Journal of Electrical Engineering, Vol. 74, No. 6, 2023, pp.

454-462

sciendo
PAPERS__________________________________________________________________________________________________________________

Implementing and evaluating a new Silent Rank Attack

in RPL-Contiki based IoT networks

Mehdi Rouissat1,2,*, Mohammed Belkheir3, Hichem S. A. Belkhira1,

Allel Mokaddem3, Djamila Ziani3

IoT networks are witnessing a rapid growth in various domains of our daily life, offering more attractive features in terms of
measurement accuracy, easy implementation and affordable deployment costs. This outstanding boom is not undoubtedly far
away from different challenging issues that impede the network efficiency and quality. The security concern remains one among
the prominent issues that affect both the edge and the core IoT network where risks increase in conjunction with the network
expansion. RPL is the well-known routing protocol for the edge part of the IoT network, intended to meet the requirements of
the constrained IoT devices. Despite its various advantages, RPL remains suffering from various security attacks targeting the
topology, the traffic, and the nodes resources. Our work presents a new silent decreased rank attack against RPL-Contiki, as
well as a lightweight countermeasure. The obtained results on a random studied topology show that almost half the existing
nodes in the topology were attracted by the planted malicious node, through its falsified low rank. Moreover, an increase of
12.5% in the control overhead and an increase of 15% in the total consumed energy are recorded compared to the attack-free
topology. On the other hand, the attack did not heavily affect the PDR, but the latency showed an increase of 45% compared
to the attack free case. This damaging effect makes this modified rank attack a serious threat to IoT RPL based networks.

Keywords: IoT, RPL, Contiki, security, rank

1 Introduction by artificial intelligence (AI) techniques that aim to

resolve the inefficiency of the manual traditional
The world is witnessing a huge transformation in
systems and bring an added value for more reliable and
recent times with the advent of new technologies based
predictable processes with fast and accurate real-time
on smart devices called sensors. These devices build
decisions. The purpose is to reduce downtimes and to
together to form a new kind of self-organized smart
provide more trustworthy solutions. This is witnessed by
networks named IoTs, very suitable to thoroughly
the better insights brought by AI-IOT enabled solutions,
collect real-time data from a monitored environment for
especially for detecting infections of COVID-19
further processing and decision. This revolutionized
pandemic and helping to stop the massive spread of the
technology that transforms our daily life digital, offers
pandemic in many countries. The pairing of the two
numerous features such as: easy deployment, scalability,
aforesaid technologies has given rise to a new trend of
measurement accuracy, affordable installation cost, high
an inspiring digitization model called “digital twin”
salability, and many other advantages that allowed it to
which is considered as one among the current interesting
grow rapidly from the research stage to the industry
research topics.
field. For that purpose, IoTs networks have been widely
involved in many sensitive domains such as smart cities, Despite the enormous growth of IoT networks and
smart industry 4.0, smart healthcare, smart education, their wide range of applications, they remain suffering
smart agriculture, smart oil/gas fields, smart transporta- from various weaknesses that should be addressed, such
tion, and other various monitored environments. In as QoS, energy efficiency, wireless networking issues
addition, and for more benefits, IoT networks are often and security challenges [1-4]. This is mainly due to the
connected through internet to cloud platforms that intend limited characteristics of the IoT devices in terms of
to store the huge amount of the gathered data for more energy saving, processing, and storage memory,
efficiency in terms of analysis and decision making. knowing that sensors are considered as constrained
Furthermore, IoT networks are more and more federated devices and run a lightweight unsecured operating
_______________________
1University Center Nour Bachir El-Bayadh, 32000, El-Bayadh, Algeria
2 STIC Laboratory, University Aboubekr Belkaid, Tlemcen, Algeria
3 LIMA Laboratory , Univeristy Center Nour Bachir, El-Bayadh, Algeria

* [email protected]

https://doi.org/10.2478/jee-2023-0053, Print (till 2015) ISSN 1335-3632, On-line ISSN 1339-309X

© This is an open access article licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives License
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
 Journal of Electrical Engineering, Vol. 74, No. 6, 2023 455

system [5, 6]. Moreover, the upsurge number of the it includes signature rules to detect any malicious
connected devices abroad a large spectrum of areas leads behavior in the network. In another work, [19] proposes
IoT networks to face various challenges in different a secured version of the basic RPL protocol named
scales of the network (edge, cloud, fog, etc) [7, 8]. SRPL-RP, where a timestamp and a threshold are
Statistics in [9] reveal a gigantic evolution in the IoT included to detect the legitimacy of a DIO sender. Each
market where 7 billion of IoT devices are interconnected node is identified by an ID and the root node ID is
over the world with 400 platforms. This number is encrypted to avoid its imitation, in addition, a moni-
expected to reach 25 billion by 2030. According to [10], toring table is established in conjunction with the
cyberattacks in IoT reached 1 billion in 2021, since the DODAG building process, contains nodes parameters
third of the connected devices are infected, where (ID, Rank) that help the receiving node to detect further
phishing attacks and DDoS are the most dominant. changing of nodes behaviors, each node sets a blacklist
Given this statistics, the security concern is still table to avoid any new rooting rules coming from an
a challenging research topic. intruder.
Our present work focuses on securing the edge part All the proposed solutions in the literature to thwart
of the IoT network, especially the routing stack. As the rank attack have demonstrated improvements in
known, IoT constrained devices are considered as LLNs different studied metrics compared to the basic RPL.
(Low power and Lossy Networks) and run a networking However, it is strongly recommended to develop
protocol named RPL (Routing Protocol for LLNs) lightweight solutions which meet the constrained
described in the IETF RFC 6550 [11]. Despite, the latter properties of the IoT network in terms of power saving,
seems very suitable for saving the intrinsic properties of computational and storage limitations of the constrained
the IoT devices; it suffers from a wide range of devices. Heavy cryptographic solutions are not suitable
vulnerabilities widely discussed and categorized by for the edge network components and rapidly exhaust
researchers. These attacks target both the topology and the network resources leading to reduce its lifetime.
network resources [12, 13].
The effects of rank attacks are: (i) routing loops,
In RPL nodes are organized in a tree topology, where (ii) unoptimized route formation, (iii) decreased packet
the root node (gateway) gains the governing position, delivery ratio, (iv) increased delay and (v) increased
and it is responsible for applying and spreading RPL transmission of DIO messages, lead to more
routing rules over the network. The other IoT nodes are consumption of resources [20-23]. In this paper we
positioned in the network according to their capabilities present and analyze a new special silent decreased rank
expressed by a value called “rank”, which is clearly attack, where an intruder advertises a better fake rank
explained in the following sections. A malicious node value than its parent, without prompting any loops,
can take part of the network and advertises a fake rank making the attack silent and hard to detect. At first, we
value to its neighborhood aiming to attract other nodes clearly describe the harmful defeat caused by this attack
to relay their data through it. This attack has been widely by simulations using Cooja under Contiki 3.0.
discussed by several recent researches, since RPL does Thereafter, we explain our countermeasure approach
not have any countermeasure to deny this harmful action and its better insight allowing detecting this potential
triggered by an intruder. In the literature various attack.
researches have addresses the rank attack, where in [14]
authors presented a study using Friedman test to
compare recent researches related to the proposed 2 RPL protocol and rank value
mitigation solutions against the rank attack. In [15], The core functioning of RPL is defined in the RFC
authors proposed a new objective function named 6550 [11], it consists of a tree-based topology named
EMBOF for Echelon Metric Based Objective Function DODAG (Direct Oriented Destination Acyclic Graph)
which involves an echelon value triggered and as illustrated in Fig. 1.
exchanged between the root and the corresponding
parent node, by which a faked rank value is detected. The root node is responsible for building the topology
Authors in [16] proposed an updated version of the basic and disseminating the routing rules over the entire
RPL protocol named MFO-RPL for Moth-Flame, they network via control messages. The other nodes are
implement a petal algorithm for the parent selection positioned according to their rank value defined by an
process and the rank attack is detected using Moth- objective function. The root node has the lower rank
Flame algorithm to prevent any falsified rank value. value which increases downward the DODAG.
Another work in [17] consists on a security system called RPL defines four control messages, where DIS
SRF-IoT based on an external intrusion detection system (DODAG Information Solicitation) message is sent by
to prevent any illegitimate nodes to intrude the network. a node to request joining the DODAG, it is sent by
In [18], authors proposed an IDS system called a node when no DIO message is received within a time
DETONAR for DETection of rOuting Attacks in RPL, interval (5 s is the default value in RPL Contiki) [24].
 456 Mehdi Rouissat et al.: A lightweight countermeasure solution against a new Silent Rank Attack n RPL-Contiki based IoT networks

DIO (DODAG Information Object) message is other harmful behavior such as Blackhole, targeting
responsible to broadcast DODAG parameters, this thereby the network traffic, topology or the nodes
includes the Instance ID, DODAG ID, and Version resources.
Number (VN), which are used for DODAG
identification and topological update tracking [25].
DAO (Destination Advertisement Object) message is 3 Results and discussion
used for building downward routes from the root node. In this section, simulated results and overall analysis
Finally, DAO-Ack, an optional message used to of the obtained results are discussed. At the beginning,
acknowledge a DAO message. we simulated a basic scenario (attack free case), used as
a reference to judge the effect of the attack on the
performance of the network. The studied topology is
shown in Fig. 2, it consists of 1 sink node, 23 fair nodes
and a single malicious node (node "25") placed two hops
from the sink node.

Fig. 1. RPL topology and control messages

In order to select the optimum path to the root, a node

uses the objective function to calculate and translate one
or more metrics into rank value, based on its parent's,
and it advertises it in its DIO messages. Based on the
different received rank values, a node chooses the node
advertising the lower rank value as its preferred parent,
a gate toward the root. Two objective functions are
defined in RPL standards:
• Objective Function Zero (OF0) [26], based on which
the rank is based mainly on the number of hops
Fig. 2. Studied network topology
toward the root. The goal is to minimize the number
of hops from nodes to the sink
The UDP client-server model is used with Zolertia Z1
• Minimum Rank with Hysteresis Objective Function motes. The reason behind choosing Z1 motes is that
(MRHOF): is the default O.F in the used Contiki O.S, Contiki RPL mote Sky engenders bugs during the
the rank is also based the link's quality; ETX compilation, because of the total code would not fit into
(Expected Transmission Count) [27]. Thus, the rank the program memory of the Tmote Sky (48 kB). The
is calculated as follows: different simulation parameters are summarized in
Tab. 1.
new_rank = base_rank + rank_increase (1)
As discussed in the previous section, the rank value
where
in RPL-Contiki is not related only to the number of hops
− new_rank is the node's Rank,
toward the sink, but it also depends on the link's quality,
− base_rank: is the DIO sender's Rank, for this reason, the rank value changes continuously in
− rank_increase is a value that shows the property a static topology. The evolution of the node's “25” rank
of the link (path cost) between the node and the during an attack free simulation is shown in Fig. 3.
DIO sender. According to the figure, the rank's value kept conti-
RPL in its basic scheme does not implement any nuously changing during the simulation, reaching
mechanism to deny advertising a faked rank value. Thus, a maximum value of 1384 and a minimum value of
a malicious node can take advantage of this gap to 1161.
broadcast a better falsified rank value aiming to be
a parent of a large part of the network's nodes. Upon
having a good position, an intruder can establish any
 Journal of Electrical Engineering, Vol. 74, No. 6, 2023 457

Table 1. Simulation parameters types of attack such as isolation attacks, IP spoofing,

Selective forwarding attack, etc.
Parameter Values
In this LPRA, we propose and implement a special
Network layer Protocol RPL case of the well-known decreased rank attack, where the
Operating System Contiki 3.0. malicious node advertises the same rank value as its
preferred parent. For this purpose, we modified the
Simulator Contiki Cooja RPL's objective function of the malicious node, file
Emulated nodes Z1 “rpl-mrhof.c”, where the value of the calculated rank is
modified to
MAC layer Protocol IEEE 802.15.4
𝑛𝑒𝑤_𝑟𝑎𝑛𝑘 = 𝑏𝑎𝑠𝑒_𝑟𝑎𝑛𝑘 (2)
Radio model UDGM
Simulation area 200 m × 200 m 14

Simulation time 20 minutes 12

Data transmission 1 Packet / 60s 10

Number of children
Objective function MRHOF 8

6
Attack free
LPRA
4
1400

2
1350

1300 0

0 2 4 6 8 10 12 14 16 18 20
1250 Time (minutes)
Rank

1200 Fig. 4. Number of children of node 25 through time

1150
Figure 4 shows that the number of direct or indirect
1100 children the malicious node owns in its sub-DODAGs
for the attack free case and for the proposed attack Like
1050
parent's rank attack (LPRA). Regarding the attack free
0 2 4 6 8 10 12 14 16 18 20 case, the number of children shows a change in the first
Time (minutes)
10 minutes and it drops to be zero in the second 10
Fig. 3. Rank value of node "25" during simulation minutes of the simulation. When it comes to LPRA
attack, the malicious node has most of the time 8 to 10
In the following, we present two special and silent children, which reflect the success of the conducted
rank attacks against RPL-Contiki. In RPL-Contiki, attack. This behavior in the network leads to the creation
loops, sign of rank attack, are detected based on of non-optimal paths, since all the upward traffic sent by
comparing the number of hops toward the sink, not the the children passes through the malicious node.
rank value itself. Consquently, the only case where a
loop is detected is when a child node with strict lower
number of hops sends a DAO message to its parent that 3.1.1 Impact of LPRA on the control overhead
has a higher number of hops. In the first proposed attack, Exchanging control messages in the proactive RPL
we take advantage of the above montionned feature, protocol is vital to keep the nodes noticed about any
where the intruder node advertises a rank value equal to changes in the network, and keep their routing tables up
its own preferred paret. to date. However, an extra overhead makes the nodes
process more control packets and exhaust their energy
and processing resources. The obtained results related to
3.1 Like parent's rank attack (LPRA) the control message overhead, basically DIO and DAO
The selection of a preferred parent by a given node is messages, are summarized in Tab. 2. The results show
simply choosing the node with the best rank (minimum that the attack did not engender an increase in the total
rank). If a malicious node modifies its advertised rank control overhead.
value, then it causes an adverse effect on topology [28].
Usually, rank attack is combined and followed by other
 458 Mehdi Rouissat et al.: A lightweight countermeasure solution against a new Silent Rank Attack n RPL-Contiki based IoT networks

Table 2. Control overhead in the attack free case Table 3. Energy consumption
and in LPRA by the four modes in (mJ)

Sent messages LPM CPU TX RX Total

Generated Forwarded Attack
free 4340.2 4338.5 15572.3 24288.7 48863.9
DIO DAO DAO Total
LPRA 4336 4478.4 16537.3 24843.5 50519.3
Attack-free 548 284 430 1262
LPRA 564 275 423 1262
3.2 Better than parent's rank attack (BPRA)
3.1.2 Impact of LPRA on the energy consumption In RPL-Contiki, inconsistencies in rank values are
detected based on comparing the number of hops toward
The energy of a node is an important resource in LLN the sink. If a child node with lower number of hops sends
networks, where network efficiency and lifetime is a DAO message to its parent that has higher number of
usually affected by the nodes energy consumption, hops then a loop is detected.
which is a major concern. In our comparison, we did not
take into consideration the root node, as naturally it is On the other hand, the rank value is not necessarily
the most loaded node with the greater energy resources an exact multiple of 256, since it is not based only on the
and processing capabilities. Thereby, energy consum- number of hops, but also on to the link's quality (ETX).
ption is more critical for the rest of the operating nodes, Based on these features, a malicious node can take
usually on their batteries. advantage of this conception and advertises a better rank
than its own preferred parent, by considering only the
Figure 5 depicts the total consumed energy by the last multiple of 256 of its parent's rank.
nodes, in four modes: CPU, LPM, TX and energy
consumed in the RX mode. In this regard, it can be
observed in Fig. 5 that a slight increase of 3.4% in the
1 Rank1 = 256
total consumed energy is recorded, from 48.9 J to 50.5 J,
result of the attack.
4
x 10
4.5
LPM 8 Rank8 = 514 = Rank1 + 258 = (2*256) + 2
4 CPU
Radio TX
3.5 Radio RX

3
Energy (mj)

2.5 48863.91 50519.33 10 Rank10 = 813 = Rank8 + 299 = (3*256) + 45

2

1.5

1 25 Rank25 = 1182 = Rank10 + 369 = (4*256) + 158

0.5
Fig. 6. Rank values of the malicious node
0
Attack free LPRA and its parents
To give a better understanding of the discussed
Fig. 5. Energy consumption in the attack free case concept, Fig.6 shows the rank values of the malicious
and in LPRA node “node 25” and its parents after 135 seconds of
simulation. The malicious node's parent, node 10, is
The increase in the consumed energy is mainly due advertising a rank of 813 = 768 + 45. If the malicious
to the increase in the TX mode, where it jumped from node advertises a rank value of 768, it will advertise
15.57 J in the attack free case to 16.53 J in the case of a rank value better than its preferred parent, without
LPRA, as table 3 details. engendering any loops, since, as discussed earlier it is
not violating the hop counts.
To implement this special attack, we modified the
RPL “rpl-mrhof.c” file of the malicious node, where the
value of the calculated rank is modified to
𝑅𝑎𝑛𝑘𝑁𝑜𝑑𝑒 = 𝑎𝑏𝑠(𝑏𝑎𝑠𝑒_𝑟𝑎𝑛𝑘 / 256 ) ∗ 256 (3)
 Journal of Electrical Engineering, Vol. 74, No. 6, 2023 459

By doing so, a modified rank attack is conducted This successful attack has reflected on the number of the
without triggering any loops in the topology, which forwarded DAO messages by the malicious node. In the
make it silent and hard to detect. attack free case, it forwarded 10 DAO messages, in
LPRA it has forwarded 59 DAO messages, while in
14 BPRA 71 DAO messages have been forwarded by the
malicious node, which again explain the strategic
12
position it gained in the network. The recorded increase
10 in the control messages would impose higher amounts of
resource consumption, particularly in terms of energy
Number of children

8
and links availability.
Attack free
6 LPRA
BPRA

4 3.2.2 Impact of BPRA on the energy consumption

2 Regarding the energy, which is a significant and
decisive metric on increasing the network lifetime,
0
Fig. 8, depicts the total consumed energy, in four modes:
0 2 4 6 8 10 12
Time (minutes)
14 16 18 20
CPU, LPM, TX and energy consumed in the RX mode.
The figure depicts an increase of 15% on the energy
Fig. 7. Number of children of node 25 through time for recorded by BPRA compared to the attack free case,
the three cases where LPRA shows an increase of only 3.4%.
4
x 10
4.5
Figure 7 illustrates the effect of BPRA, where it LPM
4 CPU
shows an identical behavior compared to LPRA in the Radio TX
first 10 minutes. This behavior did not remain during the 3.5 Radio RX

second 10 minutes, where the number of children the 3

56186.42
malicious node has is higher in BPRA, it successfully
Energy (mj)

48863.91 50519.33
2.5
could have up to 12 children nodes. Based on these
results, this second modified rank attack is considered as 2

more successful compared to LPRA. 1.5

0.5
3.2.1 Impact of BPRA on the control overhead
0
Table 4 summarizes the exchanged control messages Attack free LPRA BPRA

for the three cases. It can be seen that the BPRA

engenders an increase in the overhead, from 1262 to Fig. 8. Energy consumption for the three scenarios
1420 messages, an increase of 12.5%.

Table 4. Control overhead for the three cases The contribution of the different modes in the total
consumed energy is detailed in table 5. The TX mode is
Sent messages the main contributor, it showed an increase of 25.7%
compared to the attack free case.
Generated Forwarded
DIO DAO DAO Total
Table 5. Energy consumption in the three cases in mJ
Attack
548 284 430 1262
free
LPM CPU TX RX Total
LPRA 564 275 423 1262 Attack
4340.2 4338.5 15572.3 24288.7 48863.9
BPRA 603 326 491 1420 free
LPRA 4336 4478.4 16537.3 24843.5 50519.3
BPRA 4321.8 4945.3 19566.7 27028.5 56186.4
The main factor that leads to this increase is the
forwarded DAO messages. This increase reflects a non-
optimal topology, due to the non-optimally created paths
Based on its strategic position in the topology and its
by the attack, where around half the existing nodes are
new forwarding tasks, the malicious node is the main
direct or indirect children attached to the malicious node.
 460 Mehdi Rouissat et al.: A lightweight countermeasure solution against a new Silent Rank Attack n RPL-Contiki based IoT networks

contributor in the total consumed energy. According to Table 7. PDR and latency for the three cases
Fig. 9, the total consumed energy of node 25 has in-
creased compared to the attack free case, by 98% and Latency (s) PDR (%)
118% in the LPRA, and BPRA respectively.
Attack free 0.42 97

LPM CPU Radio TX Radio RX

LPRA 0.50 95
6000 BPRA 0.61 94
5000 4865.08
4417.63

4000
Table 7 also shows the recorded latency in the three
Energy (mj)

cases. The latency has been adversely affected by the

3000 attack, where it raised from 0.42 s in the attack free case
2228.53 to 0.61 s in the case of BPRA, which presents a signi-
2000 ficant increase of 45 %. This deterioration can be
attributed to the non-optimal topology and the non-
1000
optimal paths created by advertising a falsified rank
0
value by the malicious node to gain a strategic false
Attack free LPRA BPRA
position within the targeted topology.

Fig. 9. Energy consumption of node 25

3.3 Countermeasure
for the three cases
The two discussed rank attacks have been described
as silent, because no trace is left behind. Since the
The most affected mode by the attack is the TX,
detection of the rank attack in many works is based on
where it shows an increase of 227% in the case of BPRA
the detection of loops [31], it is sufficient to engender
compared to the attack free case, as Tab. 6 details.
loops in the topology in order to make the detection of
the attack possible. For this purpose, we modified the
“rpl-icmp6.c” file of the nodes, where the code of loops
Table 6. Energy consumption of node 25 in the three
detection has been changed to
cases in mJ
DAG _RANK(p->rank, instance) =
LPM CPU TX RX Total < DAG_RANK(dag->rank, instance)
Attack instead of
180.09 205.71 707.71 1135.02 2228.53
free (DAG _RANK(p->rank, instance) < DAG_RANK(dag-
LPRA 173.78 414.58 2072.14 1757.12 4417.63 >rank, instance).
This modification triggers loops if a child node
BPRA 172.54 454.94 2316.73 1920.87 4865.08
pretends to have the same number of hops toward the
sink like its parents, regardless the rank value it is
advertising. The obtained results regarding the number
3.2.3 Impact on the PDR and on the latency of loops recorder in the network after implementing the
The PDR (Packet Delivery Ratio) presents the ratio proposed change are summarized in Tab. 8.
of the total number of received data packets by the root
node to the total number of sent data packets by the rest Table 8. Recorded loops for the three cases
of the network's nodes [29], while the latency is the
average time taken for a given number of successful data Number of loops
packets transmitted by the network's nodes to be
received by the root node [30]. Table 7 shows the PDR Attack free 0
for the three studied cases. The figure shows that the LPRA 1
PDR has been slightly affected under the attack, where a
BPRA 603
ratio of 95% is recorded in LPRA and a ratio of 94% is
recorded in the case of BPRA. This is mainly due to the
congestion incurred due to non-optimal paths created by Thereby, the network is more usually advertised by
the attack, in particular at the level of the forwarder the inconsistency engendered by the harmful behavior of
nodes. the malicious node. Thus, further isolation or elimi-
nation process will be easily triggered.
 Journal of Electrical Engineering, Vol. 74, No. 6, 2023 461

4 Conclusion [6] V. Adat and B. B. Gupta, “Security in Internet of

Things: issues, challenges, taxonomy, and
In this paper, we presented and studied two special
architecture,” Telecommunication Systems, vol. 67,
silent rank attacks against RPL-Contiki protocol. The
no. 3, pp. 423-441, Jun. 2017,
demonstrated results in a random topology show how the
doi: https://doi.org/10.1007/s11235-017-0345-9.
intruder node could attract half the existing nodes in the
[7] J. Neeli and S. Patil, “Insight to Security Paradigm
topology to join its sub-DODAGs, without leaving any
, Research Trend & Statistics in Internet of
traces behind. Usually, rank attack is combined and
Things(IoT),” Global Transitions Proceedings,
followed by other types of attack such as isolation
Jan. 2021,
attacks, IP spoofing, selective forwarding attack, etc.
doi: https://doi.org/10.1016/j.gltp.2021.01.012.
The obtained results show that the default RPL-Contiki
[8] N. Miloslavskaya and A. Tolstoy, “Internet of
requires many improvements in terms of security to face
Things: information security challenges and
the potential threads in IoT networks.
solutions,” Cluster Computing, vol. 22, no. 1, pp.
In order to make the detection of the new variant of 103-119, Jul. 2018,
rank attack possible, a slight modification in RPL doi: https://doi.org/10.1007/s10586-018-2823-6.
functioning code is proposed to allow triggering loops in [9] J. Howarth, “80+ Amazing IoT Statistics (2022-
the case where a malicious node advertises a similar rank 2030),” Exploding Topics, Dec. 22, 2021.
value as its preferred parent. Combined to existing RPL https://explodingtopics.com/blog/iot-stats
rank attack countermeasure based on the detection of [10] Liebermann. N, “2021 IoT Security Landscape -
loops, it becomes easy to detect the attack and neutralize SAM Seamless Network,” Apr. 07, 2022.
the intruder node. This implementation presents a viable https://securingsam.com/2021-iot-security-
addition to enhance the security aspect in default RPL landscape
Contiki. [11] T. Winter , “RPL: IPv6 Routing Protocol for Low-
Power and Lossy Networks,” tools.ietf.org.
https://tools.ietf.org/html/rfc6550
References [12] Gupta, M., Jain, S., & Patel, R. B, “Security issues
[1] M. Majid et al., “Applications of wireless sensor in internet of things: principles, challenges,
networks and internet of things frameworks in the taxonomy. In: Singh, P.K., Singh, Y., Kolekar,
industry revolution 4.0: A systematic literature M.H., Kar, A.K., Chhabra, J.K., Sen, A. (eds)
review,” Sensors, vol. 22, no. 6, p. 2087, 2022. Recent Innovations in Computing. ICRIC 2020.
doi:10.3390/s22062087. Lecture Notes in Electrical Engineering, Springer,
[2] S. M. Tahsien, H. Karimipour, and P. Spachos, Singapore 701. https://doi.org/10.1007/978-981-
“Machine learning based solutions for security of 15-8297-4_52, 2021
Internet of Things (IoT): A survey,” Journal of [13] R. R. Krishna, A. Priyadarshini, A. V. Jha,
Network and Computer Applications, vol. 161, B. Appasani, A. Srinivasulu, and N. Bizon, “State-
p. 102630, Jul. 2020, of-the-Art Review on IoT Threats and Attacks:
doi: https://doi.org/10.1016/j.jnca.2020.102630. Taxonomy, Challenges and Solutions,”
[3] M. Saleh et al., “Deep reinforcement learning based Sustainability, vol. 13, no. 16, p. 9463, Jan. 2021,
transmission policy enforcement and multi-hop doi: https://doi.org/10.3390/su13169463.
routing in QoS aware LoRa IoT networks,” [14] M. A. Boudouaia, A. Ali-Pacha, A. Abouaissa, and
Computer Communications, vol. 183, pp. 33-50, P. Lorenz, “Security Against Rank Attack in RPL
Feb. 2022, Protocol,” IEEE Network, vol. 34, no. 4, pp. 133-
doi: 139, Jul. 2020,
https://doi.org/10.1016/j.comcom.2021.11.010. doi: https://doi.org/10.1109/mnet.011.1900651.
[4] A. Raghuvanshi, U. Kumar Singh, M. Shuaib, and [15] A. O. Bang and U. P. Rao, “EMBOF-RPL: Impro-
S. Alam, “An investigation of various applications ved RPL for early detection and isolation of rank
and related security challenges of Internet of attack in RPL-based internet of things,” Peer-to-
things,” Materials Today: Proceedings, Mar. 2021, Peer Networking and Applications, vol. 15, no. 1,
doi: https://doi.org/10.1016/j.matpr.2021.01.821. pp. 642-665, Jan. 2022,
[5] M. N. Khan, A. Rao, and S. Camtepe, “Lightweight doi: https://doi.org/10.1007/s12083-021-01275-3.
Cryptographic Protocols for IoT Constrained [16] A. Seyfollahi, M. Moodi, and A. Ghaffari, “MFO-
Devices: A Survey,” IEEE Internet of Things RPL: A secure RPL-based routing protocol utili-
Journal, pp. 4132-4156, 2020, zing moth-flame optimizer for the IoT appli-
doi: https://doi.org/10.1109/jiot.2020.3026493. cations,” Computer Standards & Interfaces, vol.
82, p. 103622, Aug. 2022,
doi: https://doi.org/10.1016/j.csi.2022.103622.
 462 Mehdi Rouissat et al.: A lightweight countermeasure solution against a new Silent Rank Attack n RPL-Contiki based IoT networks

[17] P. P. Ioulianou, V. G. Vassilakis, and S. F. Shahan- [25] M. Rouissat, M. Belkheir, I. S. Alsukayti, and
dashti, “A Trust-Based Intrusion Detection System A. Mokaddem, “A Lightweight Mitigation Ap-
for RPL Networks: Detecting a Combination of proach against a New Inundation Attack in RPL-
Rank and Blackhole Attacks,” Journal of Cyber- Based IoT Networks,” Applied Sciences, vol. 13,
security and Privacy, vol. 2, no. 1, pp. 124-152, no. 18, p. 10366, Sep. 2023,
Mar. 2022, doi: 10.3390/app131810366.
doi: https://doi.org/10.3390/jcp2010009. [26] P. Thubert, (2012, March). Objective function zero
[18] A. Agiollo, M. Conti, P. Kaliyar, T.-N. Lin, and for the routing protocol for low-power and lossy
L. Pajola, “DETONAR: Detection of Routing networks (RPL). Internet Engineering Task Force.
Attacks in RPL-Based IoT,” IEEE Transactions on Retrieved July 20, 2022.
Network and Service Management, vol. 18, no. 2, http://www. ietf.org/rfc/rfc6552.txt
pp. 1178-1190, Jun. 2021, [27] O. Gnawali, P. Levis, (2012 September), The
doi: https://doi.org/10.1109/tnsm.2021.3075496. minimum rank with hysteresis objective function.
[19] Z. A. Almusaylim, N. Jhanjhi, and A. Alhumam, Internet Engineering Task Force. Retrieved July 20,
“Detection and Mitigation of RPL Rank and 2022. URL http://www.ietf.org/rfc/rfc6719.txt
Version Number Attacks in the Internet of Things: [28] D. Airehrour, J. Gutierrez, and S. K. Ray, “Secure
SRPL-RP,” Sensors, vol. 20, no. 21, p. 5997, Oct. routing for internet of things: A survey,” Journal of
2020, doi: https://doi.org/10.3390/s20215997. Network and Computer Applications, vol. 66, pp.
[20] P. S. Nandhini, S. Kuppuswami, and S. Malliga, 198-213, May 2016,
“Energy efficient thwarting rank attack from RPL doi: https://doi.org/10.1016/j.jnca.2016.03.006.
based IoT networks: A review,” Materials Today: [29] M. Amirinasab Nasab, S. Shamshirband, A. Chro-
Proceedings, May 2021, nopoulos, A. Mosavi, and N. Nabipour, “Energy-
doi: https://doi.org/10.1016/j.matpr.2021.04.167. Efficient Method for Wireless Sensor Networks
[21] A. Mayzaud, R. Badonnel, I, Chrisment, “A taxo- Low-Power Radio Operation in Internet of Things,”
nomy of attacks in RPL-based internet of things”, Electronics, vol. 9, no. 2, p. 320, Feb. 2020,
International journal of network security, 18 (3), doi: https://doi.org/10.3390/electronics9020320.
459-473, 2016, DOI： 10.6633/IJNS.201605 [30] S. S. Solapure and H. H. Kenchannavar, “Design
[22] H. Kumar Saini and M. Poriye, “Threats, Detection and analysis of RPL objective functions using
and Mitigation of Rank Attack: A Survey,” SSRN variant routing metrics for IoT applications,”
Electronic Journal, 2021, Wireless Networks, vol. 26, no. 6, pp. 4637-4656,
doi: https://doi.org/10.2139/ssrn.3884409. May 2020, doi: https://doi.org/10.1007/s11276-
[23] A. Le, J. Loo, A. Lasebae, A. Vinel, Y. Chen, and 020-02348-6.
M. Chai, “The Impact of Rank Attack on Network [31] R. Stephen, L. Arockiam, (2018) “RIAIDRPL:
Topology of Routing Protocol for Low-Power and Rank Increased Attacks (RIA) Identification
Lossy Networks,” IEEE Sensors Journal, vol. 13, algorithm for avoiding loop in the RPL DODAg”,
no. 10, pp. 3685-3692, Oct. 2013, International Journal of Pure and Applied
doi: https://doi.org/10.1109/jsen.2013.2266399. Mathematics, 119 (16), 1203-1210, 2018.
[24] M. Rouissat, M. Belkheir, and H. S. A. Belkhira,
“A potential flooding version number attack against Received 23 October 2023
RPL based IOT networks,” Journal of Electrical ________________________________
Engineering, vol. 73, no. 4, pp. 267-275, Aug.
2022, doi: https://doi.org/10.2478/jee-2022-0035.