Received: 19 November 2020 Revised: 27 December 2020 Accepted: 28 December 2020

DOI: 10.1002/ett.4224

RESEARCH ARTICLE

CTrust-RPL: A control layer-based trust mechanism for

supporting secure routing in routing protocol for low power
and lossy networks-based Internet of Things applications

Temur ul Hassan1 Muhammad Asim1 Thar Baker2 Jawad Hassan1

Noshina Tariq1

1
Department of Computer Science,
National University of Computer and Abstract
Emerging Sciences, Islamabad, Pakistan Internet of Things (IoT) fostered a new epoch of innovation by interconnect-
2
Department of Computer Science, ing digital devices to make human life more convenient and attractive. These
University of Sharjah, Sharjah, UAE
smart objects are largely deployed as low power and lossy networks (LLNs) and
Correspondence to: use routing protocol for LLNs (RPL) for routing. Unfortunately, it is extremely
Muhammad Asim, Department of vulnerable to a large variety of external and internal attacks to cause devas-
Computer Science, National University of
Computer and Emerging Sciences,
tating and calamitous effects. However, this article’s scope revolves around
Islamabad, Pakistan. internal attacks only, where nodes are already part of a legitimate network. Var-
Email:[email protected] ious trust-based mechanisms have been proposed to secure the RPL protocol
from insider attackers. Existing trust mechanisms cause high energy depletion
due to complex computation on the node level, which consequently decreases
the performance of LLNs. Therefore, this article presents a novel hierarchical
trust-based mechanism “CTrust-RPL” by assessing the trust of nodes based on
their forwarding behaviors. This study ships complex trust-related computations
to the higher layer, known as the controller, to save computational, storage, and
energy resources at the node level. We also compare the proposed mechanism
with a state-of-the-art technique called Sec-trust. Our mechanism demonstrates
superior performance in detecting and isolating blackhole attacks. The results
depict that CTrust-RPL detects and isolates 10% more malicious nodes than
Sec-trust in the same time-lapse. The average packet loss ratio difference is less
for our proposed mechanism, with 35% more energy efficiency.

1 I N T RO DU CT ION

Internet of Things (IoT) has become an essential part of our personal lives. It is emerging as an epoch of innovation, where
devices belonging to digital and machine ecosystems are interconnected over the Internet to yield efficacy and conve-
nience in academia, industries, and human lives.1-3 Technologies like 5G and 6G enable the next generation of wireless
communication systems in compliance with sophisticated techniques for security.4,5 They support the huge IoT infras-
tructures, which can be defined as a link, management, and communication of a large number of smart and sensing things
(also known as objects). These things are capable of interacting with each other, especially, for transferring information
in a network.6-8 They are largely deployed as low power and lossy networks (LLN). The LLN is a class of networks where

Trans Emerging Tel Tech. 2021;32:e4224. wileyonlinelibrary.com/journal/ett © 2021 John Wiley & Sons, Ltd. 1 of 20
https://doi.org/10.1002/ett.4224
2 of 20 UL HASSAN et al.

the interconnected devices are highly resource-constrained (ie, low in power, memory, and processing) and are charac-
terized by high loss rates, low data rates, and instability in the communication links.9 Many LLN-based IoT devices are
expected to communicate in three different ways: point-to-point, point-to-multipoint, and multipoint-to-point patterns.
Unfortunately, existing routing protocols are unable to deal with such traffic patterns effectively.10 Hence, Routing pro-
tocol for low power and lossy networks (RPL) routing protocol, a proactive and lightweight IPv6 network protocol, is the
solution to such scenarios. It is designed by Internet Engineering Task Force (IETF), which patterns reliable routes over
LLN wireless link to facilitate connectivity among IoT devices11,12 (Table 1).
The RPL adopts destination oriented directed acyclic graph (DODAG) for topology creation.13 It supports LLNs to
handle the routing among devices with high packet error rate and lossy links. The adoption of RPL-based LLN is growing
because it is adaptive to variable network topologies.14 However, RPL does not support strong security mechanisms, which
expose it to internal and external attacks.15 Various cryptography techniques, such as encryption and authorization, can

T A B L E 1 Table of abbreviations
Abbreviation Explanation

IoT Internet of Things

SDN Software defined networking
LLNs Low power and lossy networks
RPL Routing protocol for low power and lossy networks
IETF ROLL Internet Engineering Task Force for routing over low power and lossy networks
DODAG Destination oriented directed acyclic graph
DIO DODAG information object
DAO Destination advertisement object
DIS DODAG information solicitation
OF Objective function
MRHOF Minimum rank with hysteresis objective function
PER Packet error rate
IDC International Data Cooperation
DDoS Distributed denial of service
MRTS Metric-based RPL trustworthiness scheme
TPM Trusted platform module
AIM Automated metering infrastructure
QoS Quality of service
p2p peer to peer
PLR Packet loss rate
FD Forwarding delay
Pd Packet dropped
Pt Total packets
PRt Packet receive time
PFt Packet forwarding time
b belief
d disbelief
u uncertainty
p positivity
n negativity
TX Transmission ratio
UL HASSAN et al. 3 of 20

prevent only external attacks effectively.16 They are not helpful in case of internal attacks, where nodes are already part of
the network.17,18 Moreover, the IoT applications are resource-constrained, as afore-mentioned. They are not suitable for
generating cryptography keys to compute complex cryptography operations due to immense computation, energy, and
storage overhead.9 Thereby, to handle internal attacks, there exist solutions in the form of trust evaluation mechanisms.
These mechanisms are an efficacious addition to the conventional security mechanisms. These mechanisms are assertive
and contribute in many domains, such as reliable data fusion and mining, user privacy protection, information security
enhancement, secure data flow, and service assurance.19-21
As discussed later in Section 3, different trust-based security mechanisms have been proposed in the literature to
secure RPL-based networks from internal attacks. However, most of them ignore these networks’ resource-constrained
nature, which consequently drains the limited resources of IoT nodes. For instance, computation, storage, and energy.
On account of this, we conferred on a hierarchical trust-based mechanism “CTrust-RPL.” It encompasses three distinct
layers: a node, a sink, and a control layer. To conserve the node’s restrained resources, the behavior details are collected
from each node and are shipped to the control layer. Because of nodes’ constrained nature, all trust-related calculations
and analysis are performed in the control layer. It is also responsible for making decisions about the isolation of detected
malicious nodes. In this way, we transpose heavy computations to the control layer, letting IoT nodes to indulge in their
predefined tasks. Besides, it also conserves their limited energy and storage.
Below are the contributions of this research:

1. A novel control layer-based architecture to support secure routing in RPL-based IoT applications.
2. An energy-efficient trust mechanism for the detection and isolation of blackhole attacks.
3. A proof-of-concept of the proposed trust assessment mechanism along with some benchmark results.

1.1 Problem statement

According to International Data Cooperation, there will be 20 to 50 billion wireless connected IoT devices by 2020, gen-
erating a massive amount of sensitive data.22 With the emergence of 5G/6G networks, crucial and useful applications
are already provided for IoT networks. It ensures minimized delay and latency with more reliable, efficient, and secure
services.5 Therefore, this paradigm opens new directions for research, including security, interoperability, software devel-
opment, and so on.23 However, security definition has also been changed due to multiple entry points distributed in
such networks. Moreover, the attacks‘/attackers’ behavior has also been changed along with technological advancements,
which does not correlate with the existing attack phenomenon. The introduction of 5G/6G in IoT infrastructures has
brought new opportunities. However, it has also raised many concerns that cyber-criminals are taking advantage of Vimal
et al.24 For example, attackers can hack these IoT devices and use them for malicious purposes.25-27
RPL is a conventional routing protocol for IoT.28 It is also vulnerable to many internal attacks, such as blackhole and
sinkhole.2,29 Traditional security mechanisms, such as authorization, encryption, and cryptography, mitigate external
attacks only. Therefore, trust-based mechanisms are recommended to avoid such attacks.30 These mechanisms work on
the trust analysis of the network nodes. The detection is carried out by monitoring various trust parameters (for instance,
previous history and current underlying nodes) behavior. Once the malicious nodes are detected, the rest of the nodes use
it to secure their communication. However, the conventional trust evaluation mechanisms consume many resources due
to the message, computation, and communication overheads that consequently degrade the performance and shorten the
lifetime of resource-constrained IoT devices.31 Hence, there is a need to devise an energy-efficient trust-based mechanism
with low processing overhead, reduced delay, and minimum memory consumption. We propose an effective yet secure
mechanism to mitigate these issues. This article focuses on the potential internal attacks, specifically the blackhole attack
in RPL-based IoT applications.

1.2 Organization of article

The rest of the article is structured as follows: Section 2 presents the background of RPL and security issues. Section 3
presents related work. The proposed trust-based mechanism is explained in Section 4. Section 5 presents the experi-
mentation and simulation details followed by the performance evaluation and results. The article concludes the work in
Section 6.
4 of 20 UL HASSAN et al.

2 BAC KG RO U N D

This section first discusses the RPL concerning its working and then presents a discussion on its security issues.

2.1 Routing protocol for low power and lossy networks

For routing, the standard protocol in LLNs is RPL. On its initiation, a tree-like topology called a directed acyclic
graph (DAG) is created, and a parent node is selected by each node.30 The selected parent node works as a sink node or
a gateway for its corresponding child node. All information regarding DAG is managed as a DODAG graph. In DODAG,
each node has a path to its parent. The control messages, DODAG information object (DIO), DODAG information solicita-
tion (DIS), destination advertisement object (DAO), and DIO-Acknowledgement are used during DODAG construction.14
It is constructed step by step by using all these messages. First, the parent or sink node broadcasts DIO messages. It
helps nodes know about their parent node or the sink node. After getting a parent node, all nodes determine their ranks
according to an objective function (OF), such as hop count, energy reduction, and latency.
The OF defines the output of a process used by an RPL instance to establish routes in the whole network.2 According
to this, the RPL instances join the DODAG more than once. It is also responsible for node rank computation and in
determining the number of routers or sink nodes. The rank value of a node shows its distance from its sink or the parent
node. The child nodes in DODAG should always have a greater rank than the parents, which is exchanged in the whole
network to avoid loops.32 Nodes, in an IoT network, that has received DIO messages are replied by DAO message. By
contrast, nodes that are not a part of DODAG topology can join the topology by broadcasting DIS control messages. The
DODAG construction in RPL is further explained in Figure 1.

2.2 Trust and security in RPL

Many authors have discussed the security of the RPL protocol in different ways. Many authors have explained different
threat models for RPL, which describe all security-related threats and their analysis. For instance, in Reference 33, the
authors identified RPL threats with the help of the ISO 7498-2 security reference model. This model defines the secure
features of a communication channel such as integrity, authorization, authentication, access control, and encryption. It
also categorized hazardous RPL-based attacks along with data integrity and confidentiality issues. Moreover, the Internet
Engineering Task Force for routing over low power and lossy networks work-group presented a brief report on the secu-
rity of RPL. The report shows that the RPL resisted external attacks by using existing authentication and cryptography
mechanisms. However, it could not resist internal attacks where nodes may start behaving maliciously and may degrade
a network’s performance, such as more energy depletion and packets drop, and so on.
To handle internal attacks, trust-based security (aka collaborative or soft security) has gained focus in both IoT and
academia. The main purpose is to detect and isolate malicious nodes approved to participate in IoT applications based on

F I G U R E 1 DODAG construction: A, DIO message broadcast by root node, B, neighbor nodes participation in the DODAG, root/parent
selection, and broadcast of DIO messages, C, DODAG built after nodes’ involvement. DIO, DODAG information object; DODAG, destination
oriented directed acyclic graph
UL HASSAN et al. 5 of 20

their legitimate identities.17,34 Trust-based mechanisms provide apt security measures against such attacks.35 Trust can
be defined as a relationship between a trustor and the trustee.36 A trustor node relies on a trustee node in the network
to perform some tasks. For such reason, it examines the trustworthiness of the trustee to perform certain operations.
The trustworthiness of a node is predicted by observing its behavior towards its neighbors. It considers belief, reliability,
confidence level, and integrity.
This article focuses on malicious forwarding attacks, which can be mitigated by a trust-based mechanism that would
ensure communication reliability, correctness, and availability. Malicious forwarding attacks deteriorate network data
delivery ratio by dropping data packets instead of forwarding them to neighboring nodes. Thus, resending undelivered
packets result in more energy consumption. Examples of such attacks include blackhole and grayhole. The blackhole
attack is analogous to a big sucking hole, which sucks everything into it. In this attack, a malicious node drops all received
data packets instead of forwarding them. For example, in Figure 2, when node 4 and 5 send data packets to the root node
“R”. Node 1, being compromised, can drop all the packets it receives from 4 and 5instead of forwarding packets to R. In
the blackhole attack, the adversary joins in the data forwarding routes that may instigate distributed denial of service
adversaries to join in the active route.37 It may also impair the network lifetime and performance.3,38 Similarly, a selective
forwarding attack or grayhole attack also behaves as a blackhole attack. However, in these attacks, few packets are sent
forward to avoid detection. If a malicious node drops more packets and forwards less, a large portion of the RPL-network
could become unreachable and isolated. As illustrated in the blackhole zone in Figure 2, some active nodes are not part
of the RPL network because of intermediate malicious nodes. Therefore, this type of internal attack is a serious threat to
a network.39 Some internal attacks in RPL-based IoT applications are defined in Table 2.

FIGURE 2 A blackhole attack scenario

T A B L E 2 Primary internal
Attack Description
attacks in routing protocol for low
power and lossy networks-based Blackhole All received packets are dropped instead of forwarding to destination.
Internet of Things applications Grayhole Modified form of blackhole attack where some of the packets are forwarded, and
rest are dropped.
Wormhole A malicious node forms conspiring and covert paths with other nodes’ collusion
in the network with malign intentions.
Sinkhole The malicious node pretends to has an optimized path from source to
destination to get more traffic in the network.
Sybil The attacker node makes different fake identities for malicious activities.
Rank Node changes its rank to gain more traffic towards it.
6 of 20 UL HASSAN et al.

3 LITERATURE REVIEW

It is inevitable to proffer scalability and resilience to handle and communicate the massive amount of heteroge-
neous data, which is unfortunately not offered by traditional communication systems.40,41 It raises the need for
efficient and robust data communication methods while preserving QoS, lower computational cost, and security.
For instance, the data generated and communicated by a smart city and healthcare applications are highly sen-
sitive and must be protected against forgery.42-44 The risks and threats become hazardous if adequate security
measures are not ensured.45 In this section, different trust-based schemes are discussed with their respective pros
and cons.
In Reference 46, a metric-based RPL trustworthiness scheme is proposed to secure RPL by constructing a secure
routing path. However, it is inefficient because of an additional layer of communication and computation overhead
due to the extension of DIO control messages. Trusted platform module, used in this mechanism, is also a costly
and nonefficient module for computation. Since the control message flooding makes these mechanisms inefficient
for resource-constrained IoT devices, a crosslayer trust-based NUD mechanism is proposed without fixed intervals
in Reference 13. Besides, a secure parent node selection mechanism is presented in Reference 47, where every node
may choose a legal node as its “parent.” Particularly, a malicious parent node intends to have a false rank than a
legal parent node. In this work, the authors discussed the fact that the child nodes may select a malicious parent. In
addition, parameters to find parents are not adaptive due to the network’s fluctuating dynamics.30 Another authen-
tication mechanism, Automated Metering Infrastructure is known as AMI-based RPL, is proposed in Reference 48.
This article also explains the drawbacks of the AMI network caused by distance spoofing and sinkhole attacks. How-
ever, the proposed approach is vulnerable to compromised internal nodes and is inappropriate for battery-powered RPL
networks.
Another resilient and light-weight trust-based routing mechanism is proposed in Reference 49 for IoT. This mecha-
nism uses the behavior of nodes as a trust parameter. Accordingly, nodes are regarded as positive or negative based on
their behavior. This mechanism provides resilience against self-promoting, bad-mouthing, and ballot-stuffing attacks.
However, a single trust evaluator is a single-point-of-failure and can cause the whole network to go down. To improve
RPL routing protocol security (in the mobile network), a mechanism is proposed and evaluated in Reference 50. It is a
hierarchical trust model, which evaluates trust level by considering the Quality of Services (QoS), p2p secure commu-
nication, and appropriate information in the IoT network. These parameters calculate the trust level of an individual
node and the whole IoT network. The proposed mechanism’s effectiveness is compared with standard MRHOF OF14 to
show its efficacy with regards to end-to-end delay and packet loss. However, this mechanism is computationally com-
plex and can decrease the performance of the whole network. Moreover, it is not well-suited for resource-constrained
IoT devices. Another trust-based defense mechanism is proposed in Reference 14 to mitigate selective forwarding and
blackhole attacks in the RPL-based IoT network. However, this mechanism is prone to single-point-of-failure because
of one dedicated node in the network for trust aggregation. Furthermore, trust parameters are not scalable in large
networks. Besides, this proposed mechanism is not energy-efficient. Same authors proposed an improved version, as
a time-based trust-aware RPL (SecTrust-RPL),30 to secure RPL-based IoT network from insider attacks, particularly
rank and Sybil attacks. Each node in this mechanism calculates its adjacent nodes’ trustworthy behavior by comput-
ing direct and indirect trust values. However, the level of uncertainty of recommendation is not considered. Similarly,
for the isolation of trust-based attacks, such as wormhole and grayhole, a solution is presented in Reference 2. They
used indirect trust, which is calculated based on the neighbors. After calculating direct and indirect trust, total or com-
prehensive trust is calculated by a summation aggregation method. However, this mechanism does not consider the
uncertainty of neighbors’ opinions. Besides, trust parameters are not adaptable due to the fluctuating dynamics of the
network.30
In summary, most of the existing studies suffer from message flooding, complex computations, and memory over-
heads. It makes them in-efficacious in resource-constrained IoT environments. A summary of comparison among
different trust-based mechanisms is presented in Table 3 to secure RPL. Some parameters are not adaptive and scalable
due to the network’s fluctuating dynamics, which pose undesirable effects in battery-powered RPL-based networks. Con-
sequently, all these limitations deplete more energy at the node level and decrease the whole network’s performance.
Therefore, the selection of trust parameters and trust mechanisms is challenging in such resource-constrained, low power,
and lossy networks.
UL HASSAN et al. 7 of 20

T A B L E 3 Comparison matrix: Trust-based mechanisms to secure RPL

Article Year Mechanism Attack addressed Evaluation parameters Limitations

30 2019 SecTrust-RPL Rank and Sybil Packet transmission delay, Static topology, Not energy efficient,
Packet loss rate Single point of failure, Does not
consider the uncertainty of
recommendations
2 2018 Lightweight Wormhole and Throughput and packet loss Not adaptable due to changing
Trust-based Grayhole rate network dynamics Not scalable
mechanism
50 2018 DCTM-RPL Rank, Sybil and Packet loss rate, end-to-end Static topology No energy efficient
blackhole delay and average energy Single point of failure
consumption
14 2018 Trust-based RPL Blackhole and Throughput and percentage Trust parameters are not adoptable
Protocol Selective Forwarding of packet loss Static
topology
46 2017 Metric-based RPL Trust-based Issues ETX and ENRT metrics Additional layer of computation and
trustworthiness during routing communication overhead No
scheme trust-based attacks
49 2017 A trust-based approach Network level attacks No. of malicious nodes Centralized approach No specific
Average delivery ratio trust-based attacks addressed
13 2016 A crosslayer Improves network Packet transmission delay No specific attacks Trust parameters
Trust-based NUD reachability and Packet delivery rate are not adaptable
Mechanism availability
47 2015 A Trust-based Rank attack Avoidance rate and Nattacked Susceptible to attacks like Sybil and
threshold Blackhole Attacks
mechanism for node
evaluation
48 2015 Key encryption and Sinkhole and Distance Secure and unsecure network System susceptible to internal node
data mining spoofing Attacks and time sequence compromises; No Energy Efficient

Abbreviation: RPL, routing protocol for low power and lossy networks.

4 CTRUST-RPL S Y ST E M

This section discusses the proposed CTrust-RPL mechanism that supports analyzing the trust behavior of IoT nodes and
rectifying network management behaviors by detecting and isolating the malicious nodes, should suspicious signs be
detected.

4.1 System architecture

Figure 3 illustrates the proposed architecture, which consists of three layers: a device layer, a sink layer, and a control layer.
We presumed that the control and sink layers are always trusted and offer unlimited computation and energy capabilities
while analyzing the proposed mechanism.3,30 The functionality of all layers is elaborated in Figure 4. It shows an expanded
view of the data flow and trust-related data collection, calculation, propagation, and isolation of the malicious node steps.
The flow of trust-related data gathering, computation, propagation, and isolation of malicious nodes, concerning the
corresponding layers, is further discussed in detail as following:

Device Layer: In this layer, IoT devices, such as sensor nodes and actuators, are deployed as LLNs. These
devices/nodes are responsible for sensing, collecting, and preprocessing data gathered from the
surroundings and sending them to the root node. They are static and deployed randomly, considering
nodes’ homogeneity. Each node is connected to all neighboring nodes through symmetric links.30 For
routing and communicating the gathered data, devices rely on the RPL routing protocol. All devices
8 of 20 UL HASSAN et al.

FIGURE 3 System architecture of CTrust-RPL

are connected as well as with the border router using RPL. There is no malicious node at the initial
stage in the RPL-based IoT network. The two main processes of this layer include network
establishment and network discovery. They are being part of the network, nodes not only sense data
from surroundings but also from neighboring nodes. They also observe the neighboring node’s traffic
activity in their idle time. A node particularly listens to the neighboring node’s behavior to check if it
forwards the received packets to the destination correctly and timely. In this way, the node marks its
neighboring nodes as positive or negative according to their forwarding behaviors.49
Sink layer: This layer comprises sink nodes that forward/receive data to/from the control layer without
processing. More than one sink node in this layer is responsible for respective interconnected nodes
with different DODAGs.51 Multiple sink nodes enhance this layer’s bandwidth and decrease the
chances of packet drop sent from the device layer. These sink nodes are also connected to data
synchronization. After trust calculation, sink nodes receive trust values from the control layer and
communicate them back to the device layer. The main task of this layer is trust parameter propagation.
Control layer: The complete trust model is implemented in the control layer to calculate corresponding trust values
to minimize energy consumption and memory overhead. All steps of the proposed trust model, such
as trust calculation, aggregation, rating, and updating, are done in this layer. The main objective of
this layer is to handle the complex computations associated with the trust assessment. This layer also
supports the device layer to safely carry out all of its operations and enhance its capability in
communication. All the steps that are handled by the control layer are described briefly in the
preceding subsections.
UL HASSAN et al. 9 of 20

FIGURE 4 CTrust-RPL
framework

4.2 Trust model

This section encompasses the proposed trust model details. While establishing trust between entities, and the trust level,
the level of uncertainty should also be checked. This study, therefore, targets these aspects keeping the trustworthiness
of nodes and checking the uncertainty of the trust value. The trust model that considers uncertainty in terms of opin-
ion is the “subjective logic” trust model.52 Therefore, we preferred Subjective logic that involves the key elements of the
Dempster-Shafer belief theory53 while computing trust among entities. It shows the compatibility with probability cal-
culus and binary logic.54 A good trust-based model must be energy-efficient and must not strain IoT devices. Therefore,
this study shipped all the complex calculations to the control layer. We also chose adaptive trust parameters. Algorithm 1
illustrates the calculation of nodes’ trustworthiness. PLR and F D (discussed in more detail in subsequent sections) are the
input parameters, and the output is the malicious node detection and its removal from the topology. After initiation of
certain thresholds, the trust parameters are computed for each node in the network.
10 of 20 UL HASSAN et al.

Algorithm 1. Trust computation of all nodes

input: PLR , FD
Result: Malicious Node detected
1 Initialize t1 , t2 , p=0, n=0 and k=2
2 Set PLR and FD thresholds as t1 and t2 begin
3 Compute:
4 PLR = PD / PT
5 FD = PRt - PFt
6 for All Nodes do
7 if PLR < t1 and FD < t2 then
8 p=p+1
9 else
10 n=n+1
11 end
12 Calculate u, d and b as :
13 u =k/ (p + n + k)
14 d =n/ (p + n + k)
15 b =p/ (p + n + k)
16 if d > 0.5 then
17 N_status ← Not trusted
18 else if b > 0.5 then
19 N_status ← Trusted
20 else
21 N_status ← Not verified
22 end
23 if N_status = Trusted then
24 Select for routing
25 else
26 Remove from topology
27 end
28 end
29 end

The trust parameter computation is shown in lines 4 and 5, which is calculated in linear time. Based on these parame-
ters (as shown in line 12 (to 15)), the algorithm uses a subjective logic model to calculate the values of u, d, and b for each
node having linear time (details are given in following sections). These values are then compared with the set threshold
to determine the trust rating of the node(s), as shown on line 16 (to 22). It is again done in linear time. After rating, trust
propagation is being done from line 23 to 27. If the value of b is greater than the defined threshold (ie, 0.5), it is assumed
to be a trusted node, and that node is considered a legitimate part of the routing. From lines 6 to 28, the whole algorithm
computes trust, isolates malicious nodes, and disseminates trust value using a single “For” loop; hence, while trust com-
putation of all nodes, the complexity of the proposed algorithm is O(n). We also compared the complexity of proposed
algorithm with Reference 30. Both algorithms have the same complexity; however, ours outperformed the other in less
message overhead and energy depletion because of novel and adaptive parameters.

4.2.1 Trust calculation

As mentioned earlier that we used subjective logic52 trust model for calculating trust values (the discrete values between
0 and 1). These values are expressed by an opinion triangle, as shown in Figure 5.49 In fact, opinion is a three-dimensional
metric, which can be stated as follows: Let W = (b, d, u), express the opinion of node A about the trustworthiness of node
B in an IoT network. Where, b, d, and u represent belief, disbelief , and uncertainty, respectively. The sum of all these three
parameters is always equal to 1, as represented in Equation (1).
UL HASSAN et al. 11 of 20

FIGURE 5 Opinion triangle

b + d + u = 1. (1)

As per opinion definition, belief represents B’s trustworthiness probability, and disbelief expresses the probability that
node A cannot trust node B. In addition, uncertainty is used to relieve the gap in the absence of b and d.55

4.2.2 Relationship between trust parameters and opinion spaces

Many trust parameters evaluate the trust level of a node in the IoT network. However, computation and implementation
of the trust model depending on the chosen trust parameters. Nevertheless, the introduction of more trust parameters can
make the mechanism more complicated and complex. Therefore, the CTrust-RPL uses only two QoS parameters (“Packet
loss rate” and “forwarding delay”) for efficient and simple trust composition and calculation. Equations (2) and (3) are
used to calculate the QoS parameters. In the chosen QoS parameters, the “packet loss rate” (PLR ) is the ratio of packets
dropped (Pd ) by the recipient node to the total packets (Pt ) sent from the sender node. The “forwarding delay,” (F D ), is
the time interval between receiving a packet from the sender and then forwarding it to the next node. Packet received and
packet forwarding time is represented by (PRt ) and (PFt ), respectively.

Pd
PLR = , (2)
Pt

FD = PRt − PFt . (3)

On the basis of these two parameters, positive p and negative n, interactions are computed as depicted in Equations (4)
to (6). bij represents the degree of the belief of nodei have on nodej by taking the ratio of positive parameter with the
sum of all the three parameters; the positive, the negative, and the constant k. dij and uij are computed in the same
fashion.
p
bij = , (4)
p+n+k

n
dij = , (5)
p+n+k

k
uij = , (6)
p+n+k

where, k is a constant to simplify the computations. Its value is set as 1 or 2.56

12 of 20 UL HASSAN et al.

4.2.3 Trust aggregation

All IoT network nodes send p and n values of their neighbors to the control layer after specific time intervals. The control
layer aggregates all the gathered evaluations for a specific node to derive its global trust value. We used a special operator,
⨁
the consensus operator ( ) for aggregation, the same as used in Reference 56. It can be used as follows:
Suppose, V ij (bij , V ij , uij ) represents a trust value vector of nodei regarding nodej and V ik (bik , dik , uik ) expresses the
trust value vector of nodei with respect to nodek , then the aggregated trust value vector of j and k with respect to i is V i =
⨁
V ij V ik , which is computed as shown in Equations (7) and (8).
( )
bij ujk + bik uij dij ujk + djk uij uij ujk
, , , (7)
k k k

where,

k = uij + ujk + uij ujk . (8)

⨁
Therefore, due to commutative and associative properties, the consensus operator, , is used to aggregate the nodes’
overall values. The aggregated value represents the global trust level of a node. For instance, it represents the global
⨁
reputation of nodei as good if the aggregated trust using reflects greater trust value than the threshold set or vice versa.

4.2.4 Trust rating

The objective of the rating process is to discriminate against the trusted nodes from malicious ones. For a rating of IoT
nodes, the CTrust-RPL mechanism uses a threshold-based system. The nodes are rated on the values of belief, disbelief , and
uncertainty, which are calculated in the trust calculation process. We use a threshold of value 0.5 for all three parameters.
Nodes are ranked into two categories: trusted and not trusted. The trusted nodes are then used for secure communication
and routing. Table 4 shows the rating criteria.

4.2.5 Trust value propagation

The trust parameters are sent to the control layer for calculating trust values that are lately propagated to all nodes via
the sink. Trust propagation means how to transfer trust values to all the authentic sensor nodes in the network. Trust is
propagated either to the central node (physical server) or to all the sensor nodes distributively. In both ways, the trust value
is disseminated using flooding mechanics.14 The proposed mechanism relies on centralized trust propagation because
the trust values are calculated and propagated by the control layer.

4.2.6 Trust update

Generally, there are two methods to update the trust of the nodes in the RPL network, event-based and time-based update
method.57 For Instance, in event-based method, nodes send the trust parameters to the control layer via sink when nodes
observe an abnormal behavior based on PLR and F D . This method of updating is not very efficient as the malicious nodes

T A B L E 4 Rating of nodes
belief (b) disbelief (d) uncertainty (u) Action

>0.5 <0.5 <0.5 Trusted

<0.5 >0.5 <0.5 Not trusted
<0.5 <0.5 >0.5 Not verified
≤ 0.5 ≤ 0.5 ≤ 0.5 Not verified yet
UL HASSAN et al. 13 of 20

FIGURE 6 Sequence diagram of CTrust-RPL system

are not detected in time. Besides, this method has some other flaws, such as more computational and memory resources
are required. Furthermore, the trust values are also stored in a database for making further decisions. Whereas, the
time-based method covers most of these issues. The trust values are updated after a periodic time interval in the time-based
method. The CTrust-RPL mechanism is also based on the time-based trust update method.
A sequence of trust parameter gathering, trust calculation, trust propagation, and isolation of malicious node(s) is
detailed in Figure 6. The diagram represents all the steps performed by the nodes, border router, and the controller sequen-
tially. The nodes gather parameters and send them to the border router in the sink layer. The border router forwards the
parameters to the controller. After trust assessment in the control layer, the controller ships node Ids (Identifications) of
malicious nodes to the device layer via the sink.

5 EXPERIMENTAT ION A ND SIMULATION DETAILS

In this section, the evaluation and simulation details of the proposed mechanism are described. The performance of the
proposed mechanism is validated and evaluated by simulations. We purposely choose the same environment and simu-
lation parameters as in Reference 30 for the comparison and validation. We use Cooja 2.7 for simulating the CTrust-RPL
environment. The simulation is done on Intel(R) 2.54 GHz CPU with 1 GB RAM using Linux platform. Our mechanism
is based on multipoint-to-point traffic flow, and the operation mode of RPL is set to No Downward. In this mode, sensor
nodes forward packets to the sink node, and the sink node forwards this data to the server for calculating the trust values.
As mentioned earlier, the IoT devices compose LLN networks, which are lossy. Therefore, the transmission ratio (TX) is
set to 100% for a lost-free transmission, and the reception (RX) ratio is set from 30% to 100%, which shows that the loss
is at the reception layer and not at the transmission. However, the reception and the transmission for the attacker nodes
are set to 100% with an interference range of 55 m.
14 of 20 UL HASSAN et al.

A screenshot of Contiki/Cooja of the proposed scheme topology is shown in Figure 7, where node 33 represents the
controller. Node 31 and node 32 represent the sink nodes, node 28 to 30 are malicious nodes, and the rest are the normal
nodes. Table 5 highlights the details of simulation parameters, along with the values used in the simulation environment.
As mentioned earlier, we use the same parameters (such as simulation coverage area, RX, TX, and TX range) in the
simulation, as in Reference 30, to preserve comparison consistency. We simulate all the three mechanisms using these
parameters and then compare and validate the results, as shown in Section 5.1. To validate our mechanism, we use packet
loss ratio, attack detection, residual energy, attack detection time, and power consumption parameters. Attack detection

FIGURE 7 Topology screenshot of

Contiki/Cooja

T A B L E 5 Simulation parameters and

Simulation parameter Value
their values
Simulation tool Contiki/Cooja 2.7
Simulation coverage area 70 m * 70 m
Total number of nodes 30
Malicious nodes 3 (Nodes 28, 29, and 30)
Malicious to legitimate node ratio 1:10
RX ratio 30% to 100%
TX ratio 100%
TX range 50 m
Interference range 55 m
Routing protocols MRHOF-RPL, SecTrust-RPL, and CTrust-RPL
Network protocol IP based
Simulation time 60 minutes
UL HASSAN et al. 15 of 20

means accurate detection of blackhole attack in the network at any given time. The packet loss ratio defines the packet
dropped rate by the (attacker) nodes in the network. Residual energy is the remaining energy of each node in the network
at a particular time when the trust is calculated. And the last one; power consumption is the consumption of energy of
each node while performing any activity (malicious or nonmalicious) during the span of experimentation.

5.1 Performance evaluation and results

This section illustrates attack implementation, its detection, packet loss ratio, and the power consumption during
evaluation. We compared CTrust-RPL technique with two different mechanisms: SecTrust-RPL30 and MRHOF-RPL.58
MRHOF-RPL uses an OF that selects routes, which minimizes a metric. It works with additive metrics that the RPL DIO
messages advertise. It is a default routing OF of RPL protocol. However, it does not support blackhole detection. There-
fore, we used only the packet loss ratio and the energy consumption for comparison. Whereas blackhole detection, packet
loss ratio, and power consumption (all three) parameters were used for the comparison with Reference 30. The results
obtained show better outcomes relative to the selected parameters.

5.2 Blackhole attack implementation

During the simulation, a malicious node (blackhole) initially shows good behavior for about 5 to 10 seconds (tends to
be a trustworthy path) and starts the malicious behavior soon after. It receives data packets from the adjacent nodes and
drops all the packets without forwarding them to the destined nodes. It significantly degrades an RPL-based network and
depletes nodes’ power since the nodes continue sending data packets to the malicious nodes.

5.3 Blackhole detection

Figure 8 shows that CTrust-RPL technique successfully detected and isolated the blackhole attacks during routing. To
check its efficiency, we compared the results with SecTrust-RPL,30 which also detected and mitigated these attacks suc-
cessfully. However, its detection ratio is lower than that of ours. It also failed to detect all the attacks, as portrayed in
Figure 8. The figure shows that both mechanisms detected many malicious nodes initially since there were more mali-
cious entities. Once the trust was fully enacted, the number of malicious nodes decreased significantly. It happened due
to the proactive nature of the RPL, where removal of the adversary nodes from its topology (DODAG) is done successfully,
and new routes are established before the network gets exhausted. For instance, at time 1, the number of malicious nodes

FIGURE 8 Attack detection

16 of 20 UL HASSAN et al.

detected for CTrust-RPL was 380, and it was 340 for Reference 30. On time 29, the number of malicious nodes detected,
reduced to 60 and 40 for CTrust-RPL and SecTrust-RPL, respectively. In the lapse of 30 minutes, CTrust-RPL detected 221
malicious nodes on average (per minute), whereas Reference 30 detected only 191 malicious nodes in the same period.

5.4 Attack detection time

In order to check the time efficiency of the proposed mechanism, we analyzed the time taken by both the mechanisms
(the SecTrust-RPL and the CTrust-RPL) in detecting malicious nodes. Figure 9 depicts the attack detection time. We
started with one malicious node initially and then increased their numbers chronologically. The detection time of first
malicious node in CTrust-RPL is 0.34 second and it is 0.8 second for SecTrust-RPL. In the figure, we can see that after
second node detection, the time taken by both the mechanisms kept increasing. It is since time is proportional to the
number of attacks. With an increase in malicious node, the time also increases. However, it may be seen that the proposed
CTrust-RPL took less time (to detect) as compared with SecTrust-RPL. The time to detect 31 malicious nodes is 40 and
47 seconds for CTrust-RPL and SecTrust-RPL, respectively. The average time to detect 31 malicious nodes for CTrust-RPL
is 21.04 seconds. Whereas, SecTrust-RPL detected the same number of malicious nodes with an average time of 25.53
seconds. The difference in average detection time shows that the proposed scheme is more efficient than state-of-the-art.

5.5 Packet loss ratio

CTrust-RPL detected a blackhole attack (malicious node) timely and then mitigated malicious nodes within the network.
This way, it decreased undue overhead because of the packet loss rate caused by a blackhole attack. Figure 10 elaborates
packet loss rate to examine how CTrust-RPL mechanism performed better than References 30 and 58. The packet loss
ratio of the mechanism is 10% less than that of in Reference 30 even under the same network parameters. Somehow few
patterns of both the mechanisms seem to be neutral due to the same network parameters used. On average, the packet
loss ratio for CTrust-RPL was 0.38, for Reference 30 it was 0.48 and it was 0.61 for Reference 58. Thus, our technique
provides better defense against blackhole attacks with the lowest packet loss ratio.

5.6 Energy consumption

Figure 11 shows that the proposed trust-based mechanism for LLNs is 35% more energy-efficient when compared with
References 30 and 58. The average power consumption (mW) used is 0.32, 0.82, and 1.60 for CTrust-RPL,30,58 respectively.
The power consumption for Reference 58 is higher than the other two because the OF does not use any attack mitigation
mechanism to handle increased packet drops by malicious nodes in the network. Whereas, the Reference 30 consumed
less power than the Reference 58, as it has the mechanism to handle malicious nodes. However, the Reference 30 still

FIGURE 9 Attack detection time

UL HASSAN et al. 17 of 20

F I G U R E 10 Packet loss ratio

F I G U R E 11 Energy consumption

consumes higher power than the CTrust-RPL. It is so that without the sink and the control layer, the energy consumption
of IoT devices is very high, as the trust mechanism is running at the node level, and a resource-constrained node itself
does all computations. Due to this processing, the victim nodes’ lifetime gets pared down, and the network falls after the
depleted energy of the victim nodes. Hence, the proposed approach performed well to utilize network resources efficiently.

5.7 Average residual energy

Figure 12 illustrates the average residual energy of nodes, saved during the simulation while malicious node detection. It
represents the efficiency of References 30,58, and the proposed approach in terms of saved energy. The results show that
nodes have 40% more residual energy saved with our mechanism than that of Reference 30 and 45% more than Reference
58. The figure shows that at time 1 (first minute), nodes’ average residual energy for all three competing mechanisms was
initially set to 1 mJ. With the course of the simulation, the energy (of nodes) started to deplete. At time 29 (29th minute),
as shown in the figure, a drastic energy depletion can be viewed in References 30 and 58, whereas, for our scheme, this
depletion remained gradual. The residual energy was 0.72, 0.3, and 0.2 mJ for CTrust-RPL,30,58 respectively. The reason
for having more residual energy, in our case, is that all processes of our trust model are running on the controller in the
18 of 20 UL HASSAN et al.

F I G U R E 12 Average residual
energy

control layer, and the IoT nodes are not involved in this process. The controller manages the energy overhead of all nodes.
The nodes are responsible for gathering parameters about their neighbors only and then sending them to the controller
via the sink node.

6 CO N C LU S I O N

The myriad of IoT devices has captured the Internet with a massive amount of sensitive data. These devices are intercon-
nected as LLNs and use RPL as a standardized routing protocol. Unfortunately, these devices are vulnerable to several
routing, security, and privacy threats. Although RPL provides standardized routing, it does not deliver good protection
against a range of routing attacks. It could result in devastating and damaging consequences in such networks. Therefore,
we proposed CTrust-RPL: an energy-efficient, control layer-based trust mechanism in the RPL routing protocol for the
detection and isolation of blackhole attacks. In order to conserve the energy of the constrained IoT devices, the trust val-
ues were computed by a control layer. It was done after observing packet exchange between nodes, which was then used
to detect and to isolate malicious nodes from the network. The simulation results showed that the proposed system was
more efficient than the Sec-trustand MRHOF-RPL in terms of low processing and storage overhead, less packet drop ratio,
and enhanced the lifetime of the network while mitigating blackhole attacks. CTrust-RPLdetected and isolated 10% more
malicious nodes than Sec-trust in the same time-lapse. The average packet loss ratio difference was less for our proposed
mechanism, with 35% more energy efficiency.
Besides energy preservation, scalability and decentralization are two important factors in designing security mech-
anisms for massively interconnected IoT devices deployed distributively. These two factors play quite a vital role in
maximized network lifetime and minimized network congestion and latency. Thus, the need to devise decentralized and
scalable security solutions cannot be ignored. Although, the proposed trust mechanism supports the same (huge) num-
ber of interconnected nodes/devices, as does the RPL. However, it may also confront the issues as mentioned above. In
the future, we plan to design a distributed and more scalable trust-based mechanism to meet the growing needs of dis-
tributive IoT deployments and tackle other RPL attacks, such as Sybil, rank, and selective forwarding attacks along with
the blackhole. We would also explore the vulnerabilities related to trust model.

DATA AVAILABILITY STATEMENT

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study

ORCID
Muhammad Asim https://orcid.org/0000-0002-2894-7891
UL HASSAN et al. 19 of 20

REFERENCES
1. Almusaylim ZA, Alhumam A, Mansoor W, Chatterjee P, Jhanjhi NZ. Detection and mitigation of rpl rank and version number attacks in
smart internet of things; 2020.
2. Mehta R, Parmar M. Trust based mechanism for securing iot routing protocol rpl against wormhole & grayhole attacks. Paper presented
at: Proceedings of the 2018 3rd International Conference for Convergence in Technology (I2CT), Pune, India; 2018:1-6.
3. Tariq N, Asim M, Maamar Z, Farooqi MZ, Faci N, Baker T. A mobile code-driven trust mechanism for detecting internal attacks in sensor
node-powered iot. J Parall Distrib Comput. 2019;134:198-206.
4. Wang M, Zhu T, Zhang T, Zhang J, Yu S, Zhou W. Security and privacy in 6g networks: new areas and new challenges. Digital Commun
Netw. 2020;6:281-291.
5. Giordani M, Polese M, Mezzavilla M, Rangan S, Zorzi M. Toward 6g networks: use cases and technologies. IEEE Commun Mag.
2020;58(3):55-61.
6. Aziz MF, Khan AN, Shuja J, Khan IA, Khan FG. A lightweight and compromise-resilient authentication scheme for iots. Trans Emerg
Telecommun Technol. 2019;e3813.
7. Kołodziej J, Jaatun MG, Khan SU, Koeppen M. security-aware and data intensive low-cost mobile systems editorial. Mob Netw Appl.
2013;18(5):591-593.
8. Subbulakshmi P, Vimal S. Secure data packet transmission in manet using enhanced identity-based cryptography (eibc). Int J New Technol
Sci Eng. 2016;3(12):35-42.
9. Abbas N, Asim M, Tariq N, Baker T, Abbas S. A mechanism for securing iot-enabled applications at the fog layer. J Sens Actuat Netw.
2019;8(1):16.
10. Mayzaud A. Monitoring and Security for the Rpl-Based Internet of Things [Ph.D. thesis]; 2016.
11. Agustin JPC, Jacinto JH, Limjoco WJR, Pedrasa JRI. Ipv6 routing protocol for low-power and lossy networks implementation in net-
work simulator–3. Paper presented at: Proceedings of the 2017 IEEE Region 10 Conference on TENCON 2017, Penang, Malaysia;
2017:3129-3134.
12. Hassan QF, Madani SA. Internet of Things: Challenges, Advances, and Applications. Hoboken, NJ: Chapman & Hall/CRC Press; 2017.
13. Guclu SO, Ozcelebi T, Lukkien JJ. Trust-based neighbor unreachability detection for rpl. Paper presented at: Proceedings of the 2016 25th
International Conference on Computer Communication and Networks (ICCCN), Waikoloa, HI; 2016:1-6.
14. Airehrour D, Gutierrez J, Ray SK, et al. A trust-based defence scheme for mitigating blackhole and selective forwarding attacks in the rpl
routing protocol. Austr J Telecommun Dig Economy. 2018;6(1):41.
15. Boudouaia MA, Ali-Pacha A, Abouaissa A, Lorenz P. Security against rank attack in RPL protocol. IEEE Netw. 2020;34(4):133-139.
16. Li J, Tang X, Wei Z, Wang Y, Chen W, Tan Y-a. Identity-based multi-recipient public key encryption scheme and its application in iot. Mob
Netw Appl. 2019;1-8.
17. Tariq N, Asim M, Al-Obeidat F, et al. The security of big data in fog-enabled iot applications including blockchain: a survey. Sensors.
2019;19(8):1788.
18. Al-khafajiy M, Baker T, Asim M, et al. Comitment: a fog computing trust management approach. J Parall Distrib Comput. 2020;137:1-16.
19. Zhou J, Cao Z, Dong X, Vasilakos AV. Security and privacy for cloud-based iot: challenges. IEEE Commun Mag. 2017;55(1):26-33.
20. Kim H-W, Jeong Y-S. Secure authentication-management human-centric scheme for trusting personal resource information on mobile
cloud computing with blockchain. Human-Centric Comput Inf Sci. 2018;8 (1):11.
21. Khurshid A, Khan AN, Khan FG, Ali M, Shuja J, Khan AUR. Secure-camflow: a device-oriented security model to assist information flow
control systems in cloud environments for iots. Concurr Comput Pract Exper. 2019;31(8):e4729.
22. Federal Trade Commission. Internet of Things: Privacy & Security in a Connected World. Washington, DC: Federal Trade Commission. 2015.
23. Tariq N, Qamar A, Asim M, Khan FA. Blockchain and smart healthcare security: a survey. Proc Comput Sci. 2020;175:615-620.
24. Vimal S, Kalaivani L, Kaliappan M. Collaborative approach on mitigating spectrum sensing data hijack attack and dynamic spectrum
allocation based on casg modeling in wireless cognitive radio networks. Clust Comput. 2019;22(5):10491-10501.
25. Baker T, Asim M, MacDermott Á, et al. A secure fog-based platform for scada-based iot critical infrastructure. Softw Pract Exper.
2020;50:503-518.
26. Noura M, Atiquzzaman M, Gaedke M. Interoperability in internet of things: taxonomies and open challenges. Mob Netw Appl.
2019;24(3):796-809.
27. Vimal S, Kalaivani L, Kaliappan M, Suresh A, Gao X-Z, Varatharajan R. Development of secured data transmission using machine
learning-based discrete-time partially observed Markov model and energy optimization in cognitive radio networks. Neural Comput Appl.
2020;32(1):151-161.
28. Winter T, Thubert P, Brandt A, et al. Rpl: Ipv6 routing protocol for low-power and lossy networks, Technical report; 2012.
29. Yadollahzadeh Tabari M, Mataji Z. Detecting sinkhole attack in rpl-based internet of things routing protocol. J AI Data Mining. 2020.
30. Airehrour D, Gutierrez JA, Ray SK. Sectrust-rpl: a secure trust-aware rpl routing protocol for internet of things. Future Generat Comput
Syst. 2019;93:860-876.
31. Khan R, Khan SU, Zaheer R, Khan S. Future internet: the internet of things architecture, possible applications and key challenges.
Paper presented at: Proceedings of the 2012 10th International Conference on Frontiers of Information Technology, Islamabad, Pakistan;
2012:257-260.
32. Kim H-S, Ko J, Culler DE, Paek J. Challenging the ipv6 routing protocol for low-power and lossy networks (rpl): a survey. IEEE Commun
Surv Tutor. 2017;19(4):2502-2525.
20 of 20 UL HASSAN et al.

33. Tsao T, Alexander R, Dohler M, Daza V, Lozano A, Richardson M. A security threat analysis for the routing protocol for low-power and
lossy networks (rpls), RFC7416; 2015:131.
34. Baker T, Rana OF, Calinescu R, Tolosana-Calasanz R, Bañares JÁ. Towards autonomic cloud services engineering via intention workflow
model. Paper presented at: Proceedings of the International Conference on Grid Economics and Business Models; 2013:212-227; Springer,
New York, NY.
35. Chen H-C. Collaboration iot-based rbac with trust evaluation algorithm model for massive iot integrated application. Mob Netw Appl.
2019;24(3):839-852.
36. Yan Z, Zhang P, Vasilakos AV. A survey on trust management for internet of things. J Netw Comput Appl. 2014;42:120-134.
37. Reddy KG, Thilagam PS. Na𝚤 ̈ ve bayes classifier to mitigate the ddos attacks severity in ad-hoc networks. Int J Commun Netw Inf Secur.
2020;12(2):221-226.
38. Tariq N, Asim M, Khan FA, Baker T, Khalid U, Derhab A. A blockchain-based multi-mobile code-driven trust mechanism for detecting
internal attacks in internet of things. Sensors. 2021;20(1):23.
39. Kamgueu PO, Nataf E, Ndie TD. Survey on rpl enhancements: a focus on topology, security and mobility. Comput Commun. 2018;120:10-21.
40. Al-Turjman F. Intelligence and security in big 5g-oriented IONT: an overview. Future Generat Comput Syst. 2020;102:357-368.
41. Al-Turjman F, Mostarda L. A hash-based rfid authentication mechanism for context-aware management in iot-based multimedia systems.
Sensors. 2019;19(18):3821.
42. Al-Turjman F, Zahmatkesh H, Shahroze R. An overview of security and privacy in smart cities’ iot communications. Trans Emerg
Telecommun Technol. 2019;e3677
43. Al-Turjman F, Alturjman S. Context-sensitive access in industrial internet of things (iiot) healthcare applications. IEEE Trans Ind Inform.
2018;14(6):2736-2744.
44. Al-Khafajiy M, Webster L, Baker T, Waraich A. Towards fog driven iot healthcare: challenges and framework of fog computing in health-
care. Paper presented at: Proceedings of the 2nd International Conference on Future Networks and Distributed Systems, Amman, Jordan;
2018:1-7.
45. Alabady SA, Al-Turjman F, Din S. A novel security model for cooperative virtual networks in the iot era. Int J Parall Program. 2018;48:1-16.
46. Djedjig N, Tandjaoui D, Medjek F, Romdhani I. New trust metric for the rpl routing protocol. Paper presented at: Proceedings of the 2017
8th International Conference on Information and Communication Systems (ICICS), Irbid, Jordan; 2017:328-335.
47. Iuchi K, Matsunaga T, Toyoda K, Sasase I. Secure parent node selection scheme in route construction to exclude attacking nodes from
rpl network. Paper presented at: Proceedings of the 2015 21st Asia-Pacific Conference on Communications (APCC), Kyoto, Japan;
2015:299-303.
48. Taylor C, Johnson T. Strong authentication countermeasures using dynamic keying for sinkhole and distance spoofing attacks in smart
grid networks. Paper presented at: Proceedings of the 2015 IEEE Wireless Communications and Networking Conference (WCNC), New
Orleans, LA; 2015:1835-1840.
49. Khan ZA, Ullrich J, Voyiatzis AG, Herrmann P. A trust-based resilient routing mechanism for the internet of things. Paper presented at:
Proceedings of the 12th International Conference on Availability, Reliability and Security; 2017:27; ACM, New York, NY.
50. Hashemi SY, Aliee FS. Dynamic and comprehensive trust model for iot and its integration into rpl. J Supercomput. 2018;75:1-30.
51. Vasseur JP, Kim M, Pister K, Dejean N, Barthel D. Routing metrics used for path calculation in low-power and lossy networks, Technical
report; 2012.
52. Jøsang A. A logic for uncertain probabilities. Int J Uncertain Fuzz Knowl Based Syst. 2001;9(03):279-311.
53. Yager RR, Liu L. Classic Works of the Dempster-Shafer Theory of Belief Functions. Vol 219. New York, NY: Springer; 2008.
54. Jøsang A. Probabilistic logic under uncertainty. Paper presented at: Proceedings of the 13th Australasian Symposium on Theory of
Computing-Volume 65, Australian Computer Society, Inc; 2007:101-110.
55. Li X, Lyu MR, Liu J. A trust model based routing protocol for secure ad hoc networks. Paper presented at: Proceedings of the 2004 IEEE
Aerospace Conferenc (IEEE Cat. No. 04TH8720), Big Sky, MT: IEEE; Vol. 2; 2004:1286-1295.
56. Jøsang A. Subjective Logic. New York, NY: Springer; 2016.
57. Wang T, Zhang G, Bhuiyan MZA, Liu A, Jia W, Xie M. A novel trust mechanism based on fog computing in sensor–cloud system. Future
Generat Comput Syst. 109:573-582.
58. Gnawali O, Levis P. The minimum rank with hysteresis objective function. RFC 6719; 2012.

How to cite this article: ul Hassan T, Asim M, Baker T, Hassan J, Tariq N. CTrust-RPL: A control layer-based
trust mechanism for supporting secure routing in routing protocol for low power and lossy networks-based
Internet of Things applications. Trans Emerging Tel Tech. 2021;32:e4224. https://doi.org/10.1002/ett.4224