Analysis of RPL Objective Functions with Security Perspective
Cansu Dogan
1 a
, Selim Yilmaz
1,2 b
and Sevil Sen
1 c
1
WISE Lab., Department of Computer Engineering, Hacettepe University, Ankara, Turkey
2
Department of Software Engineering, Mu
˘
gla Sıtkı Koc¸man University, Mu
˘
gla, Turkey
Keywords:
Internet of Things, Routing Protocol, RPL, Objective Functions, RPL Security, RPL Attacks.
Abstract:
The IPv6 Routing Protocol for Low Power Lossy Networks (RPL) is one of the standardized routing pro-
tocols for lossy networks consisting of resource-constrained Internet of Things (IoT) devices. RPL allows
to use different objective functions based on different routing metrics such as expected transmission count
(ETX), hop count, and energy to determine effective routes. In the literature, the performance of two objec-
tive functions namely Objective Function Zero (OF0), Minimum Rank with Hysteresis Objective Function
(MRHOF) are evaluated thoroughly, since they are accepted as standard objective functions in RPL. However
their performance under attack has not been evaluated comprehensively yet. Although RPL has defined some
specifications for its security, it is still vulnerable to insider attacks, which could dramatically affect the net-
work performance. Therefore, this study investigates how the performance of objective functions are affected
by RPL specific attacks. Version number, DIS flooding, and worst parent attacks are analyzed by using the
following performance metrics: packet delivery ratio, overhead, latency, and power consumption. Moreover,
how they are affected by the number of attackers in the network are analyzed. To the best of the authors’
knowledge, this is the first study that comprehensively explores RPL objective functions on networks under
attacks.
1 INTRODUCTION
The recent progress in embedded system technol-
ogy has led to the emergence of numerous sensor
devices having different architectures today. These
resource-constrained devices having small on-board
memory, low power, and low computational capabil-
ity can communicate with each other and connect to
the Internet in order to complete specific tasks. IoT
networks consisting of such sensors have grown so
rapidly in a very short time (Statistica, 2016). They
are used in various IoT applications in every aspect
of our daily life such as healthcare, agriculture, trans-
portation, security.
The Low Power and Lossy Networks (LLN) are
a special type of IoT, in which resource constrained
devices are connected over lossy links that have high
packet loss and so low throughput. In order to meet
the special requirements of such lossy and resource-
constrained networks, a new routing protocol called
RPL (Routing Protocol for Low Power and Lossy
a
https://orcid.org/0000-0002-3806-657X
b
https://orcid.org/0000-0002-9516-6892
c
https://orcid.org/0000-0001-5814-9973
Networks) is developed by IETF-ROLL (Winter et al.,
2012). Due to its efficient routing capability, RPL
has now become a standard routing protocol. The
optimal route between two end-points is found by
selecting the most appropriate communication links
between every node. To ensure that, RPL employs
different objective functions that are based on dif-
ferent routing metrics such as ETX, hop count, and
energy. Although, a number of objective functions
have been proposed till now; OF0 (Thubert, 2012)
and MRHOF (Gnawali and Levis, 2012) are regarded
as standard objective functions in RPL by the Internet
Engineering Task Force Routing over Low power and
Lossy networks (IETF-ROLL) working group and are
widely implemented in the most popular IoT applica-
tions today (Onwuegbuzie et al., 2020; Pradittasnee,
2017). OF0 considers the hop count, and hence it tar-
gets the minimum hop count between sender nodes
and the root node in the network. MRHOF, how-
ever, uses link- or node-based routing metrics. There-
fore, it provides an optimal path based on the met-
ric used. The selection of objective functions could
change routing performance, so the selection of the
appropriate objective function is usually based on the
requirements of the application at hand.
Dogan, C., Yilmaz, S. and Sen, S.
Analysis of RPL Objective Functions with Security Perspective.
DOI: 10.5220/0011011900003118
In Proceedings of the 11th International Conference on Sensor Networks (SENSORNETS 2022), pages 71-80
ISBN: 978-989-758-551-7; ISSN: 2184-4380
Copyright
c
2022 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
71
Although RPL ensures an efficient routing perfor-
mance for such constrained networks, it is vulnerable
to insider attacks. The insecure nature of this pro-
tocol is given one of the main obstacles for the fur-
ther development IoT applications (Verma and Ranga,
2019; Abhishek and Virender, 2020). RPL attacks can
easily harm the network they are incorporated by ex-
ploiting and exhausting the resources of constrained
nodes. The degree to which the network is affected
changes according to the objective functions because
they rely on different routing metrics. In this study,
we investigate how objective functions are sensitive
to different attack types and explore if any correla-
tion exists between number of attackers and perfor-
mance degradation when different objective functions
are used. In this regard, to the best of our knowledge,
this is the first study in the literature that compre-
hensively analyzes objective functions from security
point of view. Three RPL specific attacks, namely;
version number, DIS flooding, and worst parent at-
tacks are implemented on a large number of simulated
networks with varying topologies in order to ensure a
better evaluation. The simulation results are evaluated
by using the following performance metrics: packet
delivery ratio (PDR), overhead, latency, and power
consumption. A large number of experimental envi-
ronments are set in order to ensure a better evaluation.
The experimental results show that RPL attacks, espe-
cially the ones aiming to consume network resources
(i.e., version number and DIS flooding attacks) affect
the performance of networks considerably even in the
presence of a single attacker. Moreover, they affect
networks that use different objective functions differ-
ently. While OF0 is more robust to attacks, MRHOF-
ENERGY shows the worst performance in almost all
performance metrics due to the overhead caused by
attackers.
The rest of the paper is organized as follows. The
background information that covers an overview of
RPL, the standard/default objective functions used in
RPL, and attacks targeting RPL is given in Section 2.
Section 3 discusses analysis studies in the literature
that focus on objective functions and attacks in RPL
separately. Experimental settings are introduced and
the experimental results are discussed in Section 4.
Finally, Section 5 concludes the study.
2 BACKGROUND
2.1 Overview of RPL
RPL is a distance vector routing protocol. Nodes in a
RPL-based IoT network are connected to each other
through a special topology that is the combination of
tree and mesh topologies called Destination Oriented
Directed Acyclic Graphs (DODAG). A DODAG com-
prises of a root (or sink) node, which is responsible for
the initiation of DODAG building, and a number of
sensor nodes. RPL enables multipoint-to-point com-
munication (MP2P) from sensor nodes to root node,
point-to-multipoint communication (P2MP) from the
root node to sensor nodes, and point-to-point commu-
nication (P2P) between sensor nodes.
A network can operate on one or more RPL in-
stances where multiple DODAGs can take part. The
role of each instance is to define an objective function
to calculate the optimum path within the DODAG. A
DODAG is built by using the following RPL control
packets:
DODAG Information Object (DIO): It is initiated
and broadcast by only the root node. DIO pack-
ets carry network information (e.g., instance ID,
version number). Each of the receiving node adds
the sender to its parent list, calculates its own rank
value, which states its position in the graph with
respect to the root node, and finally it forwards
DIO to its neighbors. DIO packets are relayed
throughout the graph and play a major role in con-
structing the default upward routes.
DODAG Information Solicitation (DIS): It is used
as a solicitation for having DIO information when
a new node is to join the DODAG. DIS packets
are broadcast by the new node to its neighbors.
Destination Advertisement Object (DAO): It is
used for the construction of the downward routes
from the root to sensor nodes. Based on the mode
of operation, the child nodes unicast DAO packets
either to root node (in non-storing mode) or to its
selected parent (in storing mode) so that it records
downward routes in its routing table for the sub-
DODAG.
Destination Advertisement Object Acknowledge-
ment (DAO-ACK): Upon receiving DAO packets
from a parent node, DAO-ACK packets are sent
to the the sender node as an acknowledgement.
2.2 Objective Functions in RPL
Objective function (OF) is used to select optimal
routes within a DODAG by affecting the calculation
of the rank values of each participating node. As
stated earlier, it differs with respect to the RPL in-
stances, hence different objective functions could be
simultaneously used within an RPL network by dif-
ferent instances. For example, one can take hop count
into consideration to build routes of a DODAG graph,
SENSORNETS 2022 - 11th International Conference on Sensor Networks
72
the residual energy of the nodes could be used for
finding the routes of another DODAG in the same net-
work. Therefore, the selection of appropriate objec-
tive functions is crucially important, and it changes in
accordance with the requirements of the application.
Even though there have been a number of ob-
jective functions proposed in the literature so far as
discussed in Section 3, OF0 (Thubert, 2012) and
MRHOF (Gnawali and Levis, 2012) are proposed as
the default objective functions for RPL-based IoT net-
works:
2.2.1 OF0
OF0 takes the hop count between the root node and
a sensor node into account for the calculation of rank
value of that node. Therefore, it aims to minimize the
number of hops to reach to the root node by choosing
the node that has the lowest rank from its reachable
neighbors as its parent. When OF0 is used as the ob-
jective function in the network, for a given node n, the
rank of this node can be calculated by using (1).
R(n) = R(p) + RI (1)
R(n) is the new rank of node n, R(p) is the rank of
the preferred parent node, and RI stands for the rank
increase metric that is calculated by using (2).
RI = (R f × Sp +Sr) × MHRI (2)
R f is a configurable rank factor and it uses 1 as the
default value. Sp is the step of the rank, and Sr is
the maximum value assigned to the rank level. MHRI
stands for MinHopRankIncrease which is a constant
value that is defined as 256 in RFC6550 (Winter et al.,
2012).
2.2.2 MRHOF
Unlike to OF0, a number of link- and node-based ad-
ditive routing metrics can easily be integrated into
MRHOF. The rank value (R(n) in Eq. 1), and hence
the routing path, is determined according to the em-
ployed routing metric which is stored in Metric Con-
tainer suboption in the DIO packet.
While ETX, latency, packet loss rate, received sig-
nal strength indication are example to the link-based
routing metrics; the remaining energy, maximum life
time, and trustworthiness are examples to the node-
based routing metrics for LLNs. By using one of
these routing metrics, MRHOF ensures the lowest
cost path in the LLN. Two metrics are integrated into
MRHOF in this study: MRHOF with ETX (MRHOF-
ETX), which is a link-based metric, and MRHOF
with energy (MRHOF-ENERGY), which is a node-
based metric.
MRHOF-ETX chooses the paths with the lowest
number of transmission value by considering ETX
values of the links. The ETX value of the links is
calculated using (3).
ET X =
1
D f × Dr
(3)
D f is the probability of the packet being reached by
the neighbor, and Dr is the probability of the acknowl-
edgment packet being received.
MRHOF-ENERGY chooses the path that pro-
vides maximum remaining energy for the RPL nodes.
Energy metric of the nodes is calculated using (4).
ENERGY =
P
max
P
now
(4)
where P
max
is defined as the targeted maximum
power, and it is calculated from the initial energy of
node divided by the targeted lifetime; P
now
, however,
is the actual power of node.
2.3 RPL Specific Attacks
One of the main drawbacks of the RPL protocol is
that it is vulnerable to a variety of attacks due to its
nature. Moreover, resource-constraint characteristic
of LLNs makes them vulnerable to Denial-of-Service
(DoS) attacks. RPL specific attacks can vary accord-
ing to what they primarily target, and these attacks are
divided into three categories: attacks targeting net-
work resources, network topology, and network traf-
fic (Mayzaud et al., 2016). In this study, we have stud-
ied version number, DIS flooding, and worst parent
attacks:
Version Number: Version number in DIO packets
is used by the DODAG root in order to perform
global repair, and it is increased by only the root
node. In the attack scenario, the malicious node il-
legitimately increases the incoming version num-
ber, causing unnecessary rebuilding of DODAG.
DIS Flooding: It is a typical RPL specific DoS at-
tack that targets consuming network resources. In
order to make nodes or links unavailable in LLNs,
attacker continuously sends large amount of con-
trol packets. This attack is often performed by
sending broadcast or unicast DIS packets after re-
ceiving a DIO packet from a node. By doing so,
the DIS flooding attack brings about network con-
gestion or overloading of RPL nodes.
Worst Parent: As stated earlier, an RPL node
chooses its own parent node according to the rank
Analysis of RPL Objective Functions with Security Perspective
73
value that is determined by the objective func-
tion, which ensures the ‘best parent’ for that RPL
node. However, in this attack scenario, attacker
node contrarily chooses the worst parent, resulting
in non-optimized routing path and hence leading
LLN to show very poor performance.
3 RELATED WORKS
By building efficient routes among sensor nodes, the
objective functions defined on LLNs have a great im-
pact on the Quality of Service (QoS) in an IoT net-
work. That’s why, the attention towards RPL objec-
tive functions has been growing; and until now, there
is a good deal of studies proposed on this research
area in the literature. In this section, we mainly dis-
cuss the studies that analyze the existing standard OFs
and that propose new OFs by integrating different
routing metrics. Lastly, the studies analysing routing
attacks against RPL are summarized and the contribu-
tion of the current study is emphasized.
As stressed earlier, OF0 and MRHOF objective
functions are the basic functions defined in the RPL
protocol. That’s why they have been investigated
by far in the literature. In (Kechiche et al., 2017),
how efficient routing built by OF0, MRHOF-ETX,
and MRHOF-ENERGY as a response to different net-
work densities is investigated. The analysis in this
study shows that MRHOF-ENERGY performs poorly
in high and low traffic, while OF0 and MRHOF-ETX
perform better on low traffic. A very similar analy-
sis that shows the impact of network density on dif-
ferent OFs is given in (Qasem et al., 2015). Unlike
to (Kechiche et al., 2017), the authors here took the
topology (grid and random) and the packet reception
ratio (RX) into consideration as key parameters since
they might play a major role on the performance of
OFs. For both topologies, a higher PDR performance
and lower battery level are obtained as RX increases
and as the network becomes more dense. The routing
performances of OF0 and MRHOF (with ETX and ex-
pected lifetime [ELT] metrics) are analyzed in (Jamil
et al., 2019) according to the distance of sensor nodes
to the root node. The findings reveal that, for all OFs,
the network performance (PDR, overhead, latency)
gradually decreases with the increase in the average
distance of nodes to the root. The parameters like
node density, transmission range, and time interval
for sending control packets are considered in (Mar-
dini et al., 2018) for the performance analysis of OF0
and MRHOF-ETX. As expected, the authors found
that a positive correlation between the sending inter-
val and PDR; whereas, a negative correlation between
the sending interval and power consumption. The
node density, however, has no major impact on the
performance metrics.
The fact that network requirements could vary ac-
cording to IoT applications or services, the develop-
ment of new objective functions for these applications
by integrating new RPL metrics or combining existing
metrics has become one of the most studied research
areas in RPL (Pancaroglu and Sen, 2021). In (Al-
Kashoash et al., 2016), the authors showed that a great
number of data packets are lost when the data traf-
fic is high. To ensure a reliable and efficient routing
when the network is under congestion, they took the
buffer occupancy into consideration as a routing met-
ric; and hence, Congestion-Aware Objective Func-
tion (CA-OF) is proposed. Even though the perfor-
mance of CA-OF degrades when the traffic is higher,
it still performs superior than OF0, MRHOF-ETX,
and MRHOF-ENERGY in terms of PDR, energy con-
sumption, and packet loss.
Not only a single RPL metric, but also multiple
metrics can be integrated to OFs so that it considers
more than one objective simultaneously. In (Sousa
et al., 2017), Energy Efficient and Path Reliability
Aware Objective Function (ERAOF) is proposed. By
considering two RPL metrics, energy consumed and
ETX, ERAOF disregards the routes with low energy
level and high probability of packet loss. In com-
parison to OF0 and MRHOF-ETX, ERAOF yields
higher PDR, lower number of hops, and comparable
energy consumption. In (Xiao et al., 2014), the au-
thors combined hop count and ETX metrics in order
to improve MRHOF-ETX. However, that the cumula-
tive ETX value can only be calculated along the path
(as in MRHOF-ETX) might be misleading in the se-
lection of appropriate route. Regardless of the quality
of each link, a node using MRHOF-ETX is likely to
choose the path with fewer hops where the cumulative
ETX value is relatively smaller. In order to avoid this
situation, authors calculated the ETX value per hop
by dividing ETX by the number of hops. This yield
a higher PDR, shorter latency, and lower energy con-
sumption in comparison to OF0 and MRHOF-ETX.
The results suggest that network shows better perfor-
mance when multiple RPL metrics are used together
to choose paths. These metrics could be combined
in various ways according to requirements of differ-
ent applications. In (Karkazis et al., 2012) two differ-
ent strategies for combining four different RPL met-
rics that are hop count, ETX, packet forwarding in-
dication, and remaining energy are proposed. These
are additive and lexical combinations. While these
metrics values are averaged with relative weights in
the additive combination, it is the prioritization of the
SENSORNETS 2022 - 11th International Conference on Sensor Networks
74
metrics that plays a key role in the lexical combina-
tion. For example, in the lexical combination of hop
count and packet forwarding indication, the hop count
is checked first and only if two paths have an equal
hop count metric, then the packet forwarding indica-
tion metric value is taken into consideration.
The performance of an RPL-based IoT network is
not only affected by the OFs, but also by malicious
attempts; since attackers could dramatically harm the
resource, traffic, and topology of a network. There-
fore, there are also studies in the literature that an-
alyze the effects of RPL attacks. The version num-
ber attack is studied in (Aris et al., 2016). The effect
of this attack on network performance is analyzed by
two parameters: the attacker location (with respect to
the root node) and attacking probability. They found
that the attacker location has a clear effect on PDR
and overhead, but not the key factor for power con-
sumption and latency. As expected, all performance
metrics dramatically decrease as the attacking proba-
bility increases. This study is extended in (Arıs¸ and
Oktu
˘
g, 2020) by integrating multiple attackers into
the network.
The rank value plays an important role in RPL op-
erations such as including creation of optimal topol-
ogy, prevention of loop formation in DODAG. How-
ever, it could be exploited by malicious nodes in order
to dramatically affect the network’s resources, topol-
ogy, and traffic. This illegitimate attempt is known as
rank attack. The rank attack is analyzed with differ-
ent attacker locations in (Le et al., 2013). The anal-
ysis here shows that the bigger the forwarding load
area, which is the sum of the forwarding load of all
nodes in the area, is, the more impact attack leads
to on the network performance. In addition, the co-
operation of multiple attackers gives severe damage
to the network performance. The DIO packets are
very crucial in constructing the DODAG, and find-
ing upward routing paths where the majority of ap-
plication traffic follows. In order to harm a network,
the malicious nodes could either interrupt or at least
slow down the propagation of DIO packets, which is
known as DIO suppression attack. This attack is an-
alyzed in (Perazzo et al., 2017) with five cooperative
malicious nodes. It is shown that DIO suppression at-
tack considerably affects the packet delivery ratio and
network path stretch which is a metric used in order
to show the difference between the cost of a current
route cost and the cost of the shortest path.
As shown in the literature, the performance of
an RPL-based IoT network is very sensitive to rout-
ing attacks. In addition, how a network is affected
might vary according to the OFs as RPL attacks target
network- or node-related properties. That’s why, it is
worth analyzing OFs against various RPL specific at-
tacks. This is the main motivation of this study. To the
best of our knowledge, there is only a study (Semedo
et al., 2018) in the literature that addresses this prob-
lem. However, in that study the authors assess only
OF0 and MRHOF-ETX objective functions against
rank attack. In their experiements, only one network
with a single attacker node is simulated. That’s why,
the impact of the number of attackers on different OFs
is not studied. Moreover, the simulation is run on a
very small network consisting of only 19 nodes, lead-
ing a very low traffic. Hence, how OFs perform in
dense networks is disregarded in that study. More im-
portantly, it is suggested in (Kim et al., 2017) to use
at least 25 nodes in network simulations in order to
see multi-hop characteristics of RPL. In short, a lot
of research questions are left unresolved in (Semedo
et al., 2018). In this current study, we include mul-
tiple RPL specific attacks with varying number of
attackers. Instead of a single network, we consider
ten networks with 50 nodes for each attack scenario.
The main contribution of this current study is to ana-
lyze OF0, MRHOF-ETX and MRHOF-ENERGY ob-
jective functions with security perspective thoroughly
and discuss the effects of attacks on these OFs with
comprehensive simulations.
4 ANALYSIS OF RPL OBJECTIVE
FUNCTIONS
This study analyzes the impact of routing attacks on
RPL-based IoT networks based on different objective
functions. In order to that, different attack scenarios
are implemented on simulated networks. The subse-
quent chapters introduces the details of these simu-
lated networks and the performance metrics that are
used to compare different objective functions. Finally,
the experimental results are discussed thoroughly.
4.1 Simulation Settings
RPL attacks, objective functions, and the number of
attackers are used as parameters that are to be inves-
tigated in the experiments. The combination of these
input parameters defines a network scenario. In order
to simulate each scenario, Cooja simulator (Osterlind
et al., 2006), a Java-based network simulator of sen-
sor nodes running the Contiki (version 2.7) operating
system (Contiki-Ng, 2004), is used. Each scenario is
simulated with a parameter set given in Table 1.
OF0, MRHOF-ETX, and MRHOF-ENERGY are
taken into consideration as objective functions in the
simulation environment; where version number, DIS
Analysis of RPL Objective Functions with Security Perspective
75
Table 1: Simulation Parameters.
Simulation Parameters Values
Radio Environment UDGM: Distance Loss
Objective Functions OF0, MRHOF-ETX, MRHOF-ENERGY
TX Range 50m
INT Range 100m
Simulation Time 1 hour
Area of Deployment 200x200
Number of Sink Node 1 node
Number of Sensor Node 50 nodes
Platform Sky mote
Traffic Pattern UDP packets, every 60 sec. by sensor nodes
flooding, and worst parent attacks are considered to
perform malicious activities in the network. The ef-
fects of these attacks on the objective functions are
also investigated with respect to the number of attack-
ers. To do that, the simulations are first run without
any attacker node to create the baseline performance
of the objective functions. Then, the same simulation
scenarios are run with one (2%), three (6%), and five
(10%) attacker nodes.
For each scenario, the simulation is run with 10
different network topologies where the nodes are ran-
domly localized. The simulation is run for one hour
for every scenario considered. Unit Disk Graph
Medium (UDGM) is used to simulate the real lossy
environment in LLN.
4.2 Performance Evaluation Metrics
In order to reveal how the network is affected by the
objective function employed, four performance met-
rics are used in this study: packet delivery ratio, traf-
fic overhead, average latency, and average power. In
the following, these metrics are explained:
Overhead (OVR): It represents the total number
of control packets propagated in DODAG to build
network. Therefore, it is expected for a network
to have less OVR value when it is operated under
ideal conditions. The calculation of OVR is given
in (5).
OV R =
CP (5)
where CP {DIO,DIS, DAO} stands for control
packets propagated in the network.
Packet delivery ratio (PDR): It is defined as the
ratio of the total number of packets received by
the root node to the the total number of packets
sent to the root node. PDR shows how reliable the
network is, and the greater the PDR value is, the
more reliable the network is. The calculation of
PDR is given in (6).
PDR =
RT P
ST P
× 100 (6)
where RT P and ST P represent the total number of
packets that are, respectively, received and sent by
the root node.
Average Latency (ALT): Packet latency is the time
between sending a packet and reaching its destina-
tion. ALT is calculated by the total latency (T L)
of each packet divided by the total received pack-
ets (T RP), which is given in (7).
ALT =
T L
T RP
(7)
Power Consumption (PC): One of the most impor-
tant performance metrics in LLNs is power con-
sumption, since power is one of the main con-
straints in sensor devices. PC stands for the power
measured from nodes during the lifetime of the
network. The average power consumption of the
network is calculated by the ratio of the total PC
of the network to the total number of nodes. Pow-
ertrace (Dunkels et al., 2011) is integrated into the
Contiki OS in order to calculate PC of nodes by
using the (8).
Energy(mJ) = (Transmit × 19.5mA+
Listen ×21.5mA +CPU × 1.8mA
+LPM ×0.0545mA) × 3V /32768
(8)
PC(mW ) = Energy(m j)/Time(sec) (9)
4.3 Simulation Results
Ten random networks are simulated for each attack
type with different percentage of attackers (2%, 6%,
10%), and the average results of these ten runs are
shown in the results. Please note that while each
topology is randomly created, the same ten network
topologies in which attackers are located in the same
SENSORNETS 2022 - 11th International Conference on Sensor Networks
76
positions are run for each attack type/attacker percent-
age for a fair comparison.
Firstly, the average overhead introduced by at-
tacks is shown in Figure 1. As expected, version num-
ber attack results in a big increase in the number of
control packets, since it triggers the reconstruction of
DODAG. DIS flooding attack by its nature also in-
creases the number of DIS packets and the number
of DIO packets as a response to these DIS requests
considerably. The maximum number of control pack-
ets is produced by MRHOF-ENERGY in all attack
types. Moreover, MRHOF-ENERGY shows dramatic
increase in the number of control packets, particularly
at DIS flooding attacks. As more attackers take part
in the network; MRHOF-ETX results in lower over-
head than OF0; while OF0 still shows an incline to
increase in the number of control packets, MRHOF-
ETX is less affected by the number of attackers.
The average PDR of simulated networks with dif-
ferent objective functions under version number, DIS
flooding, and worst parent attacks is shown in Fig-
ure 2. As shown in the figure, OF0 and MRHOF-
ETX reach to nearly the optimal PDR when there
is no attack in the network. MRHOF-ENERGY not
only shows the lowest PDR, but also shows the fastest
decrease in PDR when the network is under attack.
However, all objective functions are considerably af-
fected by the version number attack even when there
is only one attacker in the network. As shown in the
results, while OF0 and MRHOF-ETX show compara-
ble performance against attacks, OF0 is slightly better
than MRHOF-ETX in each scenario.
The comparative average latency values (in sec) of
simulated networks under version number, DIS flood-
ing, and worst parent attacks are given in Figure 3.
There is no considerable performance difference of
OFs in terms of latency when they operate completely
in a benign network. However, they behave differ-
ently on networks under attacks. Version number and
DIS flooding attacks result in much more latency on
networks than worst parent attack, because of the ex-
tra overhead introduced by these attacks. Further-
more, in these attack scenarios, packet arrival times
are often delayed as the number of attackers increases.
Again here, OF0 shows much better performance than
the MRHOF objective functions. An interesting re-
sult here is that, MRHOF-ENERGY shows a slight
decrease in overhead and a decrease in latency in
the version attack scenarios, when the number of at-
tackers gets bigger than 6% of all nodes in the net-
work. Since it shows a big reaction to even a single
attacker (2%), this attack is less affected by the fur-
ther increase in the number of attackers. MRHOF-
ENERGY also performs a smaller rate of increase in
latency for DIS flooding attack, when the number of
attackers are increased up to 10% of all nodes. In the
future, this can be further explored by simulating net-
works with more attackers.
Average power consumption of simulated net-
works is demonstrated in Figure 4. Even when there
is no attacker in the network, MRHOF-ENERGY re-
sults in more power consumption than other objec-
tive functions. Again, MRHOF-ENERGY shows a
big jump in power consumption even when a single
attacker is introduced into the network. While the
MRHOF objective functions behave similarly under
attacks, MRHOF-ENERGY always consumes more
energy than MRHOF-ETX. Again, version number
and DIS flooding attacks consume much more en-
ergy than worst parent attack due to the higher num-
ber of control packets they introduced. Please note
that MRHOF-ENERGY prefer routes that consist of
nodes with higher remaining energy among alterna-
tive routes between the same end-points, it does not
aim to decrease energy consumption but to guarantee
the delivery of packets. Hence, it might compute a
longer path for some traffic in order to increase the
network life duration (Barthel et al., 2012).
To sum up, the presence of attackers adversely af-
fects the performance of networks as expected, espe-
cially, the effects of version number and DIS flood-
ing attacks on simulated networks are more obvious
as they inherently bring more control packets to the
network. On one hand, the worst parent attack which
aims to change the network topology does not cause
a dramatic effect on networks using different objec-
tive functions. On the other hand, version and DIS
flooding attacks that target consuming of network re-
sources clearly affect networks in all performance
metrics. However, networks using different OFs are
affected differently from such attacks. Networks us-
ing OF0 are more robust to attacks. Since attacks of-
ten negatively affect the quality of the links and the
energy of the nodes, thus affect the ETX and EN-
ERGY metrics used in the MRHOF objective func-
tions, attacks can cause more changes such as fre-
quent parent changes in networks using such objec-
tive functions.
5 CONCLUSION
This is the first study that comprehensively analyzes
the standardized RPL objective functions from the se-
curity point of view. Since objective functions play a
key role in determining optimal routes between end-
points, they might be affected differently from differ-
ent type of attackers. The comprehensive simulation
Analysis of RPL Objective Functions with Security Perspective
77
0 2 6 10
0
1
2
3
4
5
6
7
8
9
Total Overhead
10
4
a) Version Number Attack
0 2 6 10
Attacker Percentage
0
1
2
3
4
5
6
7
8
9
10
4
b) DIS Flooding Attack
0 2 6 10
0
1
2
3
4
5
6
7
8
9
10
4
c) Worst Parent Attack
MRHOF-ETX
MRHOF-ENERGY
OF0
Figure 1: Average overhead of simulated networks.
0 2 6 10
20
30
40
50
60
70
80
90
100
PDR (%)
a) Version Number Attack
0 2 6 10
Attacker percentage
20
30
40
50
60
70
80
90
100
b) DIS Flooding Attack
0 2 6 10
20
30
40
50
60
70
80
90
100
c) Worst Parent Attack
MRHOF-ETX
MRHOF-ENERGY
OF0
Figure 2: Average PDR of simulated networks.
0 2 6 10
0
1
2
3
4
5
6
7
8
Latency (sec)
a) Version Number Attack
0 2 6 10
Attacker Percentage
0
1
2
3
4
5
6
7
8
b) DIS Flooding Attack
0 2 6 10
0
1
2
3
4
5
6
7
8
c) Worst Parent Attack
MRHOF-ETX
MRHOF-ENERGY
OF0
Figure 3: Average latency of simulated networks.
0 2 6 10
1
1.5
2
2.5
3
3.5
4
4.5
5
Average Power Consumption (mW)
a) Version Number Attack
0 2 6 10
Attacker Percentage
1
1.5
2
2.5
3
3.5
4
4.5
5
b) DIS Flooding Attack
0 2 6 10
1
1.5
2
2.5
3
3.5
4
4.5
5
c) Worst Parent Attack
MRHOF-ETX
MRHOF-ENERGY
OF0
Figure 4: Average power consumption of simulated networks.
SENSORNETS 2022 - 11th International Conference on Sensor Networks
78
results confirm this hypotheses and show that OF0 is
more robust to attacks than MRHOF and, MRHOF-
ENERGY obtains the worst packet delivery ratio. The
results also show that attacks targeting network re-
sources (version number and DIS flooding attacks)
have a more clear effect on all network performance
metrics. We believe researchers developing security
solutions for RPL will benefit from the results pre-
sented in this current study. The study could be ex-
tended with more attack types and more parameters
such as mobility in the future.
REFERENCES
Abhishek, V. and Virender, R. (2020). Security of rpl based
6lowpan networks in the internet of things: A review.
IEEE Sensors Journal, 20(11):5666–5690.
Al-Kashoash, H. A., Al-Nidawi, Y., and Kemp, A. H.
(2016). Congestion-aware rpl for 6l0wpan net-
works. In 2016 Wireless Telecommunications Sympo-
sium (WTS), pages 1–6. IEEE.
Arıs¸, A. and Oktu
˘
g, S. F. (2020). Analysis of the rpl version
number attack with multiple attackers. In 2020 Inter-
national Conference on Cyber Situational Awareness,
Data Analytics and Assessment (CyberSA), pages 1–8.
IEEE.
Aris, A., Oktug, S. F., and Yalcin, S. B. O. (2016). Rpl
version number attacks: In-depth study. In NOMS
2016-2016 IEEE/IFIP Network Operations and Man-
agement Symposium, pages 776–779. IEEE.
Barthel, D., Vasseur, J., Pister, K., Kim, M., and Dejean, N.
(2012). Routing Metrics Used for Path Calculation in
Low-Power and Lossy Networks. RFC 6551.
Contiki-Ng (2004). contiki-ng/contiki-ng. [accessed 13-
July-2021].
Dunkels, A., Eriksson, J., Finne, N., and Tsiftes, N. (2011).
Powertrace: Network-level power profiling for low-
power wireless networks.
Gnawali, O. and Levis, P. (2012). The minimum rank with
hysteresis objective function. RFC 6719.
Jamil, M. Z., Khan, D., Saleem, A., Mehmood, K., and
Iqbal, A. (2019). Comparative performance analysis
of rpl for low power and lossy networks based on dif-
ferent objective functions. International Journal of
Advanced Computer Science and Applications, 10(5).
Karkazis, P., Leligou, H. C., Sarakis, L., Zahariadis, T.,
Trakadas, P., Velivassaki, T. H., and Capsalis, C.
(2012). Design of primary and composite routing met-
rics for rpl-compliant wireless sensor networks. In
2012 international conference on telecommunications
and multimedia (TEMU), pages 13–18. IEEE.
Kechiche, I., Bousnina, I., and Samet, A. (2017). A compar-
ative study of rpl objective functions. In 2017 Sixth In-
ternational Conference on Communications and Net-
working (ComNet), pages 1–6. IEEE.
Kim, H.-S., Ko, J., Culler, D. E., and Paek, J. (2017). Chal-
lenging the ipv6 routing protocol for low-power and
lossy networks (rpl): A survey. IEEE Communica-
tions Surveys & Tutorials, 19(4):2502–2525.
Le, A., Loo, J., Lasebae, A., Vinel, A., Chen, Y., and Chai,
M. (2013). The impact of rank attack on network
topology of routing protocol for low-power and lossy
networks. IEEE Sensors Journal, 13(10):3685–3692.
Mardini, W., Aljawarneh, S., Al-Abdi, A., and Taamneh, H.
(2018). Performance evaluation of rpl objective func-
tions for different sending intervals. In 2018 6th Inter-
national Symposium on Digital Forensic and Security
(ISDFS), pages 1–6.
Mayzaud, A., Badonnel, R., and Chrisment, I. (2016). A
taxonomy of attacks in rpl-based internet of things. I.
J. Network Security, 18:459–473.
Onwuegbuzie, I. U., Razak, S. A., and Isnin, I. F.
(2020). Control messages overhead impact on desti-
nation oriented directed acyclic grapha wireless sen-
sor networks objective functions performance com-
parison. Journal of Computational and Theoretical
Nanoscience, 17(2-3):1227–1235.
Osterlind, F., Dunkels, A., Eriksson, J., Finne, N., and
Voigt, T. (2006). Cross-level sensor network simu-
lation with cooja. In Proceedings. 2006 31st IEEE
Conference on Local Computer Networks, pages 641–
648.
Pancaroglu, D. and Sen, S. (2021). Load balancing for rpl-
based internet of things: A review. Ad Hoc Networks,
page 102491.
Perazzo, P., Vallati, C., Anastasi, G., and Dini, G. (2017).
Dio suppression attack against routing in the in-
ternet of things. IEEE Communications Letters,
21(11):2524–2527.
Pradittasnee, L. (2017). Improving performance of route
maintenance processes in rpl protocol. In 2017 9th
International Conference on Information Technology
and Electrical Engineering (ICITEE), pages 1–6.
Qasem, M., Altawssi, H., Yassien, M. B., and Al-Dubai,
A. (2015). Performance evaluation of rpl objective
functions. In 2015 IEEE International Conference on
Computer and Information Technology; Ubiquitous
Computing and Communications; Dependable, Auto-
nomic and Secure Computing; Pervasive Intelligence
and Computing, pages 1606–1613.
Semedo, F., Moradpoor, N., and Rafiq, M. (2018). Vulner-
ability assessment of objective function of rpl proto-
col for internet of things. In Proceedings of the 11th
International Conference on Security of Information
and Networks, pages 1–6.
Sousa, N., Sobral, J. V., Rodrigues, J. J., Rab
ˆ
elo, R. A.,
and Solic, P. (2017). Eraof: A new rpl protocol ob-
jective function for internet of things applications. In
2017 2nd International Multidisciplinary Conference
on Computer and Energy Science (SpliTech), pages 1–
5. IEEE.
Statistica (2016). Internet of things (iot) connected de-
vices installed base worldwide from 2015 to 2025 (in
billions). https://www.statista.com/statistics/471264/
iot-number-of-connected-devices-worldwide/. Ac-
cessed: 2020-04-01.
Analysis of RPL Objective Functions with Security Perspective
79
Thubert, P. (2012). Objective Function Zero for the Routing
Protocol for Low-Power and Lossy Networks (RPL).
RFC 6552.
Verma, A. and Ranga, V. (2019). Addressing flooding at-
tacks in ipv6-based low power and lossy networks. In
TENCON 2019 - 2019 IEEE Region 10 Conference
(TENCON), pages 552–557.
Winter, T., Thubert, P., Brandt, A., Hui, J. W., Kelsey, R.,
Levis, P., Pister, K., Struik, R., Vasseur, J.-P., Alexan-
der, R. K., et al. (2012). Rpl: Ipv6 routing protocol
for low-power and lossy networks. rfc, 6550:1–157.
Xiao, W., Liu, J., Jiang, N., and Shi, H. (2014). An op-
timization of the object function for routing protocol
of low-power and lossy networks. In The 2014 2nd
International Conference on Systems and Informatics
(ICSAI 2014), pages 515–519. IEEE.
SENSORNETS 2022 - 11th International Conference on Sensor Networks
80