A Novel Energy-efficient Wormhole Attack Prevention Protocol for
WSN based on Trust and Reputation Factors
Saad Al-Ahmadi
Department of Computer Science, King Saud University, Riyadh, Saudi Arabia
Keywords: Network Security, Trust Mechanism, Wireless Sensor Networks, Wormhole Attack, WSN Simulation.
Abstract: The deployment of Wireless Sensor Networks (WSNs) for the Internet of Things (IoT) is important, but this
also poses some security issues. Wireless Sensor Networks (WSNs) are vulnerable to various attacks, such
as the Wormhole attack. The Wormhole attack is one of the most severe attacks on WSNs that is particularly
challenging to defend against even when the communication is authentic, and sensors are not compromised.
Existing techniques to detect and protect against Wormhole attacks place a substantial burden on the scarce
sensor resources and do not consider the dynamic nature of the network. In this paper, a novel Energy Efficient
Wormhole Attack Prevention Protocol (EWATR) is proposed to protect WSNs against Wormhole attacks.
EWATR is based on trust and reputation among WSN nodes that consider the dynamic nature of the network.
This study also compares EWATR against several state-of-the-art trust and reputation models through
extensive simulations using the TRMSim-WSN simulator. Eventually, the simulation results show the
superiority of EWATR over other proposed protocols in terms of efficient energy consumption and shorter
path length.
1 INTRODUCTION
Wireless Sensor Networks (WSNs) consist of large
numbers of low-power sensors dynamically forming
different topologies. WSNs are notable for their
distributed cooperation, dynamic topology, lack of
central administration, open medium, and constrained
capability (computation and power constraints). Every
sensor serves as both a host and a router, forwarding
packets to and from other sensors. Sensors have many
limitations that require keeping in mind while
establishing networks, like limited memory size,
battery life, and processing performance (Akyildiz et
al., 2002). WSN is vital for the Internet of Things (IoT)
and has numerous military applications, environment
monitoring, predictive maintenance, smart grids,
transportation, buildings, healthcare, to name a few
(Agrawal et al., 2011).
WSN Security is a significant concern because
sensors interact with their surroundings and people
are often physically accessible, allowing possible
physical attacks. WSNs are more vulnerable than
traditional wired and wireless networks to security
threats. A wormhole attack is a severe threat where
powerful routers are injected in strategic locations to
tunnel packets received in one network and replay
them in another place. The attacker only needs two
transceivers and no key material.
There are many solutions proposed in the
literature to protect WSN against wormhole attacks.
This paper presents a novel Energy Efficient Worm-
hole Attack Prevention Protocol (EWATR) based on
trust and reputation among WSN nodes to combat
wormhole attacks. It uses cooperation to establish
trust among neighbouring sensors. EWATR can be
embedded in nay WSN routing protocol, such as Ad
hoc On-Demand Distance Vector (AODV), to detect
and prevent wormhole attacks. It can be extended to
other ad hoc networks such as the Wireless Ad-hoc
network (WANET), where route discovery and node
credibility are the first security requirements in
wireless routing protocols.
This paper is arranged as follows: Section 2
introduces the reader to the necessary background
information. In contrast, Section 3 covers an
extensive literature review of the state-of-the-art
research for detecting and preventing wormhole
attacks— Section 4 presents the proposed approach
EWATR. In Section 5, we show our comparison
results using simulation versus other existing
techniques. Section 6 concludes the work and sheds
light on future work.
Al-Ahmadi, S.
A Novel Energy-efficient Wormhole Attack Prevention Protocol for WSN based on Trust and Reputation Factors.
DOI: 10.5220/0010951400003118
In Proceedings of the 11th International Conference on Sensor Networks (SENSORNETS 2022), pages 191-201
ISBN: 978-989-758-551-7; ISSN: 2184-4380
Copyright
c
2022 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
191
2 BACKGROUND
A wormhole attack is a severe WSN attack in which
the attacker fakes two sensors as the wormhole tunnel
ends, one of which collects, captures the packets, and
forwards them through the tunnel to the other. By
enabling smoother transmission channels, the tunnel
allures adjacent nodes to direct their traffic via the
tunnel rather than different safe routes. When
wormhole nodes only forward packets faster, it is
considered a passive attack. When a wormhole alters
packet content or drops them, it is regarded as an
active attack. As a Denial-of-Service attack, the
wormhole distorts data aggregation and routing
protocols.
Therefore, the attacker can initialize the
wormhole using malicious nodes deployed in several
locations to capture network traffic (packets or bits).
It redirects the network packets using inside an
extended distance tunnel. The wormhole tunnel was
established using either a wired link or high-
frequency wireless links (Aliady & Al-Ahmadi,
2019).
Figure 1 shows communication between the
source node S and the base station. The safe path,
shown in dotted lines, is long and has many hops,
while the dangerous path, shown in dark thick lines,
has fewer hops and passes through a wormhole tunnel
between the two malicious nodes W1 and W2 (Chiu
& Lui, 2006). In some cases, the tunnel establishes a
shortcut route that makes the tunnelled packet arrives
with fewer hops and shorter time. The confidentiality
and authenticity of the communication channel do not
defend against such a dangerous attack (Ban et al.,
2011).
Figure 1: Wormhole attack.
2.1 Wormhole Attack Classification
Wormholes are classified into three types: closed,
half-open, and open (N. Sharma & Singh, 2014), as
shown in Figure 2. Nodes S and D are the source and
destination nodes. Nodes M1 and M2 represent the
malicious nodes. Nodes A, B are benign nodes on the
route between S and D. Curly braces “{}” are used to
hide nodes invisible nodes to S and D. because they
are in the wormhole. The word “closed” means “start
from and include,” while the word “open” means
“start from but not include” (N. Sharma & Singh,
2014).
Figure 2: Types of Wormhole attack.
In closed wormhole attack (a), M1 and M2 take
over the neighbourhood discovery beacons between S
and D, where they assume they are direct neighbours
to each other while they are not. In half-open
wormhole attack (b), M1 is a neighbour of S and
tunnels beacons through M2 to D. Only one malicious
node is visible to S and D. Similar situation can
happen to D where it can see a neighbouring
wormhole node M2 and tunnels beacons through M1
to S. In the open mode wormhole (c), both wormholes
M1 and M2 are visible direct neighbours to S and D.
2.2 Wormhole Attack Modes
Wormhole attacks are classified according to
techniques used to establish the attack, such as
encapsulation, out-of-band channel, packet relay,
protocol deviation, and high-power transmission.
Packet encapsulation is considered an easy way to
launch a wormhole attack. There is no need for extra
specialized hardware or tools, and each packet is
routed only via the legitimate path. This attack class
is based on freezing the hop counter by encapsulating
the packets that reached the wormhole ends. The
packet is transported into the original form by the
second endpoint (Ghugar & Pradhan, 2019).
Therefore, this prevents the right functional nodes
from finding the shortest paths between them and
generates many fake neighbours since the hop counter
is unrealistic.
Using Out-of-Band mode needs much effort to
lunch since it needs extra hardware. It relies on
hardware rather than software. This mode is deployed
SENSORNETS 2022 - 11th International Conference on Sensor Networks
192
with at least two malicious nodes equipped with a
long-range directional wireless link or a direct wired
link (Ghugar & Pradhan, 2019).
Packet relay mode can be deployed with at least
one malicious node without specialized hardware by
initializing a fake neighbour relation between two
right functional indirect nodes and fabricating a
neighbour relation. The malicious node tunnels
packets between indirect nodes, thus controlling all
the traffic without changing the routing protocol.
Protocol deviation attacks can cause serious harm
to the ad hoc network and WSN (Padmavathi &
Shanmugapriya, 2009). It acts as a massive Denial of
Service (DoS) attack. This attack is also known as the
Rushing Attack. It is launched against many existing
on-demand routing protocols by using one or more
malicious nodes. In many existing routing protocols
such as AODV and DSR, a node can discover the
destination route by sending multiple RREQ packets
to know the suitable route to the destination node.
Each node forwards the RREQ packet only once to
prevent this flood’s high overhead. The intruder will
abuse this property in this attack by forwarding the
RREQ route discovery packet to other nodes. It will
cause any node to receive the rushed RREQ will
discard any further RREQ from the same route
discovery.
The high-power transmission attack’s main idea
is to use a high-power source such as an
electromagnetic radio frequency signal to send a
high-power broadcast to various network nodes.
Therefore, nodes that received the transmitted high-
power broadcast rebroadcast it to the destination
node. This method deploys the wormhole tunnel
simply without using any colluding node. Also,
malicious nodes are on the route between the source
and the destination nodes (Johnson et al., n.d.).
2.3 Countermeasures to Wormhole
Attacks
This section will briefly describe other detection and
prevention techniques of wormhole attacks in WSN
proposed by several researchers.
2.3.1 Secure Geographic Routing Protocol
Detecting wormholes based on a node’s geographical
location is a typical detection and prevention
technique—the coordinates of every node within the
WSN are used for secure geographic routing. The
sender node includes its current geographical location
in the packet and the time stamp so the receiver node
can verify communication and falsify malicious
nodes. The receiver node calculates the possible and
estimated transmission range and checks the
eligibility of transmission time. This technique
requires an accurate and synchronized timer with a
geographic positioning system such as GPS to obtain
good results. The usage of additional hardware has
many disadvantages, such as exceeding the overall
network cost and reducing battery life, especially
within the WSN that uses small or low power sensors.
There are many existing geographic protocols, for
example, Greedy Perimeter Stateless Routing
(GPSR) (Ratnasamy et al., 2001). GPSR is inspired
by a greedy algorithm where every node within the
network knows its current location by using beacons
to identify the closest neighbour and near the ultimate
destination.
2.3.2 Statistical Analysis Approach
One possible solution to detect and forestall the
wormhole attack is using statistics and probabilities.
We can utilize this vital information to see the
wormhole tunnel by studying nodes’ routes,
frequency, and neighbour discovery. The statistical
analysis approach can provide the right solution
depending on the sensors and network configuration.
For example, if the sensors have a set of radio
frequency ranges within fixed boundaries, detecting
the malicious node that uses a high-power
transmission would be more straightforward and
notable. The researchers proposed a new statistical
analysis scheme that explores the network routes by a
statistical routing protocol (Somu & Mallapur, 2019).
It can spot the attack by inspecting the behaviour of
the sensors and the transmitted information. Other
researchers have suggested statistical techniques that
focus intensely on investigating the state of sensor
nodes and their closest neighbours. Based on the
received neighbourhood information, the base
station(s) can discover the presence of wormholes
probabilistically using hypothesis testing (Buttyán et
al., 2005).
2.3.3 Cryptography Key
Using public or symmetric key cryptography is a
method proposed by many researchers. Keys building
and exchange techniques for secure transmission in
distributed environments is a challenging task.
Encryption and decryption in WSN consume large
amounts of those limited sensor resources such as
processing capabilities, memory available, and little
energy (Babaeer & Al-Ahmadi, 2020). In (Zhang et
al., 2006), a new scheme based on public and private
authentication keys was proposed to eliminate
A Novel Energy-efficient Wormhole Attack Prevention Protocol for WSN based on Trust and Reputation Factors
193
malicious nodes before packet sending. The
researchers have suggested a key management
scheme based on the public key exchanging methods
to improve security matters in ad hoc networks and
WSNs (Salam et al., 2010). In (Ju, 2012), Elliptic
Curve Cryptography (ECC) is used as a routing-
driven key management scheme to help the exchange
mechanism between neighboring nodes. Nodes only
accept packets from authenticated neighbors with a
valid key and ignore other packets coming from fake
nodes. In this way, malicious nodes cannot connect
with rightful nodes to gain trust and validity. Another
critical management scheme solution was proposed in
(P. Sharma, 2012), where three keys were used for
different purposes. A global key (network key) is
generated periodically by the base station. All nodes
share it in the network and are used to encrypt all
broadcasted messages within the network. The last
two keys are pair-wise keys used by network nodes
for authentication and exchanging secure messages
purposes. An asymmetric cryptography-based key
management scheme was suggested in (Khalil et al.,
2007). It adopts a new fundamental management
approach by inheriting the advantages of asymmetric
cryptography and employs it efficiently for delivering
session keys to sensor nodes.
2.3.4 Trust Mechanism
Trusted-based is another approach that has been
proposed by many researchers as a solution to
wormhole attacks as well as other attacks like sink-
hole attacks, black hole attacks, and DoS attacks. The
trust is built in the network layer routing protocol by
building trusted links among all nodes within the
WSN before establishing communication. In
(Simaremare et al., 2013), the researchers have
proposed a model improving security in ad hoc On-
Demand Distance Vector (AODV) routing protocol
by adding a trust factor that consists of three primary
agents: trust agent, reputation agent, and combined
agent. The trust agent model records the node’s trust
information so the reputation agent can spread the
recorded data to all nodes. The combiner agent
generates consolidated trust factors from the previous
agents’ received information. Based on AODV, a
local and global trust calculation method is proposed
in (P. Sharma, 2012) to detect and prevent ad hoc
network attacks. The Trust AODV (TAODV)
calculates trust by gathering all activity information
from surrounding nodes. It enhances the existing
AODV by introducing new fields into the node’s
routing table to record other nodes’ positive and
negative feedback. A node is known for its
trustworthiness if the input from other nodes is
positive. Opinions among nodes change dynamically
with the rise of successful or failed communication
times. The AODV routing protocol is subject to
extensive research and has many variations, yet it is
still an open research area because of its flexibility.
3 LITERATURE REVIEW
The wormhole attack problem got the attention of
many researchers because it is a crucial issue in WSN
and it affects routing protocols and network lifetime.
In the literature, many wormholes detection
approaches have been proposed. The authors
proposed a robust trust model for a cluster-based
network in (Gautam & Kumar, 2018) to secure
against collusion attacks. The model is based on
direct trust using a time-lapse function. The function
takes advantage of the forgetting curve to calculate
direct trust—another reputation function used for
indirect trust (recommendation-based trust)
calculation. The cluster head plays the
recommendation manager’s role and builds indirect
trust by collecting information from all neighbours of
the target node. Simple mathematical operations
calculate direct trust, and it entirely depends upon the
packet transaction among nodes. Active and trust
communication among nodes is achieved when the
trust value is high. This model does not provide
complete trust information because it only considers
communication trust.
A MANET is a decentralized, ordinary, and self-
orbiting type of wireless network that has the
capability of managing mobile nodes. To detect
wormhole attacks in MANET, a lightweight approach
has been proposed in (Zardari et al., 2018). All replies
(RREP) packets with sequence numbers are stored
temporarily at the source node. The source node
calculates an average of all sequence number and then
store it. If any of the nodes exceeds the average, it will
discard all replies packets. Hence, wormhole nodes
can be eliminated from the network, and those nodes
will be communicating in the network who are
trusted. The proposed approach in this study is power-
efficient, less complex, and increases network
lifetime in large networks.
A similar task has been done in (An & Cho, 2021)
to detect wormhole attacks in WSN. When events
occur in the WSN, the sensor node in the vicinity of
the generation event recognizes the event, generates a
report of the detected event, and sends it to the base
station. A wormhole attack, where in the report is sent
over a various path than the original path, can happen
SENSORNETS 2022 - 11th International Conference on Sensor Networks
194
during the transmission process. As a result, a
wormhole attack is detected, including encrypted
node ID and hop count into the report content during
the statistical en-route filtering methods report
production process. Furthermore, in this study, they
proposed a wormhole detection technique that can
enhance security. The proposed approach detects both
wormhole attacks and false report injection together.
Ukil in (Ukil, 2010) presents a simple yet efficient
model based on collaborative computing through
trust and reputation. It gives the possibility to find an
optimal routing path that optimizes both reliability
and communication efficiency. In the case of a trusted
environment, no authentication is needed for a node
to communicate with other nodes and to unwrap
hidden keys. Reputation is a global phenomenon that
tends to the history of the trustworthiness of nodes.
The proposed model gives reliability preference over
efficiency to protect and secure the network from
insider threats. A worm propagation detection scheme
was presented in (Ho & Wright, 2017), according to
which a randomized-SPRT (Sequential Probability
Ratio Test) was built on sequential traffic analysis to
detect worm propagation through legitimate traffic.
Detector nodes observe communication patterns in
the network and identify the worm propagation
pattern. After the initial remote attestations to detect
infected nodes, detector nodes create chains of
connections that would not be seen in regular traffic
due to its spreading hop-by-hop. When some nodes
are compromised, a self-propagating malicious code
can take over the network and cause damage. The
authors presented a methodology based on round trip
time (RTT) and hop count to identify wormhole
attacks (M. A. Patel & Patel, 2018). This approach
requires two steps; the first step uses a hop counting
mechanism based on the RTT algorithm. In every
sensor node, built local maps to detect abnormalities
and use a diameter feature transmission range. In the
second step, both neighbours’ ids and RTT are
collected between every two successive nodes. This
technique does not need any specialized hardware and
saves energy consumption. Simulation results show
that this scheme has high detection accuracy.
The study (N. Sharma et al., 2020) proposes a
technique in which each node's trust value is
calculated by watching N2N packet delay during
packet transaction between neighbouring nodes, and
malicious node identification is conducted based on
this trust value. The trust value can also be used to
choose a secure way and make a routing decision. For
network performance valuation, the DSR routing
protocol is utilized in the existing malicious nodes.
The researchers produced a defence method to
protect wormhole attacks using the neighbour
information collected for all nodes in WSN (Chu et
al., 2019). This mechanism does not require extra
hardware, complex calculation and does not consume
significant resources. The proposed method depends
on a Quantum-inspired Tabu Search (QTS)
algorithm. This algorithm relies on finding
combinations of the nodes to indicate the detection of
wormholes. The experimental results show that the
detection of wormhole attacks is highly successful.
A similar situation appears in ad hoc network,
Mobile ad-hoc networks (MANETs), needs
protection from various attacks like the wormhole
that benefits from the absence of centralized
infrastructure. In MANETs, routing suffers when a
participating node does not set its intended function
and performs malicious activity. The attacker can
record packets at one location in the network, transfer
them to another location, and retransmits them into
the networks. Researchers introduced a modified
AODV to discover wormhole attacks without
specialized hardware such as a directional antenna
and a specific synchronized clock (Gupta et al.,
2011). Their protocol is independent of the physical
medium of the wireless network. In route discovery,
the source node starts the wormhole detection process
in the discovered path, which counts hop difference
among the neighbours of one of the hops away nodes.
The destination node discovers a wormhole if the hop
difference between neighbours of the nodes exceeds
the acceptable level. The algorithm uses an additional
hound packet to check if the ad hoc network is formed
between trusted parties. It does not require sending a
hound packet when the network is public, or nodes
experience a high packet dropping rate. The hound
packet is sent after the path discovery phase. The
simulation results show that the algorithm works well
in detecting wormholes of extensive tunnel lengths.
The researchers introduced Multi-Layered
Intrusion Detection (MLDW) as a method for
detecting and preventing wormhole attacks on
MANET in (C.P & Saviour Devaraj, 2013). MLDW
utilizes the AODV route discovery phase with
adequate response time and uses layer framework
information such as link latency estimator,
intermediate neighbour node discovery mechanism,
packet drop calculator, and node energy degrade
estimator followed by isolation technique. MLDW
does not require any specialized hardware or clock
synchronization and achieves good defence results.
Authors in (M. Patel et al., 2019) have presented
various detection features that help for wormhole
detection. Existing wormhole detection techniques
A Novel Energy-efficient Wormhole Attack Prevention Protocol for WSN based on Trust and Reputation Factors
195
mainly rely on four detection features: time,
neighbourhood, location, and hop count. The time
feature indicates that the suspicious path has more
average time per hop than the safe path. Techniques
based on time information require clock
synchronization—the neighbourhood feature used by
detection techniques to store information on one hop
or two hops’ neighbours. Analysing and storing two
hops neighbour’s data in the network requires more
storage processing, power, and memory. The location
feature is significant for detecting wormhole to know
the location need to direct antennas and GPS. The hop
count feature shows that wormhole attracts traffic by
advertising a shorter route with a fewer hop count—
existing techniques used hop count feature in
collaboration with location and time features.
The authors proposed a new method for
preventing wormhole attacks by using an algorithm to
manage many nodes using node ID (Buch & Jinwala,
2011). In this method, Node ID was linked to the
surrounding nodes in a binary tree structure. The
algorithm created a load balancing feature for
monitoring a hierarchical system for nodes and an
extra packet header on each node to ensure it was not
compromised. The experiment developed a
simulation using C# on the Windows platform to
study only the binary structure of how the nodes
severed in a mesh network.
A Wormhole Resistant Hybrid Technique
(WRHT) was presented as a more positive way of
detecting the presence of a wormhole attack than
existing strategies in (Singh et al., 2016). WRHT
relies on the concept of watchdog and Delphi
schemes and verifies that the wormhole will not be
left untreated in the sensor network. WRHT depends
on the dual wormhole detection mechanism of
calculating probability factor time delay and packet
loss probability of the established path to find the
value of wormhole presence probability. All nodes in
the path are given a different ranking and,
subsequently, colours according to their behaviour. It
does not require additional hardware such as a global
positioning system (GPS), timing information,
synchronized clocks, or high computational needs
like traditional cryptographic schemes. The
experimental results showed significant improvement
in the detection rate of the wormhole attack.
4 PROPOSED APPROACH
The proposed protocol is an Energy Efficient Worm-
hole Attack prevention scheme in WSN based on
Trust and Reputation factor (EWATR). It makes use
of trust and reputation information (Khalid et al.,
2013) to calculate the trustworthiness of a node to
judge a node’s misbehaviour and effectively
distinguish between normal operating and malicious
nodes. In EWATR, the positive and negative effects
of a node’s action are observed. The observations are
aggregated in a specific trust table maintained by the
node. Statistical analysis is performed on the trust
table data to generate the node reputation.
Node’s trust computation based on:
information collection,
information dissemination,
information mapping to trust model, and
Decision making.
4.1 Information Collection
Information is collected by node (say A) through
packet broadcasting. Node A monitors another node
(say B) in promiscuous mode. When node B takes the
responsibility to forward the packet received from A,
it will broadcast it so node A can hear it again. Now,
node A will verify the packet contents and update the
trust rating for node B. In this manner, nodes can
collect what is called first-hand trust information.
When a sensor node broadcasts a packet, the node
monitors the neighbouring node through a watchdog
mechanism. After sending the packet to a neighbour,
the watchdog mechanism requires that the node
monitor the neighbour in a promiscuous mode, as
illustrated in figure 3, to verify whether the neighbour
has forwarded or intentionally dropped the packet. If
the neighbouring node takes the packet forwards, the
trust rating is incremented and updated in the sender
nodes database. Conversely, if the adjacent node
drops the packet, the trust rating is reduced in the
sender node’s database.
Figure 3: Description of promiscuous mode.
4.2 Information Dissemination
In EWATR, network nodes distribute their first-hand
information to the neighbouring nodes. Such kind of
distribution information is called second-hand
information, as implemented in Figure 4. The use of
second-hand information is beneficial and makes the
reinforcement of the reputation process faster.
Second-hand information sets up a global view of
trust in the network. The nodes extend second-hand
information either proactively, after some fixed time,
SENSORNETS 2022 - 11th International Conference on Sensor Networks
196
or reactively, on the occurrence of some event or
some essential change in the network.
Figure 4: Second-hand trust information dissemination.
4.3 Information Mapping to the Trust
Model
In this phase, network nodes add first-hand and
second-hand information to produce a trust and
reputation metric. First-hand data is direct, and more
computation effort is required to combine first-hand
data into the reputation metric. The credibility of the
reporting node is made through various techniques
suggested in the literature. One such technique, called
the deviation test, is presented in (Ganeriwal et al.,
2008). The following inequality represents the
deviation test: Where d is a threshold, the value of
which is set truly. In the inequality given in Equation
(1), α and β are two parameters of the statistical Beta
distribution, which is used for decision-making.
 𝐵𝑒𝑡𝑎
∝,𝛽
𝐸
𝐵𝑒𝑡𝑎∝
,𝛽
𝑑
(1)
Here, α select right nodes while β selects nodes
with bad behaviours. The probability value E (Beta(α,
β)) is the measure of the current trust information
node A has about node B, whereas E (Beta(αf, βf))
shows the new trust information provided by some
node C to node A about node B. The reporting node
C is considered trustworthy if the left-hand side of the
difference in Equation (1) produces a value less than
threshold d.
Various trust and reputation mechanisms use
different statistical models to estimate the correctness
of second-hand information, relying on the
application and security requirements. The most used
schema is the Beta distribution, whose probability
density function is expressed using the Gamma
function as
𝑃
𝑥
𝐵𝑒𝑡𝑎
∝,𝛽
,
𝑥

1𝑥

,∀0
𝑥1,0,𝛽0
(2)
Where α refers to the acceptable behaviour, and β
represents the bad behaviour of a node. Equation (2)
suggested another way of measuring the consistency
of data by a reporting node. For example, let us
assume a node B provides trust information to node
A for P+Q times. If information is regular for p times,
then P(x) is implemented to predict its regularity in
the subsequent observation. If P(x) results in a value
equal to 1, then the reported information is regular
and authentic; otherwise, it is considered unauthentic
information.
4.4 Decision Making
The final step shows the decision-making process.
The decision relies on the computed trust values.
Node A decision may be one of the two binary values,
where a “1” refers to participating and forwarding the
packet, and a “0” means not cooperating. Figure 5
graphically illustrates the decision‐making process.
Figure 5: Trust computation process shows how weights
applied to set trust thresholds.
5 EXPERIMENT EVALUATION
We simulated three experiments to evaluate and
compare the performance of BTRM-WSN (Gómez
Mármol & Martínez Pérez, 2011), Eigen Trust
(Kamvar et al., n.d.), Peer Trust (Xiong & Liu, 2004),
Power Trust (Zhou & Hwang, 2007), Linguistic
Fuzzy Trust Model (LFTM) (Mármol et al., 2010),
and Trust and Reputation Infrastructure Based
Proposal model (TRIP) (Mármol & Pérez, 2012). The
first experiment runs on comparing the seven systems
in terms of accuracy in searching for trustworthy
sensors. This experiment estimates the level of
security provided in WSN. The second experiment
compares the average path length leading to the
trustworthy sensors selected. It assesses the
efficiency, or the facility for discovering reliable
sensors, of the seven techniques. Finally, the total
energy saving of applying these seven techniques in
WSNs is measured. The average Path length is the
A Novel Energy-efficient Wormhole Attack Prevention Protocol for WSN based on Trust and Reputation Factors
197
efficiency measurements in ease of discovering
sensor nodes.
5.1 TRMSim-WSN Simulation
We implemented the seven trust and reputation model
simulators for WSN using TRMSim-WSN (Mármol
& Pérez, 2009). It is a java-based trust and reputation
model simulator that provides a straightforward
method to test a trust and reputation model over WSN
and compare it against other models. TRMSim-WSN
users may choose between static and dynamic
networks, as well as the proportion of fraudulent
nodes, the percentage of nodes functioning as clients
or servers, etc. We set up WSN based on network
parameter settings shown in table 1. In a randomly
created WSN, 30% of all nodes are clients that will
request default services. The other 70% nodes will be
servers, which are asked to provide services upon
request. WORMHOLE NODES act in pairs for a
single transaction when two legitimate nodes are
supposed to communicate (for sharing of files
between two legitimate nodes). There may be more
than two nodes, both even or odd nodes. Statistical
analysis may give more accurate results (Johnson et
al., n.d.). If the movements are random in the
network, wormhole nodes may have trouble catching
up dynamically. Similarly, it is hard for legitimate
ones to detect untrustworthy nodes.
Furthermore, in a typical transaction, two
WORMHOLE nodes are needed for the tunnel to be
established; one may be a Server node, while the one
at the other end will be a client sensor node.
Having 70 percent server nodes and 30 percent
client nodes means there are more sensing nodes
information being transmitted by servers than by
clients, which means that a client may be sending
requests to more than one server node, at a time. A
broadcast message by the client may reach multiple
Server or even client nodes. Interestingly, a wormhole
may also send a broadcast to the serving node and its
other peer (worm) closest to the client node.
In network topologies in an infrastructure-based
client-server model, clients are more than a single
Server. For instance, a Webserver may serve
thousands of clients’ requests a second.
This may be because sensing information like
humidity, temperature, wind speed, etc. maybe being
more often transmitted to hub/gateway/hybrid nodes,
which in turn may transmit or relay to infrastructure
nodes.
Table 1: Experiment Parameters.
NumExecutions 100 %Clients 20%
NumNetworks 100 %Relay 5%
MinNumSensors {50,100,150,200} %Malicious 70%
MaxNumSensors {50,100,150,200} Radio range {10, 8, 6, 4}
5.2 Accuracy
To estimate the reliability and level of security
provided by the trust and reputation system in WSN.
The accuracy of a trust and reputation system is
implemented by the percentage of times when it
successfully selects trustworthy sensors (the former
situation) out of the total number of transactions.
When a transaction take place, it from wormholes.
It may take several rounds of transactions, after which
the culprit nodes are caught.
Figure. 6 demonstrates the accuracy of BTRM-
WSN, Eigen Trust, Peer Trust, Power Trust, LFTM,
TRIP, and EWATR Techniques with various sensor
nodes over WSN. We conclude that the EWATR
technique can approximately provide the highest
accuracy and, consequently, the highest level of
reliability and security, while TRIP offers the lowest
value. Furthermore, it has been observed that the Peer
Trust model is the most oscillated and unstable
accuracy. Fig 6 shows accuracy calculations based on
static WSN. For nodes with similar movements, it
may be difficult for wormhole nodes to catch up to
locked-in victim nodes in all the above-mentioned
techniques. Hence repeating the exercise for
movements based on WSN may be futile.
Figure 6: Selection percentage of trustworthy servers over
static WSNs.
5.3 Path Length
The path length is the rate hops that lead to the most
trustworthy sensors selected by the client in a WSN
using a particular trust and reputation system. It
assumed that less average path length indicates better
performance in efficiency and facility searching for
SENSORNETS 2022 - 11th International Conference on Sensor Networks
198
trustworthy sensors of a trust and reputation system.
The reasons behind this claim: 1) fewer
intermediaries mean higher security level; and less
energy consumption; and 2) shorter path length refers
to that it is easier to find honest nodes, and thus,
server nodes will respond quicker to client nodes.
Figure 7 compares BTRM-WSN, Eigen Trust, Peer
Trust, Power Trust, LFTM, TRIP, and EWATR in
terms of rate path length leading to trustworthy sensors
with various sensor nodes over WSN. TRIP and
EWATR have the highest performance since, in TRIP,
the rate path length remains one by increasing the
number of sensor nodes. Moreover, Eigen Trust has the
lowest performance in terms of shorting average path
length. Eigen Trust, Power Trust, and Peer Trust are
unstable and lengthiest in average path length that
BTRM, LFTM, TRIP, and EWATR models.
With TRIP, the number of the hop, i.e., average
distance, remains the same, i.e., one. This means that
the Client and Server nodes are adjacent. The
Wormholes attack strives on spoofing and fooling the
Server-Client, that the best path between these is
through these.
Figure 7: Average path-length leading to trustworthy
servers over static WSNs.
5.4 Energy Consumption
Energy consumption of the network is the overall
energy consumed in (1) client nodes sending request
messages; (2) server nodes sending response services;
(3) energy consumed by malicious nodes which
provide inadequate services; (4) relay nodes that do
not provide services; and (5) the energy to execute the
trustworthy sensor searching process of a particular
trust and reputation system. How to effectively
reduce energy consumption is the main problem in
WSN research. Wormholes increase the wireless
energy transmission to fake being closer to sensor
nodes and thus gain trust and access to genuine nodes.
This also fakes path lengths. It takes several rounds to
detect culprit nodes, by which time, critical battery
drainage may have occurred in some cases. In some
instances, mini solar panels are utilized in WSN,
which only get a few minutes of light, e.g., in the
canopy of forests.
Figure 8 compares the energy consumption of
BTRM-WSN, Eigen Trust, Peer Trust, Power Trust,
LFTM, TRIP, and EWATR, respectively, over WSN
and by various sensor nodes. It can be concluded that
EWATR is the most energy consumption model in
WSN environments, while TRIP is the least energy
consumption.
Figure 8: Network energy consumption over WSNs.
6 CONCLUSIONS
The wormhole attack employs various modes to
launch one of the most severe malicious threats to
WSN. Self-configuring and autonomous systems,
such as WSNs, rely on trust to make successful
judgments upon recognizing a misbehaving node.
The task of building trust and reputation becomes
challenging when the nodes are mobile. We have
suggested the EWATR protocol and compared it with
BTRM-WSN, Eigen-Trust, Peer-Trust, Power-Trust,
LFTM, and TRIP. This paper concluded that the
EWATR technique’s accuracy approximately
provides the best accuracy and, thus, the best level of
reliability and security. While rate path length
concludes, TRIP and EWATR have the highest
performance, and Eigen Trust has the lowest
performance in terms of shorting rate path length.
Also, the energy consumed conducted that EWATR
is the most energy consumption model in WSN
environments, while TRIP is the least energy
consumption. Future work will evaluate the proposed
algorithm for various network conditions, such as
frequent link breaks between nodes, which is a typical
problem in a practical setting.
REFERENCES
Agrawal, S., Jain, S., & Sharma, S. (2011). A survey of
routing attacks and security measures in mobile ad-hoc
networks. ArXiv Preprint ArXiv:1105.5623.
A Novel Energy-efficient Wormhole Attack Prevention Protocol for WSN based on Trust and Reputation Factors
199
Akyildiz, I. F., Su, W., Sankarasubramaniam, Y., & Cayirci,
E. (2002). Wireless sensor networks: a survey.
Computer Networks, 38(4), 393–422.
Aliady, W. A., & Al-Ahmadi, S. A. (2019). Energy
preserving secure measure against wormhole attack in
wireless sensor networks. IEEE Access, 7, 84132–
84141.
https://doi.org/10.1109/ACCESS.2019.2924283
An, G. H., & Cho, T. H. (2021). Wormhole detection using
encrypted node IDs and hop counts in the event report
of statistical en-route filtering. International Journal of
Computer Networks and Applications, 8(4), 390–399.
https://doi.org/10.22247/ijcna/2021/209705
Babaeer, H. A., & Al-Ahmadi, S. A. (2020). Efficient and
Secure Data Transmission and Sinkhole Detection in a
Multi-Clustering Wireless Sensor Network Based on
Homomorphic Encryption and Watermarking. IEEE
Access, 8, 92098–92109. https://doi.org/10.1109/
ACCESS.2020.2994587
Ban, X., Sarkar, R., & Gao, J. (2011). Local connectivity
tests to identify wormholes in wireless networks.
Proceedings of the Twelfth ACM International
Symposium on Mobile Ad Hoc Networking and
Computing, 1–11.
Buch, D., & Jinwala, D. (2011). Prevention of Wormhole
Attack in Wireless Sensor Network. International
Journal of Network Security & Its Applications, 3(5),
85–98. https://doi.org/10.5121/ijnsa.2011.3507
Buttyán, L., Dóra, L., & Vajda, I. (2005). Statistical
wormhole detection in sensor networks. European
Workshop on Security in Ad-Hoc and Sensor Networks,
128–141.
Chiu, H. S., & Lui, K.-S. (2006). DelPHI: wormhole
detection mechanism for ad hoc wireless networks.
2006 1st International Symposium on Wireless
Pervasive Computing, 6-pp.
Chu, T.-H., Kuo, S.-Y., & Chou, Y.-H. (2019). Using
Quantum-inspired Tabu Search Algorithm with Logic
Operation and Moving Average Indicator for
Wormhole Attack Detection in a WSN. Journal of
Internet Technology, 20(1), 167–176.
C.P, V., & Saviour Devaraj, A. F. (2013). MLDW- A
MultiLayered Detection mechanism for Wormhole
attack in AODV based MANET. International Journal
of Security, Privacy and Trust Management, 2(3), 29–
41. https://doi.org/10.5121/ijsptm.2013.2303
Ganeriwal, S., Balzano, L. K., & Srivastava, M. B. (2008).
Reputation-based framework for high integrity sensor
networks. ACM Transactions on Sensor Networks
(TOSN), 4(3), 1–37.
Gautam, A. K., & Kumar, R. (2018). A robust trust model
for wireless sensor networks. 2018 5th IEEE Uttar
Pradesh Section International Conference on Electrical,
Electronics and Computer Engineering (UPCON), 1–5.
Ghugar, U., & Pradhan, J. (2019). A Review on Wormhole
Attacks in Wireless Sensor Networks Intrusion
detection of various attack in WSN View project A
Review on Wormhole Attacks in Wireless Sensor
Networks. International Journal of Information
Communication Technology and Digital Convergence,
4(1), 32–45. https://www.researchgate.net/publication/
336798552
Gómez Mármol, F., & Martínez Pérez, G. (2011). Providing
trust in wireless sensor networks using a bio-inspired
technique. Telecommunication Systems, 46(2), 163–
180. https://doi.org/10.1007/s11235-010-9281-7
Gupta, S., Kar, S., & Dharmaraja, S. (2011). WHOP:
Wormhole attack detection protocol using hound
packet. 2011 International Conference on Innovations
in Information Technology, 226–231.
Ho, J. W., & Wright, M. (2017). Distributed detection of
sensor worms using sequential analysis and remote
software attestations. IEEE Access, 5, 680–695.
https://doi.org/10.1109/ACCESS.2017.2648853
Johnson, M. O., Siddiqui, A., & Karami, A. (n.d.). A
Wormhole Attack Detection and Prevention Technique
in Wireless Sensor Networks.
Ju, S. (2012). A lightweight key establishment in wireless
sensor network based on elliptic curve cryptography.
2012 IEEE International Conference on Intelligent
Control, Automatic Detection and High-End
Equipment, 138–141.
Kamvar, S. D., Schlosser, M. T., & Garcia-Molina, H. (n.d.).
The EigenTrust Algorithm for Reputation Management
in P2P Networks.
Khalid, O., Khan, S. U., Madani, S. A., Hayat, K., Khan, M.
I., Min‐Allah, N., Kolodziej, J., Wang, L., Zeadally, S.,
& Chen, D. (2013). Comparative study of trust and
reputation systems for wireless sensor networks.
Security and Communication Networks, 6(6), 669–688.
Khalil, I., Bagchi, S., & Shroff, N. B. (2007). Liteworp:
Detection and isolation of the wormhole attack in static
multihop wireless networks. Computer Networks,
51(13), 3750–3772.
Mármol, F. G., Marín-Blázquez, J. G., & Pérez, G. M.
(2010). Linguistic fuzzy logic enhancement of a trust
mechanism for distributed networks. Proceedings -
10th IEEE International Conference on Computer and
Information Technology, CIT-2010, 7th IEEE
International Conference on Embedded Software and
Systems, ICESS-2010, ScalCom-2010, 838–845.
https://doi.org/10.1109/CIT.2010.158
Mármol, F. G., & Pérez, G. M. (2009). TRMSim-WSN,
trust and reputation models simulator for wireless
sensor networks. IEEE International Conference on
Communications. https://doi.org/10.1109/ICC.2009.51
99545
Mármol, F. G., & Pérez, G. M. (2012). TRIP, a trust and
reputation infrastructure-based proposal for vehicular
ad hoc networks. Journal of Network and Computer
Applications, 35(3), 934–941.
Padmavathi, G., & Shanmugapriya, M. D. (2009). A Survey
of Attacks, Security Mechanisms and Challenges in
Wireless Sensor Networks. In IJCSIS) International
Journal of Computer Science and Information Security
(Vol. 4, Issue 1).
Patel, M. A., & Patel, M. M. (2018). Wormhole attack
detection in wireless sensor network. 2018
International Conference on Inventive Research in
Computing Applications (ICIRCA), 269–274.
SENSORNETS 2022 - 11th International Conference on Sensor Networks
200
Patel, M., Aggarwal, A., & Chaubey, N. (2019). Analysis
of Wormhole Detection Features in Wireless Sensor
Networks. International Conference on Internet of
Things and Connected Technologies, 22–29.
Ratnasamy, S., Francis, P., Handley, M., Karp, R., &
Shenker, S. (2001). A Scalable Content-Addressable
Network.
Salam, M. I., Kumar, P., & Lee, H. (2010). An efficient key
pre-distribution scheme for wireless sensor network
using public key cryptography. The 6th International
Conference on Networked Computing and Advanced
Information Management, 402–407.
Sharma, N., Sharma, M., & Sharma, D. P. (2020). A Trust
based Scheme for Spotting Malicious Node of
Wormhole in Dynamic Source Routing Protocol. 2020
Fourth International Conference on I-SMAC (IoT in
Social, Mobile, Analytics and Cloud)(I-SMAC), 1232–
1237.
Sharma, N., & Singh, U. (2014). Various Approaches to
Detect Wormhole Attack in Wireless Sensor Networks.
International Journal of Computer Science and Mobile
Computing, 3(2), 29–33.
Sharma, P. (2012). Trust based secure AODV in MANET.
Journal of Global Research in Computer Science, 3(6),
107–114.
Simaremare, H., Abouaissa, A., Sari, R. F., & Lorenz, P.
(2013). Secure AODV routing protocol based on trust
mechanism. In Wireless Networks and Security (pp. 81–
105). Springer.
Singh, R., Singh, J., & Singh, R. (2016). WRHT: A Hybrid
Technique for Detection of Wormhole Attack in
Wireless Sensor Networks. Mobile Information
Systems, 2016. https://doi.org/10.1155/2016/8354930
Somu, P., & Mallapur, D. (2019). Research Trends in
Secure Routing Protocols and Communication System
in WSNs. International Journal of Innovative
Technology and Exploring Engineering, 9(1), 3413–
3421.
Ukil, A. (2010). Trust and reputation based collaborating
computing in wireless sensor networks. 2010 Second
International Conference on Computational
Intelligence, Modelling and Simulation, 464–469.
Xiong, L., & Liu, L. (2004). Peertrust: Supporting
reputation-based trust for peer-to-peer electronic
communities. IEEE Transactions on Knowledge and
Data Engineering, 16(7), 843–857.
Zardari, Z., Memon, K., Shah, R., Dehraj, S., & Ahmed, I.
(2018). A lightweight technique for detection and
prevention of wormhole attack in MANET. ICST
Transactions on Scalable Information Systems, 165515.
https://doi.org/10.4108/eai.13-7-2018.165515
Zhang, Y., Liu, W., Lou, W., & Fang, Y. (2006). Location-
based compromise-tolerant security mechanisms for
wireless sensor networks. IEEE Journal on Selected
Areas in Communications, 24
(2), 247–260.
Zhou, R., & Hwang, K. (2007). Powertrust: A robust and
scalable reputation system for trusted peer-to-peer
computing. IEEE Transactions on Parallel and
Distributed Systems, 18(4), 460–473.
A Novel Energy-efficient Wormhole Attack Prevention Protocol for WSN based on Trust and Reputation Factors
201