters when they passively observe the forwarded pack-
ets. The direct trust value is calculated for every fixed time
interval (also called trust update interval) using these coun-
ters. The Direct Trust (DT) value calculation consists of
two components: (i) weight assessment; and (ii) expecta-
tion calculation. During weight assessment, a weight value
will be assigned to each trust metric based on Scount value
as follows:
W(mi)= Scount(mi)
P5
i=1 Scount(mi)(2)
Where, W(mi) is the weight to be assigned to the trust
metric i. With Equation (2), a trust metric having high-
est success count will be given more weight as opposed
to heuristic weight assignment. In this way, weight assess-
ment provides the priority of a trust metric over remaining
trust metrics. When multiple attacks are present in the
network, each malicious node will be having independent
attack profile. In such cases, the adaptive weighing men-
tioned in Equation (2) helps in improving the accuracy of
trust calculation instead of assigning a fixed weight value
to the trust metrics. To this end, it is also important to
assess the expected behavior of a node with respect to a
particular trust metric. To this end, the expected behavior is
calculated using Beta expectation function [29] as follows:
E(mi)= Scount(mi)+1
Scount(mi)+Fcount (mi)+2 (3)
Where, E(mi) is the expectation value. Because weight
value is the priority of trust metric over remaining met-
rics, and expectation is the behavior assessment pertaining
to a trust metric, combining these two values will provide
more effective trust value. So, using weight and expec-
tation values, the DT value of a node jis calculated as
follows:
DT(j)=
5
X
i=1
W(mi)*E(mi) (4)
Where, DT(j) is the direct trust value. Because the
weight and expectation values remain in [0,1], the DT
value also remains in [0,1].
4.3. Indirect trust calculation
The direct trust value is sufficient to assess node behav-
ior. However, during mobility, a node may discover new
nodes, and its old neighbors may disappear. In such cases,
assessing the behavior of new nodes may become diffi-
cult. In some cases, obtaining the second opinion about a
node having less trust value can also help in assessing exact
trust value. Hence, obtaining trust recommendations from
neighboring nodes and combining them with the DT value
helps in improving the quality of routing decisions. In the
existing trust models, the recommendations are obtained in
two ways: (i) using recommendation request (TRREQ) and
response messages (TRRES); (ii) periodic broadcasting
of recommendation values. However, these two methods
increase the congestion in the network and hence lead to
packet loss. Therefore, a lightweight method is required
to report the indirect trust values. To this end, the current
study presents a lightweight method to report the recom-
mendations by piggybacking the recommendations along
with outgoing data packets. At first, the calculated DT
value of node jis rounded to an integer value as shown
below:
DRR(j)=dDT(j)*10e(5)
Where, DTR(j) is the rounded integer value of DT(j).
For example, if DT(j) is 0.68 then the DT R(j) becomes
7, to store or communicate this value, 4 bits are suffi-
cient (because 4 bits supports from 0 to 15). The energy
consumption of a node depends on the number of bits
it transmits or receives. Therefore, this procedure signif-
icantly reduces the communication overhead and energy
consumption. When a node has to forward the data packet,
it prepares a list of nodes having trust value greater than
the average trust value of its neighboring nodes and pig-
gybacks the node identity and corresponding DTRvalues
along with the data packets. Let n1,n2:::nmbe the neigh-
boring nodes of a node jthen the average trust value
(DTavg) is calculated as
DTavg =Pm
k=1 DT(nk)
m(6)
Because each node works with the promiscuous use of
the network interface, it is easy to obtain the indirect trust
values and update them by combining with DT values.
It is also possible that a packet will route through bad-
mouthing attackers. In such cases, the bad-mouth node can
recommend false trust values. The quality of routing deci-
sions can degrade if such recommendations are considered
without validation. To this end, a simple validation method
is presented in the current study. Consider that node j
received the recommendations about one of its neighbor-
ing node kfrom node i, then jcalculates the recommended
trust RT(k) as follows
RT(k)=
DT(k)+DT (ji)* DT R(ik)
10 
1+DT(ji)(7)
Where, DT(k) is the DT value of node k,DT (ji)isthe
DT of jon i,DTR(ik) is the recommended DT value of
ion k. The reported value will be considered when the
following condition holds
|DT(k)–RT (k)| Rth (8)
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

Trust aware cooperative routing method for WANETs P. R. VAMSI AND K. KANT
Where, Rth is the heuristic recommendation threshold
value. Recall that the trust metric m5is related to sincerity
in providing recommendations, when the Equation (8) is
not satisfied, then the recommended value will be dis-
carded and node jincrements m5corresponding Fcount
value for node i. Otherwise, m5corresponding Scount
value will be incremented. Further, recommended value
is combined with DT value to calculate consolidated trust
value.
4.4. Consolidated trust value calculation
Once node validates the recommendation value, it is com-
bined with DT value to calculate CTV as follows:
CTV(k)=CF (k)*DT(k)+(1–CF(k))*RT(k)) (9)
Where, CTV(k), DT (k), and RT (k), respectively, are the
consolidated, direct, and recommended trust values of node
k.CF(k) is the confidence factor on node k. In the current
study, the CF value is inspired from the human trust model
in which confidence on a person improves with respect
to the number of positive interactions and degrades with
respect to the number of negative interactions. Each node
calculates the CF value using the total success count of the
trust metrics under consideration as follows
TScount(k)=
5
X
i=1
Scount(mi) (10)
Where, TScount(k) is the total success count of node k.
Using TScount value the CF value is calculated as follows:
CF(k)= TScount (k)
TScount(k)+1 (11)
It can be observed from Equation (10) that the CF value
increases with respect to increase in the total success count.
Because CF and DT values remains in [0,1], the CTV value
will also remain in [0,1]. The CTV is used in trusted path
selection as described in the next section.
4.5. Trusted path selection
Routing is the process of discovering route from the source
to the destination. The CTV is used with routing proto-
col to identify the trusted path for making efficient routing
decisions by isolating malicious nodes. To this end, the
proposed trust model is used with well-known AODV pro-
tocol. The AODV protocol is a reactive routing protocol
that discovers the route when it is required. In AODV pro-
tocol, the process of route establishment will be carried
out using route request (RREQ) and route reply (RREP)
messages. A node broadcasts RREQ message when it has
to send the packet and the route to the destination is
not available. The RREQ message consists of fields such
as <broadcast ID, source address, source sequence num-
ber, destination address, destination sequence number, hop
count>. Any node receives RREQ message searches in
their routing table for destination address. If route to the
destination is available then RREP message will be sent
to the source node. Otherwise, node rebroadcasts RREQ
packet by increasing the hop count. In this way, the RREQ
message will be flooded in the network till it reaches the
destination. The RREP message consists of fields such
as <source address, source sequence number, destination
address, destination sequence number, hop count, life-
time>. However, when a node receives multiple RREP
message for a destination, it updates the route in its routing
table having lowest hop count.
The task of trust model is to calculate the CTV based
on node’s packet forwarding behavior and use it along
with the route discovery process. Two new variables called
total_trust and avg_trust are introduced in RREQ and
RREP messages to record the trust values of intermediate
nodes between source and destination. While flooding of
RREQ message, the total_trust and avg_trust variables are
set to 0 by each node. When the destination node receives
RREQ message, it replies RREP message by initializing
the total_trust and avg_trust variable to 1. All nodes in
the reverse path from destination to source add the CTV
and remaining energy information of the replying node
to total_trust in RREP message and forward further. Let
i=1,2,..mbe the nodes in the reverse path, then total_trust
and avg_trust are calculated as follows:
total–trust+=
m
X
i=1
CTV(i)+energy(i) (12)
avg–trust =total–trust
m(13)
When a source node receives RREP messages, it
updates the route in the routing table having highest
avg_trust value in the routing table. In this way, trusted
path will be selected from the source to the destination.
Along with route discovery, AODV protocol performs
route maintenance. It is initiated when route time is expired
or link between two nodes is broken because of mobility.
Nodes running the proposed trust model calculate the CTV
of neighbors and label the nodes having CTV less than 0.5
as malicious in their neighbor table. During the route main-
tenance process, routes having such malicious nodes are
identified and purged. A route error message will be sent
out to initiate new route discovery between the source and
destination. In this way, an energy efficient and trusted path
will be maintained using route discovery and maintenance
mechanisms.
5. SIMULATION STUDY
The network simulator ns-2 [30] has been used to evalu-
ate the performance of AODV [8], TE-AODV [23], and
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

P. R. VAMSI AND K. KANT Trust aware cooperative routing method for WANETs
BT-AODV (proposed) protocols. The following network
performance metrics are considered for the evaluation.
Packet delivery fraction: It is the ratio of the num-
ber of data packets received by the destination to the
number of data packets sent by the source nodes. This
metric attributes the dependability of the trust model.
Routing load: It is the ratio of the number of control
packets to the number of data packet generated in the
network.
End-to-end delay: It is defined as the time taken for a
packet to reach the destination from its source.
Energy consumption: It is the energy consumed by
the nodes in the network in performing routing oper-
ations.
Hop count: It is the number of hops traveled by the
data packet between source and destination.
Throughput: It is the average number of bits per sec-
ond routed in the network. This metric shows the
efficiency of the trust model.
5.1. Simulation setup
Table II shows the important simulation parameters con-
Table II. Simulation parameters.
Simulator ns-2.35 [30]
Examined Protocols AODV [8],TE-AODV[23],
and BT-AODV
Mac 802.11 DCF
Simulation Time 600 s
Area 1kmx1km
Nodes 50
Propagation model Two Ray Ground reflection
Transmission range 250 m
Initial energy 10 Joules
Mobility model Random way point
Maximum speed 5, 10, 15, and 20 m/s
Traffic type CBR over UDP
Maximum connections 15
Packet size 64 bytes
Packet rate (drate) 4 packets/ond
Maximum malicious 50% of total nodes
nodes (i.e., 25 nodes)
Type of attacks Security attacks addressed
in Section 3.2.
Trust update interval 0.02 s
Rth 0.15
AODV, ad-hoc on-demand distance vector; CBR, Constant Bit Rate; DCF,
Distributed Coordinate Function TE-AODV, trust and energy based ad-hoc
on-demand distance vector; UDP, Universal Datagram Protocol.
sidered for the study. Two scenarios are considered for the
evaluation by considering acceptable mobility speed (i.e.,
16–20 m/s in real environments [31]). The mobility scenar-
ios are generated using setdest utility available in the ns-2
simulation tool. In scenario 1, the performance of protocols
is evaluated in the presence of 50% of malicious nodes and
with varying node mobility from 0 to 20 m/s with a step
size of 5 m/s. In scenario 2, the performance of the pro-
tocols is evaluated with a fixed mobility speed of 20 m/s
and with varying percentage of malicious nodes from 0%
to 50% of the total nodes with a step size of 10%. All the
security attacks addressed in Section 3.2 are considered for
the study. The attacking nodes are chosen randomly. Each
simulation scenario with all possible cases are simulated
on 50 random graphs, and the average of the data obtained
from simulation runs is presented in the result analysis.
5.2. Result analysis
5.2.1. Scenario 1: 50% of malicious nodes with
varying node mobility.
Figure 2(a)–(f) plots the node mobility speed ver-
sus performance metrics for AODV, TE-AODV, and BT-
AODV protocols in the presence of 50% of malicious
nodes in the network. The description of each performance
metric is provided in the sequel.
Figure 2(a) plots the packet delivery fraction (PDF). It
can be seen from the graph that PDF of BT-AODV remains
high across various mobility rates. It is because nodes run-
ning BT-AODV protocol establish trustworthy path using
CTV and remaining energy of nodes. The TE-AODV pro-
tocol is not resistant to multiple attacks because it does not
consider the bad-mouth attack, on-off attack, and verifica-
tion of false energy information which disrupts choosing
the best path. Because AODV protocol is trust unaware,
it has resulted into low PDF. Hence, BT-AODV results
into high PDF as compared with BT-AODV and AODV
protocols.
Figure 2(b) shows the routing load. It can be observed
from the graph that routing load is high in TE-AODV
as compared with BT-AODV and AODV protocols. It is
because during the route maintenance phase, if a node
observes malicious node in the routing table, then it purges
the corresponding route and initiates the route discovery
process. This procedure results in the generation of control
packets. Hence, the control packets count in the network
depends on the number of malicious nodes identified in
the path. In addition, TE-AODV protocol sends out special
request and response message to know about secondary
trust values. To this end, BT-AODV protocol is placing
efforts to identify the best path between source and des-
tination when there is a maximum number of malicious
nodes in the network. Hence, the routing load of BT-AODV
protocol is low as compared with TE-AODV protocol.
Figure 2(c) shows the end-to-end delay occurred in
packet delivery. The end-to-end delay increases because
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

Trust aware cooperative routing method for WANETs P. R. VAMSI AND K. KANT
Figure 2. Scenario 1: Performance metrics in the presence of 50% of malicious nodes and with varying node mobility.(a) Packet
delivery fraction, (b) Routing load, (c) End-to-end delay, (d) Energy consumption, (e) Hop count, and (f) Throughput. AODV, ad-hoc
on-demand distance vector; TE-AODV, trust and energy based ad-hoc on-demand distance vector.
of the presence of selfish nodes and false information
providers. It is because such malicious activities disrupt
the route discovery process by not responding to con-
trol packets and misguiding the path discovery through
energy holes. Nodes running BT-AODV protocol elimi-
nates choosing malicious path using CTV and validated
remaining energy information. BT-AODV protocol has
the provision of filtering false recommendations and false
energy information provided by the malicious nodes.
Because such mechanism does not exist in TE-AODV pro-
tocol, it resulted in high end-to-end delay as compared
with BT-AODV protocol. However, TE-AODV protocol
can recognize black hole, gray hole, and selfish attacks;
it can bypass such attackers from the routing path as
compared with AODV protocol. Hence, the BT-AODV
protocol results into low end-to-end delay as compared
with TE-AODV and AODV protocols.
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

P. R. VAMSI AND K. KANT Trust aware cooperative routing method for WANETs
Figure 2(d) shows the network energy consumption. It
can be seen from the graph that the energy consumption is
low in BT-AODV protocol across varying node mobility.
BT-AODV protocol utilizes piggybacking method to report
recommendation values. Further, direct trust values are
rounded to the nearest integer values (using Equation (5))
to reduce the energy and communication overheads. Along
with this, BT-AODV protocol can identify the false energy
information provided by the malicious nodes. Because
of these features, BT-AODV protocol is able to identify
the energy efficient and trusted path between source and
destination. TE-AODV protocol considers node energy
information, but it has no provision of recognizing false
energy information. Because AODV protocol is trust and
energy unaware, the energy consumption is more across
varying node mobility. Hence, BT-AODV protocol results
into low energy consumption as compared with TE-AODV
and AODV protocols.
Figure 2(e) plots the hop count. The ability to identify
the best path reflects the average number of hops traveled
by the packets during the routing process. As explained
before, and also observed from previous graphs. BT-
AODV protocol has the ability to identify multiple attacks.
It forwards packets via energy efficient path by isolat-
ing malicious nodes. This feature makes the packet travel
through additional paths. Hence, it results in a marginal
increase in hop count as compared with TE-AODV and
AODV protocols.
Figure 2(f) plots the network throughput. It can be
observed from the graph that the network throughput
depends on the PDF of the protocol under use. It can be
observed from Figure 2(a) that the PDF of BT-AODV pro-
tocol is high when compared with TE-AODV and AODV
protocol. Hence, high throughput can be observed with BT-
AODV protocol as compared with TE-AODV and AODV
protocols.
5.2.2. Scenario 2: 20 m/s node mobility with
varying percentage of malicious nodes.
Figure 3(a)–(f) plots the percentage of malicious nodes
versus performance metrics for AODV, TE-AODV, and
BT-AODV protocols in 20 m/s node mobility in the net-
work. The number of malicious nodes is varied in between
0% and 50% of the total nodes (i.e., 0–25 nodes) with a
step size of 10%. The simulation has set in a way that
the malicious nodes appear in an incremental fashion. The
malicious nodes appear in the network from 60 s of the
simulation time. From 0% malicious nodes, this percent-
age increases by 10% every 15 s. In this way, the malicious
nodes population grows up to 50% of the total nodes. This
simulation setting is used to test the dynamic behavior the
proposed trust model. The description of each performance
metric is provided in the sequel.
Figure 3(a) plots the packet delivery fraction. It is appar-
ent from the figure that the proposed BT-AODV protocol
has shown high packet delivery across varying percentage
of malicious nodes. Direct trust calculation with adaptive
weight and expectation assessment with energy awareness
resulted in the identification of malicious nodes dynami-
cally by BT-AODV protocol. Any change in node behavior
can be identified using adaptive weights and confidence
factor. Further, the proposed trust model dynamically cal-
culates the CTV using direct trust value, indirect trust
value, and confidence factor. It results into choosing the
paths dynamically by excluding malicious nodes dur-
ing the route discovery process. AODV protocol has no
method of trust calculation or bypassing the malicious
nodes, and hence, it resulted into low PDF. Whereas, in
TE-AODV protocol, the path selection is circumvented
by fake information providers. It results into choosing the
wrong paths and hence leads to low PDF as compared with
BT-AODV protocol.
It can be observed from Figure 3(b) that routing load
of TE-AODV protocol keeps increasing with respect to
increasing the percentage of malicious nodes. It is because
of special reputation request and response messages. When
malicious nodes are identified in the existing paths, TE-
AODV protocol purges such paths and initiates a new route
discovery process. Route discovery involves an exchange
of control packets (such as RREQ and RRES messages)
and thereby results in an increase in routing load. With
this, it can be noted that the number of route discov-
ery initiations increases with respect to identification of
malicious nodes in already established paths. BT-AODV
protocol reports the secondary trust opinions via piggy-
backing. Hence, such consideration results to less number
of control packets generation and thereby results in low
routing load as compared with TE-AODV protocol. AODV
protocol considers each node as trustworthy (irrespec-
tive of malicious nodes) and establishes the path. Hence,
it is completely unaware of malicious activities thereby
results into general route discovery initiations. Hence, BT-
AODV protocol incurred low routing load as compared
with TE-AODV and AODV protocols.
Figure 3(c) plots the end-to-end delay. It can be
observed from the graph that AODV protocol has a high
end-to-end delay. It is because of packet dropping by mali-
cious nodes in the network. The packet retransmissions
increase with respect to the number of packet drops and
hence increase in end-to-end delay. In TE-AODV pro-
tocol, in addition to general route discovery and route
maintenance phases, it requests for secondary trust opinion
about nodes having trust value below 0.5 by sending spe-
cial control messages called reputation request (TRREQ)
and reputation response (TRRES) messages. Each node
after sending TRREQ message has to wait for some
amount of time to receive TRRES messages. In addition,
because TE-AODV protocol has no provision of verify-
ing false recommendations, it is easy for an adversary to
appear in the routing path. The procedure of sending out
special control messages for obtaining reputation values,
waiting for reputation responses and waiting for the best-
trusted path (i.e., path with high final trust value) increases
the end-to-end delay as compared with BT-AODV proto-
col. However, BT-AODV protocol conveys the secondary
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

Trust aware cooperative routing method for WANETs P. R. VAMSI AND K. KANT
Figure 3. Scenario 2: Performance metrics with varying percentage of malicious nodes. (a) Packet delivery fraction, (b) Routing load,
(c) End-to-end delay, (d) Energy consumption, (e) Hop count, and (f) Throughput. AODV, ad-hoc on-demand distance vector; TE-AODV,
trust and energy based ad-hoc on-demand distance vector.
trust opinions by piggybacking them with outgoing data
packets. Because each node establishes path by identi-
fying benign nodes, the secondary opinions provided by
each node in the path can be considered as trustworthy.
Hence, there will be less chance of occurring bad-mouth
attacks. However, if any such attackers are present, then the
inconsistency check (using Equation (8)) can easily fil-
ter the false recommendations. Hence, BT-AODV protocol
results into low end-to-end delay as compared with TE-
AODV and AODV protocols.
Figure 3(d) shows the network energy consumption.
The energy consumption is directly proportional to the
number of packets sent and received. It can be seen from
the graph that AODV protocol has high-energy consump-
tion as compared with TE-AODV and BT-AODV protocol.
It is due to the number of packet retransmissions occurred
because of packet drops by malicious nodes in the net-
work. Although TE-AODV protocol considers the remain-
ing energy information of nodes, because of the additional
control messages sent for obtaining secondary trust values
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

P. R. VAMSI AND K. KANT Trust aware cooperative routing method for WANETs
(TRREQ and TRRES messages), the energy consumption
is high as compared with BT-AODV protocol. As the num-
ber of malicious nodes increases in the network, a node ini-
tiates more secondary trust value requests and hence results
in high-energy consumption. Whereas, in BT-AODV pro-
tocol, the secondary trust values of nodes having the trust
value above the threshold are sent by piggybacking along
the data packets once for every two trust update intervals.
Because each node operates in promiscuous mode, they
can easily validate and calculate the CTV using confidence
factor. This procedure reduces the energy consumption.
It can be observed from the figure that energy consump-
tion of BT-AODV protocol keeps decreasing until 40%
of malicious nodes. When 50% of malicious nodes are
present in the network, BT-AODV protocol has recorded a
hike in energy consumption because when there is a max-
imum number of malicious nodes BT-AODV protocol is
putting best efforts to identify trusted path. In search of
such trusted path, an additional amount of energy is con-
sumed in the network. However, as compared with AODV
and TE-AODV protocols, BT-AODV protocol has resulted
in low energy consumption.
Figure 3(e) plots the average number of hops traveled by
the packets during the routing process. It can be observed
from the graph that AODV protocol has high hop count
because of the absence of trust calculations. It is also due
to the reason that the attackers are disrupting the route
establishment phase by making the protocol choose the
longest path. In TE-AODV protocol, trusted path selec-
tion with final trust values and hop count helps in reducing
the number of hops traveled by the packets. However, BT-
AODV protocol is robust in recognizing multiple attacks;
it searches for energy efficient and trusted path between
source and destination for the best of effort delivery. The
route length is increased in order to establish such energy
efficient paths. Hence, it results in a marginal increase in
high hop count in BT-AODV protocol as compared with
TE-AODV protocol.
Figure 3(f) plots the network throughput. It can be seen
from the figure that high throughput is recorded in the net-
work when BT-AODV protocol is employed. This is due
to high packet delivery fraction by BT-AODV protocol.
The throughput is proportional to the number of packets
routed in the network. That means BT-AODV protocol is
efficient in detecting and isolating malicious nodes from
the routing path. With this, it can be concluded that net-
work running BT-AODV protocol has resulted into high
throughput across varying percentage of malicious nodes
as compared with TE-AODV and AODV protocols.
5.3. Result summary
The dependability and efficiency of any trust model is
assessed based on its performance in the presence of a
maximum number of malicious nodes present in the net-
work. Table III shows the performance metrics such as
packet delivery fraction, end-to-end delay, energy con-
sumption, and throughput of AODV, TE-AODV, and BT-
Table III. Performance comparison of AODV, TE-AODV, and
BT-AODV protocols.
Trust model
Metrics AODV TE-AODV BT-AODV
Packet delivery frac-
tion
0.519 0.636 0.761
Routing load 4.71 6.73 5.2
End-to-end delay (sec) 0.07 0.05 0.03
Energy consumption
(Joules)
19.61 19.26 18.64
Hop count 1.82 1.82 1.87
Throughput (Kbps) 13.38 16.41 19.63
AODV, ad-hoc on-demand distance vector; TE-AODV, trust and energy
based ad-hoc on-demand distance vector.
AODV protocol in the presence of as maximum of 50% of
malicious nodes in the network in a mobile network with
20 m/s maximum node mobility.
It can be interpreted from Table III that the packet
delivery fraction of BT-AODV protocol has increased by
46.6% as compared with AODV protocol and 19.6% as
compared with TE-AODV protocol. The routing load of
BT-AODV protocol has recorded low as compared with
TE-AODV protocol. It is due to the dynamic identifica-
tion of malicious nodes. The process of purging malicious
paths and special messages for obtaining secondary trust
values increased the routing load in TE-AODV protocol.
However, the end-to-end delay has recorded a substan-
tial improvement when BT-AODV protocol is employed.
It can be observed that end-to-end delay has improved
by 66% as compared with TE-AODV protocol. In the
same way, BT-AODV protocol delivered the packet with
low energy consumption as compared with TE-AODV
protocol. This is due to the feature of piggybacking the
secondary trust values in BT-AODV protocol. In search
of trusted and energy efficient path, the number of hops
traveled by the packets has recorded a marginal improve-
ment in BT-AODV protocol as compared with TE-AODV
protocol.
Finally, the BT-AODV protocol can be termed as effi-
cient because it has recorded high throughput as compared
with TE-AODV and AODV protocols. The throughput has
increased by 19.6% (in accordance with PDF) as compared
with TE-AODV protocol. However, marginal improvement
in routing load and hop count can be admitted because
of the accurate elimination of malicious nodes from rout-
ing path and care taken during the selection of the routing
path. Hence, it can be concluded that BT-AODV protocol is
more dependable and efficient for establishing cooperative
routing in WANETs.
6. CONCLUSION AND FUTURE
WORK
In this paper, behavior-based trust model for cooperative
routing in the wireless ad-hoc network has been proposed.
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

Trust aware cooperative routing method for WANETs P. R. VAMSI AND K. KANT
The proposed method has been integrated with the well-
known AODV protocol. It is said to be BT-AODV proto-
col. BT-AODV protocol calculates the Consolidated Trust
Value (CTV) of nodes by combining Direct Trust (DT) and
Indirect Trust values. The features of BT-AODV protocol
such as adaptive weight assignment and expectation calcu-
lation for each trust metric enabled it to detect and isolate
multiple attacks during the routing process. The CTV is
used to identify trustworthy nodes to establish trusted
and energy efficient path for cooperative data forwarding
between source and destination. A simulation study using
the network simulator NS-2 has been conducted to analyze
the performance of the BT-AODV protocol. BT-AODV
protocol has been compared against the recently proposed
TE-AODV protocol and standard AODV protocol. It has
been observed that using BT-AODV protocol important
network performance metrics such as packet delivery frac-
tion, end-to-end delay, energy consumption, and network
throughput has recorded a significant improvement as com-
pared with TE-AODV and AODV protocols. As a future
work, the proposed trust model will be extended to detect
malicious activities on routing such as wormhole and
Sybil attacks.
REFERENCES
1. Kiess W, Mauve M. A survey on real-world implemen-
tations of mobile ad-hoc networks. Ad Hoc Networks
2007; 5(3): 324–339.
2. Kannhavong B, Nakayama H, Nemoto Y, Kato N,
Jamalipour A. A survey of routing attacks in mobile ad
hoc networks. IEEE Wireless communications 2007;
14(5): 85–91.
3. Sanzgiri K, Dahill B, Levine BN, Shields C, Belding-
Royer EM. A secure routing protocol for ad hoc
networks. In Proceedings. 10th IEEE International
Conference on Network Protocols, 2002, IEEE, Paris,
France, 2002; 78–87.
4. Zapata MG. Secure ad hoc on-demand distance vec-
tor routing. ACM SIGMOBILE Mobile Computing and
Communications Review 2002; 6(3): 106–107.
5. Li Q, Hu Y-C, Zhao M, Perrig A, Walker J, Trappe
W. Sear: a secure efficient ad hoc on demand routing
protocol for wireless networks. In Proceedings of the
2008 ACM Symposium on Information, Computer and
Communications Security, ACM, Tokyo, Japan, 2008;
201–204.
6. Cerri D, Ghioni A. Securing AODV: the A-SAODV
secure routing prototype. IEEE Communications Mag-
azine 2008; 46(2): 120–125.
7. Capra L. Engineering human trust in mobile system
collaborations. ACM SIGSOFT Software Engineer-
ing Notes, Vol. 29, ACM, New York, USA, 2004;
107–116.
8. Perkins C, Belding-Royer E, Das S. Ad hoc on-
demand distance vector (aodv) routing. Technical
Report, IETF (Internet Engineering Task Force) RFC
3561, 2003.
9. Jøsang A, Ismail R, Boyd C. A survey of trust and rep-
utation systems for online service provision. Decision
support systems 2007; 43(2): 618–644.
10. Vamsi PR, Kant K. Systematic design of trust manage-
ment systems for wireless sensor networks: a review.
In Fourth International Conference on Advanced Com-
puting & Communication Technologies (ACCT), IEEE,
Rohtak, India, 2014; 208–215.
11. Govindan K, Mohapatra P. Trust computations and
trust dynamics in mobile adhoc networks: a survey.
IEEE Communications Surveys & Tutorials, 2012; 14
(2): 279–298.
12. Mejia M, Pena N, Munoz JL, Esparza O. A review of
trust modeling in ad hoc networks. Internet Research
2009; 19(1): 88–104.
13. Marti S, Giuli TJ, Lai K, Baker M. Mitigating routing
misbehavior in mobile ad hoc networks. In Proceed-
ings of the 6th Annual International Conference on
Mobile Computing and Networking, ACM, Boston,
MA, USA, 2000; 255–265.
14. Buchegger S, Le Boudec J-Y. Performance analysis
of the confidant protocol. Proceedings of the 3rd acm
international symposium on mobile ad hoc networking
& computing, Lausanne, Switzerland, 2002; 226–236.
15. Michiardi P, Molva R. Core: A Collaborative Rep-
utation Mechanism to Enforce Node Cooperation in
Mobile Ad Hoc Networks. In Advanced Communi-
cations and Multimedia Security. Springer: Portoroz,
Solvenia, 2002; 107–121.
16. Johnson DB, Maltz DA. Dynamic Source Routing
in Ad Hoc Wireless Networks. In Mobile Comput-
ing, Vol. 353. The Kluwer International Series in
Engineering and Computer Science, Springer, 1996;
153–181.
17. Pirzada AA, McDonald C, Datta A. Performance
comparison of trust-based reactive routing protocols.
IEEE Transactions on Mobile Computing 2006; 5(6):
695–710.
18. Park VD, Corson MS. A highly adaptive distributed
routing algorithm for mobile wireless networks. In
Proceedings IEEE INFOCOM’97. Sixteenth Annual
Joint Conference of the IEEE Computer and Commu-
nications Societies. Driving the Information Revolu-
tion, Vol, 3, Kobe, Japan, 1997; 1405–1413.
19. Venkataraman R, Pushpalatha M, Rama Rao T.
Regression-based trust model for mobile ad hoc
networks. IET Information Security 2012; 6(3):
131–140.
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec

P. R. VAMSI AND K. KANT Trust aware cooperative routing method for WANETs
20. Clausen T, Jacquet P. Optimized link state routing
protocol (olsr). Technical Report, IETF (Internet Engi-
neering Task Force) RFC 3626, 2003.
21. Mistry N, Jinwala DC, Zaveri M. Improving AODV
protocol against blackhole attacks. In Proceedings
of the International Multi Conference of Engineers
and Computer Scientists, Vol, 2, Hong Kong, 2010;
1034–1039.
22. Li X, Jia Z, Zhang P, Zhang R, Wang H. Trust-
based on-demand multipath routing in mobile ad
hoc networks. IET Information Security 2010; 4(4):
212–232.
23. Venkanna U, Agarwal JK, Velusamy RL. A coopera-
tive routing for manet based on distributed trust and
energy management. Wireless Personal Communica-
tions 2015; 81(3): 961–979.
24. Dhulipala VS, Karthik N, Chandrasekaran R. A novel
heuristic approach based trust worthy architecture for
wireless sensor networks. Wireless personal communi-
cations 2013; 70(1): 189–205.
25. Pirzada AA, McDonald C. Trusted greedy perimeter
stateless routing. In 15th IEEE International Confer-
ence on Networks, 2007. ICON 2007, IEEE, Adelaide,
Australia, 2007; 206–211.
26. Vamsi PR, Kant K. An improved trusted greedy
perimeter stateless routing for wireless sensor net-
works. International Journal of Computer Network
and Information Security (IJCNIS) 2014; 6(11):
13–19.
27. Vamsi PR, Kant K. Self adaptive trust model for
secure geographic routing in wireless sensor networks.
International Journal of Intelligent Systems and Appli-
cations (IJISA) 2015; 7(3): 21–28.
28. Jin X, Zhang R, Sun J, Zhang Y. Tight: a geographic
routing protocol for cognitive radio mobile ad hoc
networks. IEEE Transactions on Wireless Communi-
cations 2014; 13(8): 4670–4681.
29. Jsang A, Ismail R. The beta reputation system. In
Proceedings of the 15th Bled Electronic Commerce
Conference, Bled, Slovenia, 2002; 41–55.
30. Network simulator ns-2.35. https://www.isi.edu/
nsnam/ns Accessed: 05-10-2014.
31. Abbas S, Merabti M, Llewellyn-Jones D, Kifayat K.
Lightweight sybil attack detection in manets. IEEE
Systems Journal 2013; 7(2): 236–248.
Security Comm. Networks
(2017) © 2017 John Wiley & Sons, Ltd.
DOI: 10.1002/sec