ArticlePDF Available

Comparative Analysis of Selective Forwarding Attacks over Wireless Sensor Networks

Authors:

Abstract and Figures

The potential use of wireless sensor networks (WSNs) in many technological situations is extensive. WSNs are employed in numerous areas, such as battlefields, traffic surveillance, healthcare, and environmental monitoring. Many studies have been conducted to improve various aspects of the performance of WSNs, such as energy efficiency, quality of service (QoS), reliability, mobility, scalability, and extended network lifetime. However, few studies have sought to advance the security of WSNs by focusing on medium access control (MAC), physical, and transport stacks. The network layer is the middle layer that coordinates the lower layers and the upper layers. Thus, the network layer is of paramount significance to the security of WSNs to prevent exploitation of their confidentiality, privacy, availability, integrity, and authenticity. The objective of this study is to address the security laps of WSNs on the network layer, particularly selective forwarding attacks. This paper includes a survey performed from 2003 to 2014. Significant constraints and hazards of physical attacks at the network layer are extensively addressed. The survey includes a benchmark for the comparison of existing approaches for handling the security of WSNs on the network layer and their challenges. Future directions are also comprehensively highlighted in the document. This study will help researchers to understand attacks on the network layer. Keywords Wireless sensor network, selective forwarding, and denial of services attacks.
Content may be subject to copyright.
International Journal of Computer Applications (0975 8887)
1
Comparative Analysis of Selective Forwarding Attacks
over Wireless Sensor Networks
Naser M. Alajmi and Khaled M. Elleithy
Computer Science and Engineering Department, University of Bridgeport, Bridgeport, CT, USA
nalajmi@my.bridgeport.edu, elleithy@my.bridgeport.edu
ABSTRACT
The potential use of wireless sensor networks (WSNs) in
many technological situations is extensive. WSNs are
employed in numerous areas, such as battlefields, traffic
surveillance, healthcare, and environmental monitoring. Many
studies have been conducted to improve various aspects of the
performance of WSNs, such as energy efficiency, quality of
service (QoS), reliability, mobility, scalability, and extended
network lifetime. However, few studies have sought to
advance the security of WSNs by focusing on medium access
control (MAC), physical, and transport stacks. The network
layer is the middle layer that coordinates the lower layers and
the upper layers. Thus, the network layer is of paramount
significance to the security of WSNs to prevent exploitation
of their confidentiality, privacy, availability, integrity, and
authenticity. The objective of this study is to address the
security laps of WSNs on the network layer, particularly
selective forwarding attacks. This paper includes a survey
performed from 2003 to 2014. Significant constraints and
hazards of physical attacks at the network layer are
extensively addressed. The survey includes a benchmark for
the comparison of existing approaches for handling the
security of WSNs on the network layer and their challenges.
Future directions are also comprehensively highlighted in the
document. This study will help researchers to understand
attacks on the network layer.
Keywords
Wireless sensor network, selective forwarding, and denial of
services attacks.
1. INTRODUCTION
Wireless sensor networks contain numerous sensors. These
sensors communicate with a vast number of small nodes via
radio links. Sensor networks have a source and a base station.
The base station controls the sensor networks. They have the
ability to collect sensor data and send it to the base station. All
data flows between the nodes and the ends at the base station.
Nodes are densely deployed [1]. The positions of the nodes do
not need to be predetermined. Sensor nodes are deployed in
high-risk areas. The majority of WSN protocols do not have
the security to prevent simple attacks on the nodes [2]. Thus,
sensor network protocol and algorithms should be self-
organizing. Some design factors exist for sensor networks and
sensor nodes. These factors are significant in the design of
protocols or algorithms. The impact of these factors can be
used to compare different approaches [3]. The factors include
scalability, fault tolerance, network topology, power
consumption, production cost, hardware constraints,
environment, and transmission media.
The features of sensor nodes guarantee many applications.
The rapid deployment, self-organization, and fault tolerance
can provide a promising sensing method for certain functions,
such as control, communication, and computing [4], [5].
Networks have different applications. Therefore, applications
comprise several levels of monitoring, tracking, and
controlling. A group of applications is employed for specific
purposes. In military applications, sensor nodes include
monitoring, battlefield surveillance and object tracking. The
battlefield monitors utilized in military operations have
prompted the development of WSNs. In medical applications,
sensors aid in patient diagnosis and monitoring. The majority
of these applications are deployed to monitor an area and react
when a sensitive factor is recorded [6].
The security of wireless sensor networks has been extensively
investigated over the past few years. WSNs are susceptible to
many types of attacks because they serve as an open network
with limited resources of nodes. Therefore, the obstacles to
securing a wireless sensor network comprise the main
disadvantage for all devices. The most conventional threats to
the security of wireless sensor networks include
eavesdropping, node compromised, interrupt, modify or inject
malicious packets, compromised privacy and denial of service
attacks [7]. Sensor node can be simply compromised in
contrast with wired networks. A common attack in a WSN is a
DoS attack; the primary objective of the attacker in a DoS
attack is to make WSNs unavailable for users [8]. In a DoS
attack, the energy efficient protocols serve as targets in
wireless sensor networks. Sensor nodes are liable to physical
capture. Because the use of a sensor node as a target is
inexpensive, tamper-resistant hardware is unlikely to take
over. The denial of a service attack can be affected by
draining the energy of sensor nodes to prevent sensor nodes
from receiving authentic messages.
Even with physical damage to sensor nodes, which have
rendered a network inaccessible, the goal of the attacks is to
disrupt the network during the transfer of data [9]. The
majority of studies about WSNs can be classified as follows
[10]:
Key Management: Establishing and maintaining
cryptographic keys in an energy efficient manner to enable
encryption and authentication.
Secure Routing: Discovering new protection techniques and
applying them to new routing protocols without sacrificing
network connectivity, coverage or scalability.
Secure Services: Includes specialized security services such as
data aggregation, localization and time synchronization.
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
2
Wireless sensor networks are emerging as a central new stage
in the IT ecosystem and an area of active research involving
hardware and system design networking, distributed
algorithms, programming models, data management, security
and social factors [11], [12]. Wireless sensor networks are
widely employed in the area that would insure a specific task.
A wireless sensor network is an area of focus by researchers,
particularly in the security field. Therefore, one of the main
objectives of this survey is the comparison of existing
approaches for selective forwarding attacks in WSNs
employed by the researchers and developers in the security
fields. Sensor nodes have some impacts by selective
forwarding attacks. This survey suggests several resolutions
for wireless sensor networks in the security field. Section II
presents a background of WSNs. Section III describes the
denial of a service attack in layers. Section IV compares the
selective forwarding attacks between certain approaches.
Section V depicts the benchmarks for WSN. Section VI
presents the conclusion and recommendations for future
studies.
2. BACKGROUND
2.1 Security Goals for WSN
In wireless sensor networks, there are some requirements,
which represented the fundamental of security. These
requirements should be overall to protect sensitive data, which
ensures the operation of the sensor network. The objective of
security techniques in wireless sensor networks is to preserve
the data from attackers. Attacker motivation and
vulnerabilities and opportunities are two factors that give the
attacker potential influence over wireless sensor networks [2].
The security goals are categorized as major security goals and
minor security goals [13]. Confidentiality, data authentication,
data integrity, and data availability are major goals for the
security of WSNs. The minor goals include data freshness,
access control, and self-organization.
2.1.1 Data Confidentiality
Data confidentiality involves protecting the information
hideaway from attackers. The adversaries search for any data
in networks; thus, to obscure hidden data hidden from users
wishing to modify the data or illegal users, the data should be
encrypted with a secret key [14], [15] to ensure protection
from eavesdroppers.
2.1.2 Data Authentication
Message authentication is critical in numerous applications in
wireless sensor networks. Data authentication is the first
prevention that stops any illegal user from participating in the
network and original nodes must be distinguishable from
unauthorized nodes [14]. Data in this stage should be
confirmed for correctness.
2.1.3 Data Integrity
An attempt to modify data is the main issue of adversaries.
Therefore, during the transmission data integrity ensures that
a received message has not been changed.
2.1.4 Data Availability
If the base station and cluster header fails, the entire sensor
network is threatened. A node should be able to locate the
resources. Data availability is significant for networks.
Therefore, the desired network services should be available
even they exist under the denial of a service attack. DoS
attacks result in the loss of availability, which may have a
detrimental effect [7]. A WSN should introduce these services
in every case.
2.1.5 Data Freshness
Data freshness means that the data are not old. It makes sure
all data or messages that are not fresh have not been relayed.
Therefore, the data must be recent. Although data integrity
and confidentiality help to assure data, all messages should
attain optimum freshness. The malicious node does not reply
to or resend old or previous data [2].
2.1.6 Access Control
Access control restrains the resource that is unauthorized to
enter in networks. It has to construct the networks to prevent
admittance by unauthorized users.
2.1.7 Self Organization
Sensor nodes must be elastic to be self-organizing and self-
healing. The lack of self-organization in a wireless sensor
network is risky. Thus, any damage to the output from an
attack may be destructive.
!
Fig 1: WSN Security Requirements
!
2.2 Constraints for WSN
The limited resources of WSNs include memory, energy,
processing power, and transmission channels. Sensor nodes
have limited storage and limited communication bandwidth
due to the small size of wireless sensor networks. Resource
constraints represent the obstacles in wireless sensor
networks. Energy and memory are not appropriate for
traditional security measures. The condition of resources must
assess in WSNs to develop security algorithms.
2.2.1 Limited Memory
Sensor node memory typically includes flash memory and
RAM. A sensor is a tiny device with minimal memory and
storage space. Thus, a small code size with a few lines should
be used for sensor operation and security due to the small size
of the memory’s storage space [15][62]. In a smart dust
project, TinyOS reserve 4500 bytes for security and 4 K bytes
for consumer applications [16]. A common sensor type
TelosBhas a 16-bit, 8 MHz RISC CPU with 10 K RAM,
1024 K flash storage and 48 K program memory. Therefore,
the algorithms are impractical in sensors [17].
2.2.2 Limited Energy
Sensor nodes use a limited battery. Thus, locating the sensor
node in difficult areas such as a battlefield, which has a low
battery, is not feasible. Sensors are deployed in dangerous
areas in which the battery cannot be modified or exchanged.
The energy consumption in sensor nodes can be classified in
several parts: communication among sensors, sensor
transducers, and microprocessor computation. Power is
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
3
required for executing 800 to 1000 instructions, in which each
bit is transmitted in a WSN [16]. The battery power is limited
in sensor nodes. The cost of energy varies. In addition,
security levels in WSNs correspond to the energy
consumption [18]. In wireless sensor networks,
communications and calculations consume the majority of the
available energy [6]. Limited energy impacts the process
speed and capability of a sensor node.
2.2.3 Unreliable Communication
Sensor nodes use communication to send packets by multi-
hop transmission [19]. Unreliable communication hinders the
sensor network. A sensor network is susceptible to corruption
or damage. Weak packets can cause congested nodes and
channel errors [13]. The packets may conflict during a transfer
[20].
2.2.4 Higher Latency in Communication
In a wireless sensor network, multi-hop routing and higher
latency during transmission may complicate synchronization,
which may cause congestion of the network. The security
technique is dependent on cryptographic key distribution [21].
3. DENIAL OF SERVICE ATTACKS IN
WSN LAYERS
Wireless sensor networks are very weak and susceptible to
many types of security attacks on the broadcast. Security
attacks are classified into two classes: passive attacks and
active attacks [22]. In passive attacks, the attacker
compromises data confidentiality. In active attacks, the
malicious action targets data confidentiality, data integrity,
and unauthorized access. In WSNs, attacks occur in two forms
based on the type of hardware utilized by the attacker to
compromise the network [2]: mote class attackers and laptop
class attackers. Mote class attackers have access to some
sensor nodes, such as regular sensor nodes. Malicious nodes
are usually gained during node-compromised activities.
Laptop class attackers have access to more powerful devices,
such as laptops, smart phones, and workstations. The attacks
can be classified into two types of attacks: outsider attacks
and insider attacks. In an outsider attack, the adversary cannot
gain any type of access to the network communications but
the network must be pervaded before the attack can be
detected. In an insider attack, the adversary gains legitimate
and authorized access to the network by neighboring nodes.
A DoS attack is one of the security attacks that impacts
wireless sensor networks. It is dependent on the vulnerability
of each layer. A DoS attack is a multilayer attack that can
launch from any layer in the network. Different types of DoS
attacks can impact sensor nodes and can also affect idle nodes
or standby mode. A DoS attack comprises the main energy
consumption attack in WSNs. Thus, its main focus is network
resources. It can also disrupt wireless transmission and can
occur either unintentionally in the form of interference, noise
or collision at the receiver side or at in the context of an attack
[7]. Attackers must reach certain targets, such as network
access, network infrastructure, and server applications. The
DoS attack attempts to drain the resources available to the
victim node by transferring unnecessary data. Therefore, it
prevents users from accessing services. This type of attack is
intended not only for adversaries who wish to subvert, disrupt,
or destroy a network but also for any event that diminishes a
network’s capability to provide a service [23]. Thus, a DoS
attack renders a service unavailable for the users. Several
types of attacks to the networks exist, such as consumption of
bandwidth or processor time, disruption of service to a system
or user, and disruption of physical components.
Karlof [2] noted the names of some types of DoS attacks, such
as spoofed data, selective forwarding, the sinkhole attack, the
Sybil attack, the wormholes attack, the hello flood attack, and
acknowledgment spoofing. Therefore, DoS attacks are severe
impacts on most layers of sensor networks. Because a DoS
attack comprises a dangerous threat to a wireless sensor
network, studies have explored various mechanisms to detect
these types of attacks. The important aspect of a DoS attack is
the identification of the nodes that are harmed by the
adversary.
There are some methods to prevent DoS attacks such as
payment for network resources, pushback, strong
authentication and identification of traffic [23]. Some
mechanism can be implemented to safeguard the
reprogramming process, for example, authentication flows.
The choice for a DoS attack is the recreate the key request
packet, which is came from the sensor node while the lifetime
of the keys has expired. If the rate of recreate the key requests
is frequently, the sink can conclude the occurrence of a DoS
attack and drop the packet from the node for a configurable
period of time [24].
Karlof and Wagner [2] have proposed a design for sensor
network security and presented different types of threats in ad
hoc networks that may reverse the sensor security. Arazi, Qi,
and Rose have proposed a RSA based on a framework to
prevent DoS attacks and ensure that malicious nodes exhaust
the resources [25]. Advancements in wireless sensor networks
are being achieved in several fields. Therefore, security in
WSNs is an active research field [26].
Denial of services is illustrated for several types of attacks at
different layers. At the physical layer, a DoS attack may cause
jamming and tampering. At the network layer, it may cause
black holes, spoofing, replying, and homing. At the link layer,
it may cause collisions and unfairness. At the transport layer,
it may cause flooding and de-synchronization [27].
3.1 Physical layer
The physical layer is responsible for the establishment of
certain functions, such as connection, modulation, data rate,
data encryption, and signal detection. In addition, a physical
layer in a network may increase the reliability by reducing the
path loss effect and shadowing. Attacks that affect the
physical layer include jamming and tampering.
3.1.1 Jamming
Jamming is a type of DoS attack in the physical layer.
Because the physical layer is responsible for frequency
selection, carrier frequency generation, signal detection, and
data encryption and is the lowest layer, it is the first layer to
be attacked via jammers [28]. Malicious nodes use the same
frequency in the network. Therefore, jamming the entire
network will damage the network. An attacker attempts to
transfer many packets in different paths to jam the networks.
In addition, limited resources lure the adversaries to attack
networks.
3.1.2 Tampering
Tampering involves circuitry or injecting fabricated code in a
legitimate node. The intruder captures the node by changing
its programming code. Thus, attackers may attempt to acquire
sensitive information, such as the cryptography key from a
node, by destroying it to gain access to a higher level of
communication [29]. It affects the physical layer when the
sensor nodes are unattended after deployment.
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
4
3.2 Data link layer
The guarantee interoperability communication between sensor
nodes is the main objective of the data link layer. It is
responsible for error detection, multiplexing, prevention of
collision of packets and transmission. The data link layer
impacts security attacks as follows:
3.2.1 Collision
A collision attack executes between two nodes during
transmission in a channel. It occurs when the two nodes
attempt to transmit using the same frequency. It affects all
transmitted packets. The error correction code can be
employed to protect the link layer from this type of attack
[30].
3.2.2 Medium Access Control (MAC)
MAC layer protocols are designed for wireless sensor
networks. A MAC protocol employs different algorithms to
conserve battery power [31]. Using low power modes for a
radio conserves the battery power when sending or receiving
data. The main attack in a MAC protocol is the denial of sleep
attacks. The denial of sleep attacks is against the S-MAC
protocol [32], which employ a static cycle. Sensor nodes in
this protocol are organized into virtual clusters using SYNC
messages. Using this protocol, radio networks will be asleep
90% of the time. The T-MAC protocol [33] is superior to the
S-MAC protocol by focusing all traffic at the beginning of the
period. Because it used the same SYNC technique, the
technique enables nodes to transition to sleep mode. B-MAC
[34] used a technique that is referred to as low-power listening
(LPL). This technique is used to reduce energy consumption.
G-MAC [35] is an energy efficient MAC protocol that is
designed to coordinate transmissions in clusters.
3.3 Transport layer
The main objective of the transport layer is to provide
reliability and congestion. Many protocols are designed to
provide these two tasks. However, they employ different
techniques.
3.3.1 Flooding
Flooding attack is a problem that impacts the transport layer.
The attacker establishes a connection request until resources
are drained. Connection via a puzzle is a potential solution
[29].
3.3.2 De-synchronization attack
De-synchronization is another attack in the transport layer.
This attack occurs when there is connection between two
endpoints. Therefore, the adversary creates inaccurate
messages at the endpoints [29].
3.4 Application layer
Collection, management and processing of the data are the
main functions of the application layer. The objective of the
application layer is to render the final output. Therefore, it
ensures that the information flows to lower layers. The
application layer consists of user data and supports many
protocols, such as HTTP, FTP, SMTP, AND TELNET, which
provides vulnerability to attacks. The major attack in the
application layer is an attack on reliability and data
aggregation distortion. The adversary in this attack needs to
determine the communication route to alter data using that
route. In addition, the adversary creates faulty data via
network connections. Therefore, sensor nodes will be harmed
by the energy consumption attack when responding to the
false data. Ensuring reliability acknowledgment for all
received data is the main task of security to prevent attacks on
reliability [36].
3.5 Network layer
The goal of the network layer is to establish a path for
efficient routing techniques. Therefore, the main function in
the network layer is routing. The network layer is also
responsible for some network tasks, such as routing the data
between nodes, routing the data from a node to the base
station, and routing data between a node and a cluster head.
Several challenges at the network layer exist based on
applications. These challenges include limited memory,
efficiency, reliability, scalability, and energy consumption
[37]. The attack of the network layer from several types of
attacks impacts these challenges. Routing in the network layer
may have caused one of these attacks. Consequently, most
network layer attacks against sensor networks may be
categorized as one of the following attacks: spoofing or
replaying information, selective forwarding, the sinkhole
attack, the Sybil attack, and the wormhole attack.
3.5.1 Spoofing or replaying information
This type of attack directly impacts routing information.
Creating routing loops, extending or shortening service routes,
generating false error messages, and increasing end-to-end
latency are caused by the spoofing attack [2].
3.5.2 Selective forwarding
In networks, attackers focus on two types of attacks: a data
attack and a routing attack. A selective forwarding attack is a
type of data attack, in which a compromised node selectively
drops a packet. Thus, a selective forwarding attack may
damage the network efficiency. This type of attack can be
considered to be a special case of black hole attack. Karlof
and Wagner [2] described a selective forwarding attack. They
noted that a selective forwarding attack is also known as a
malicious node. It is located in the path of data flow that can
reject a certain or particular sensitive message that originates
from other nodes to forward it to the base station [9].
Subsequently, a malicious node can cut off certain nodes from
the base station. A malicious node selectively drops sensitive
packets, such as a packet that reports enemy tank movement.
Malicious sensor nodes can function as normal sensor nodes
[38]. Bysani and Turuk [19] characterize selective forwarding
attacks into two classes: drop packets in certain nodes and
drop packets of certain types.
Fig 2: Sensor network under selective forwarding attacks
Redrawn [39]
Military surveillance is very important in battlefields. In the
military applications, a sensor node is battery powered and
has a constraint of sensing. It also has wireless
communication abilities. When need to follow activities, such
as tank movements, (Figure 2) sensor readings gather
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
5
information to create a report. The report will be sent to the
base station. Therefore, it can transfer some actions to the
objective area. Errors in these actions are possible via a
selective forwarding attack. In a selective forwarding attack,
malicious nodes serve as standard nodes.
Fig 3: Selective Forwarding based on node Redrawn [39]
Three fundamental methods apply to any malicious node
(Figure 3) [39]. These methods can impact networks. The first
method involves a single malicious node. The node, which is
located in the middle of the path, can selectively forward
packets to the base station. The second method involves two
consecutive malicious nodes. In this method, packet dropping
is hard to detect because more than one malicious node is
forwarded. The last method involves the surrounding
malicious nodes, which can reject any packet to the base
station.
Fig 4: Sinkhole attack in wireless sensor networks. (a)
Using an artificial high quality route; (b) Using a
wormhole Redrawn [49]
3.5.3 Sinkhole attack
One of the DoS attacks types is sinkhole attack. Sinkhole
attack is a significant attack. It makes the base station to
acquire the whole data and thus create a severe major threat to
layer application (Figure 4). Sinkhole attacks comprise one
type of network layer attack. In addition, a compromised node
sends false routing information to its neighbors to attract
network traffic to itself [40]. In a sinkhole attack, the
attacker’s objective is to tempt the majority of the traffic from
a specific location in the network via a compromised node,
which generates a metaphorical sinkhole with the attacker at
the center [2]. For example, when the node at the coordinator
is attacked from an adversary, all other nodes follow it into
the sinkhole.
The major chance in a sinkhole attack is eavesdropping. A
compromised sensor node attempts to impact the information
sent to it from any neighboring node. Therefore, a sensor node
eavesdrops on the information is being communicated to its
neighboring sensor nodes. Sinkhole attacks ordinarily
function by establishing a compromised node that seems
attractive to surround the nodes with regard to the routing
metric. For instance, the attacker could spoof or reply to an
advertisement for a very high quality route to the base station
[2]. Karlof and Wagner noted that the sensor nodes are
vulnerable to the sinkhole attack based on the communication
design. Some security attacks, such as selective forwarding or
eavesdropping, can be started during a sinkhole attack. The
malicious node or even the attacker can do anything in the
network as long as the data are routed through the malicious
node. WSNs are vulnerable to sinkhole attacks because many
nodes transfer data to a single base station [41]. A sinkhole
attack is always searching for nodes located near the base
station.
3.5.4 Sybil attack
The Sybil attack is defined as a malicious device that assumes
numerous identities [42]. Malicious nodes can allege to have
many identities. WSNs are vulnerable to the Sybil attack. In
this a case, a node can act as more than one node using
different identities of legitimate nodes. Therefore, a single
node presents multiple identities to other nodes in the network
[43]. The Sybil attack has attempted to degrade the integrity
of data, security and resource utilization [11]. James et al. [44]
developed a classification for the Sybil attack. They presented
direct vs. indirect communication, fabrication vs. stolen
identities, and simultaneity as three dimensions. They
proposed defenses against the Sybil attack in sensor network,
including radio resource testing, verification of key sets for
random key pre-distribution, and registration and position
verification.
3.5.5 Wormhole attacks
This type of attack is an important and also it is a dangerous
attack. The attacker could record a packet at a single location
in the network, tunnels them to another location, and resends
them into the network [2]. The attacker can replay messages
to any part of the network. In wormhole attacks, malicious
nodes can create a hidden channel between sensor nodes [45].
A wormhole attack is an important threat to a wireless sensor
network because this type of attack does not require that a
sensor in the network be compromised. This type of attack
affects the network layer by continuously hearing and
recording data [46]. It can be implemented in the initial phase
when the sensor launches to discover information.
A wormhole attack is not easy to detect because an attacker
uses a private band, of which the network is not aware [2].
The technique for detecting a wormhole attack was proposed
by Perrig et al. [45]. It is based on packet leashes, and a
message includes a timestamp and the location of the sender.
However, it requires strict time synchronization and is
infeasible for most sensor networks. Wormhole and sinkhole
attacks are sometimes combined to attack the sensor
networks. These two types of attacks render the networks hard
to defend against attacks [2].
3.5.6 HELLO flood attacks
The attackers in HELLO flood attacks send or replay a routing
protocol [2]. This protocol consists of HELLO packets, which
transmit between sensor nodes with extra energy. HELLO
packets are employed as a weapon to encourage the sensors in
WSNs. The victim nodes attempt to go through the attacker
because they think that the attacker is their neighbor [2].
base station
(a) Single malicious node (b) Two consecutive malicious nodes (c) Surrounding malicious nodes
Figure 2. Deployment of malicious nodes.
versary only needs to ensure the presence of one compro-
mised node in each path. Traditional transport layer pro-
tocols [11,12] for WSNs also fail to guarantee that packets
are not maliciously dropped. They are not designed to deal
with malicious attacks.
In this paper, we propose a lightweight security scheme
that detects selective forwarding attacks by using a multi-
hop acknowledgement technique that increases detection
accuracy yet lowers overhead. The scheme allows both the
base station and source nodes to collect attack alarm in-
formation from intermediate nodes. This means that even
when the base station is deafened by surrounding malicious
nodes, the source nodes can still make decisions and re-
sponses. The scheme can efficiently obtain those alarm
information whenever intermediate nodes in a packet for-
warding path detect any malicious packet dropping. Simu-
lation results show that the communication overhead of our
scheme is usually less than 2 times the overhead of the com-
mon one-path packet delivery process, and the detection ac-
curacy is over 95% even when the channel error rate is a
harsh 15%. To the best of our knowledge, this is the first
paper that presents a detailed detection scheme in response
to selective forwarding attacks.
The remainder of this paper is organized as follows. Sec-
tion 2 introduces the attack model and our design goals.
Section 3 presents our detection scheme based on multi-hop
acknowledgement in detail. Section 4 first proposes several
evaluation metrics for our detection scheme and then shows
the simulation results in terms of these metrics. Finally, we
introduce the related work in Section 5, and conclude our
work in Section 6.
2. Attack Model
We consider a military application of sensor networks
for reconnaissance of opposing forces, as shown in Figure
1. Each sensor node is battery-powered and has limited
sensing, computation and wireless communication capabili-
ties. When the activities of the opposing forces such as tank
movement are detected, sensor readings are aggregated to
generate a report, which will be forwarded to the base sta-
tion through multi-hops. The sink is a data collection center
equipped with sufficient computation and storage capabili-
ties. Once the base station receives the report, it can take
action by, for example, sending soldiers or missiles to the
target field.
In a military application such as this, prompt detection
and reporting of each relevant event in the field are impor-
tant, but these processes can be easily corrupted by selective
forwarding attacks. In such attacks, malicious nodes may
refuse to forward certain packets and simply drop them, en-
suring that they are not propagated any further. An adver-
sary will not, however, drop every packet. To avoid raising
suspicions, the adversary instead selectively drops packets
originating from a few selected nodes and forwards the re-
maining traffic. As shown in Figure 1, the adversary may
attack in two ways, from inside the network via compro-
mised nodes or from outside the network by jamming the
communication channels between uncompromised nodes.
Figure 2 shows three basic ways in which malicious
nodes can be distributed in a network for different tacti-
cal purposes. Figure 2(a) shows a single malicious node
located in the middle of a forwarding path. This node can
selectively forward packets to the base station. Figure 2(b)
shows two or more malicious nodes chained along a for-
warding path. This can make it more difficult to detect
packet dropping. Figure 2(c) shows a number of compro-
mised nodes surrounding a base station. This arrangement
can be used to deafen a base station by refusing to forward
any packets at all.
In this paper, our goal is to design a scheme that detects
selective forwarding attacks and identifies malicious nodes.
Once the ids of suspect nodes are known, routing proto-
cols can exclude them from routing paths. Furthermore,
professionals can be sent to the battlefield to examine the
nodes physically and even physically remove the compro-
mised nodes. The detection scheme should have the follow-
ing properties. First, the scheme should be able to quickly
detect any malicious packet dropping. Second, detection
accuracy should be guaranteed even when radio conditions
are poor. Finally, the scheme should cause little additional
communication overhead.
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
6
Fig 5: Taxonomy of attacks
4. SELECTIVE FORWARDING
ATTACKS –DETECTION APPROACHES
In this section, the discussion of countermeasure for selective
forwarding attacks is the main goal of this survey. The denial
of service attack creates assortments to attack wireless sensor
networks. These assortments may temper sensor nodes and
the function of networks. Consequently, some attackers target
layers, such as the physical layer, the network layer, the link
layer, and the transport layer, and some attackers target the
routing layer. A DoS attack may occur in any layer of an OSI
layer. All DoS attacks are dependent on the vulnerability of
each layer in the architecture of wireless sensor networks. In a
DoS attack, adversaries attempt to decipher a system but are
unsuccessful. The selective forwarding attack is such an
attack.
A selective forwarding attack is hard to detect due to
unreliable sensor wireless communications. Karlof and
Wagner [2] discussed the selective forwarding attack. In this
type of attack, malicious nodes have attempted to stop the
packets in a network by rejecting message forwarding.
According to [2], selective forwarding attacks can impact
some multi-hop routing protocols, such as TinyOS beaconing,
DSR, PSFQ, directed diffusion and its multipath variant, and
geographic routing (GPSR and GEAR). During the launch of
a selective forwarding attack, a compromised node has
notable prospects, including itself, along the path of data.
Based on previous studies, this type of attack makes the
sensor network rely on the redundancy forwarding via
broadcast for data to spread during the network.
Karlof et al. [2] suggested the prevention of selective
forwarding attacks by counting the selective forwarding
attacks using multipath routing between nodes, which has
disadvantages. Communication overhead is increasing, which
causes an increase in the number of paths [39]. In addition,
the security of WSNs cannot be resilient. The malicious node
forwards several packets to the neighbors but drops some
them, which causes a significant loss of data. The Low Energy
Adaptive Clustering Hierarchy (LEACH) protocol, which was
improved by Wu, Hu, and Ni to the SS-LEACH algorithm,
can prevent a selective forwarding attack using a sequence
number. Thus, the cluster head is responsible for sending a
packet. In addition, the cluster head can discover the attack
and send a warning to the base station [47].
Yu and Xiao [39] proposed a new approach based on
lightweight security to detect a selective forwarding attack in
the environment of sensor networks. The detection utilized a
multi-hop acknowledgment to launch alarms by obtaining
responses from the nodes that are located in the middle of
paths. Yu and Xiao assumed that the approach could identify
malicious sensor nodes. The aim of the approach is to send an
alarm when a malicious node is discovered, which indicates a
selective forwarding attack. The authors noted that the
detection accuracy of their approach exceeds 95% with an
error rate of 15%.
The detection approach enables the base station and source
nodes to aggregate attack alarm information from a specific
node. Thus, the specific node that is located in the middle of a
route can detect an attack and send an alarm. The detection
contains three types of packets during transmission: report,
ACK, and alarm. When the node sends a packet, three values
are reported: ACK_Cnt, ACK_Span, and ACK_TTL.
Fig 6: An example of multi-hop acknowledgement with
ACK_Span=3, ACK_TTL=6. -Redrawn [39]
Yu and Xiao employed two detection processes in the
scheme: a downstream process (the direction on the way to
the base station) and an upstream process (the direction on the
way to the source node). In the upstream process, a report
packet is created and sent to the base station hop by hop when
nodes detect a malicious node. ACK_Cnt is set to ACK_Span,
which is a predefined metric. The node that is referred to as
the intermediate node saves the packet report as soon as it is
received in its cache decreases the ACK_Cnt by one or resets
ACK_Cnt to its initial value. The packet report may be sent
downstream if ACK_Cnt is 0. Simultaneously, an ACK
packet is created and the TTL in the ACK packet is set to
ACK_TTL, which is a predefined metric. In the remaining
detection process, which occurs downstream, packet loss may
occur if the intermediate node receives a report packet that
should have Packet_ID for a certain source node. In this case,
the node creates an alarm packet, in which
Lost_Packet_ID_Beg and Lost_Packet_ID_End describe the
range of the lost Packet_IDs, and Suspicious_Node_ID is set
to the upstream node, where the report with the discontinuous
Packet_ID originated. The base station will receive the alarm
packet and forward multiple hops that are produced by the
node.
The detection accuracy is increased by a detecting selective
forwarding attack scheme. Although the radio frequency
status is poor, detection accuracy is guaranteed. The scheme
authorizes the base station and source nodes to collect attack
BS S
u1
u2u3u4u5u6u7u8
ACK
ACK
ACK
ACK
u9
ACK
compromised nodesuncompromised nodes where ACK dropped
Figure 3. An example of multi-hop acknowledgement with ACK Span =3,ACK TTL =6. Node
u3,u6,u9are ACK nodes, which are required to send out ACK packets.
3. A Multi-hop Acknowledgement-Based De-
tection Scheme
In this section, we present the design of our multi-hop
acknowledgement-based detection scheme. In our scheme,
each intermediate node along the forwarding path is in
charge of detecting malicious nodes. If an intermediate
node detects the misbehavior of its downstream (upstream)
nodes, it will generate an alarm packet and deliver it to
the source node (the base station) through multiple hops.
In this paper, downstream denotes the direction toward the
base station, and upstream denotes the direction toward the
source node. The base station and the source node can then
use more complicated IDS (Intrusion Detection System) al-
gorithms to make decisions and responses.
3.1. Assumptions
The following five assumptions are appropriate to use of
the proposed detection scheme in a mission-critical applica-
tion such as military reconnaissance, as opposed to a civil-
ian application such as temperature monitoring. Our first
assumption is that during the deployment phase each sen-
sor can acquire its geographical position and loosely syn-
chronize its time with the base station. Secure positioning
and time synchronization discussed in [15,16,17] are also
required by other purposes in a military mission, so our
assumption does not add additional overhead. Note that
our scheme can function over various existing routing al-
gorithms but does not rely on geographic routing. Second,
we assume that the adversary cannot successfully compro-
mise a node during the short deployment phase. Some exist-
ing work [7] has made similar assumptions and argued that
such attacks can indeed be prevented in real-life scenarios
when appropriate network planning and deployment keep
away attackers during the bootstrapping process. Third, we
assume that malicious nodes, in order to allay suspicions,
selectively drop only a small proportion of all packets pass-
ing by rather than every packet. Fourth, we assume that
each node shares a master secret key with the base station
and a node can establish a pairwise key with another node
that is multiple hops away. Finally, we assume that rout-
ing and transport protocols such as Directed Diffusion[9]
and PSFQ[10] have been implemented in sensor nodes. Our
scheme can function over these protocols.
Although the routing layer of WSNs is threatened by
various attacks, here we are considering only selective for-
warding attacks. It is out of the scope of the paper to con-
sider the issue of verifying whether an sensor report is mod-
ified or injected. The interested will find discussions of this
in [4,5,6,7]. Nor do we consider the detection of link-layer
jamming attacks [14], which are also able to cause packet
loss.
3.2. Node Initialization and Deployment
Before deployment, the key server loads every node with
a unique secret key and a symmetric bivariate polynomial
f(u, v).
The unique key is shared with each sensor node and the
base station and can be used to encrypt sensor reports and
generate MACs (Message Authentication Codes) for the re-
ports.
The symmetric bivariate polynomial is used to estab-
lish a pairwise key between any two sensor nodes in
the network. Before deployment, the key server gener-
ates a symmetric bivariate k-degree polynomial f(u, v)=
Σk
i,j=0 aij uivjover a finite field Fq, where qis a prime num-
ber that is large enough to accommodate a cryptographic
key. A polynomial f(u, v)is said to be symmetrical if
f(u, v)=f(v, u). The key server then loads each node
with this polynomial. After deployment, each node regen-
erates a polynomial g(x)using its node id:g(x)=f(id, x),
and erases f(u, v)from its memory forever. Suppose that
node aowns ga(x)=f(a, x), node bowns gb(x)=
f(b, x), and node awants to send a packet to node b. First
node acalculates k1=ga(b)as the key to generate a MAC
for the packet and then sends the packet together with the
MAC and its id to node b. After receiving the packet, node
bcalculates k2=gb(a),(k1=k2), as the decryption key
and verify the MAC. If the verification is passed, node b
believes the packet comes from the authentic node a.
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
7
alarm information from the intermediate nodes. Because the
sensor node requires more effort, it will impact the efficiency
of scheme. The scheme cannot develop a countermeasure for
other types of attacks; thus, the scalability will be decreased.
Fig 7: Suspect nodes identification-Redrawn [48]
The identification of suspect nodes is reported via an
intermediate node. Figure 7 provides an example of a node
that is suspect and detected by an intermediate node. First,
Xiao, Yu, and Gao [48] proposed a checkpoint-based method.
In this approach, a node is randomly selected as the
checkpoint to send an acknowledgement message for
detecting the adversary. It is a mechanism used to identify
suspect nodes in a selective forwarding attack. They have
attempted to improve the technique by detecting an abnormal
packet in sensor networks [39]. They assumed that any
compromised nodes could not create alert packets with the
aim of maliciously prosecuting other nodes. In the previous
example, node µ3 produced an alert packet to prosecute node
µ4. Thus, the prosecuted node is a compromised node. After
collecting evidence to determine whether the node is a
malicious node, the source nodes determine the position of the
suspect node according to the location. However, it is no
guarantee for reliable transmission of messages even though
the adversary is positioned by acknowledgement. An
acknowledgement packet, an alert packet and a one-way key
packet will drain the energy during detection [49].
Fig 8: Original sensor nodes after divided into two
dataflow topologies-Redrawn [50]
Hung-Min, etal. [50] proposed a multidataflow topology
(MDT) scheme against a selective forwarding attack. This
scheme divided networks into various data topology, in which
a sensor node can communicate with other nodes in the same
topology. In figure 6, the sensor nodes are divided into two
topologies via the base station. The dataflow topology
contains A and B; they encompass the monitored area. One
report derives from A or B to control the entire project. The
sensing information can be retained via the topology if it does
not contain a malicious node; for example, if the attack occurs
in topology A, topology B is to transfer the information to the
base station because the two topologies are overlapped.
The MDT scheme has some advantages and disadvantages in
its defense against a selective forwarding attack. The main
advantage of this scheme is that the base station can obtain
packets even if the networks are attacked via a selective
forwarding attack. Thus, the base station continues to receive
information although the node is rendered malicious by an
attacker. The disadvantages of a MDT scheme are the high
network cost and the low network lifetime. The high cost is
attributed to the division of the network into additional data
topologies. The network lifetime is low due to multiple data
transmissions.
Xin etal, [49] proposed a lightweight defense against a
selective forwarding attack. This defense is dependent on the
neighboring nodes. Thus, neighboring nodes monitor the
packets during packet transmission, assess the attacker’s
location and retransfer the packets dropped by the attackers.
As a result, they suggested that this approach consumes less
energy and storage. Its efficiency in detecting selective
forwarding attacks ensures packet delivery to the base station.
The defense scheme employs a hexagonal WSN mesh
topology. According to the hexagonal mesh topology, the
authors specified some processes of topology construction,
such as node initialization, cell partition, active node election,
and secure architecture construction. In node initialization, the
node determines its location and its neighbors’ locations. In
cell partition, the node determines the association with the
RC. Active node selection involves contact with nodes of
other RCs. The communication relation is determined
between RCs to establish and construct a secure architecture.
Fig 9: An example for the monitor node gets the attacker-
Redrawn [49]
Xin, etal, [49] discussed the two phases of a defense scheme.
The first phase is routing discovery and selection. The second
phase is data transmission with attack defense. Based on the
network model, routing discovery and selection method are
designed to prevent selective forwarding attacks. It calculates
the number of hops between the source node and the
destination node. According to the policy of the probability, it
selects the transmission route. A method is employed to
B. Xiao et al. / J. Parallel Distrib. Comput. 67 (2007) 1218– 1230 1225
2 3 4 5 6 7 8 9 10
0.5
0.6
0.7
0.8
0.9
1
Number of malicious nodes (m)
Detection Probability
q=0
q=2
q=4
2 3 4 5 6 7 8 9 10
0
0.2
0.4
0.6
0.8
1
Number of malicious nodes (m)
Detection probability
k=1
k=2
k=3
2 3 4 5 6 7 8 9 10
0.4
0.5
0.6
0.7
0.8
0.9
1
Number of malicious nodes (m)
Detection probability
k=2, α= 0.05
k=2, α= 0.1
k=2, α= 0.15
k=3, α= 0.05
k=3, α= 0.1
k=3, α= 0.15
Fig. 5. Impact of q,k,!and mon detection probability: (a) impact of qand m, given k=1, !=0.1, and n=100; (b) impact of kand m, given q=0,
!=0.1, and n=100; (c) impact of !and m, given q=0 and n=100.
along a forwarding path. Second, k!2 is the minimal security
requirement for our scheme. Finally, under perfect channel con-
ditions, the selection of fewer checkpoint nodes produce better
detection performance.
4. Identification of suspect nodes
In this section, we discuss the identification of suspect nodes.
Since this paper does not focus on the decision and response
phase introduced in Section 2.3, our purpose here is to discuss
the policy which identifies suspect nodes in terms of a single
alert packet reported by an intermediate node. This policy al-
lows the source node, after receiving enough evidence, to make
some more tactical decisions based on this policy, such as re-
ducing the traffic going through the suspect nodes.
We take Fig. 6 as an example showing the identification of
suspect nodes. First, we suppose that compromised nodes do
not generate alert packets with the goal of maliciously pros-
ecuting other normal nodes. Suppose that node u4 is a com-
promised node. Node u3 generates an alert packet prosecuting
node u4 as a suspect node. In this case, the prosecuted node is a
compromised node. Next, we suppose that compromised nodes
are smart and can generate alert packets in order to maliciously
prosecute innocent neighboring nodes. For instance, node u4
prosecutes node u5 by generating an alert packet. In this case,
the prosecuting node is a compromised node, while the prose-
cuted node in the packet is innocent. As long as it remains unde-
tected, it is a simple matter for a compromised-yet-undetected
node to maliciously prosecute an innocent neighboring node. It
is quite difficult, however, to distinguish between the compro-
mised and the innocent. Consequently, in practice, in terms of
a single alert packet, we should tag both the prosecuting node
and the prosecuted node in the alert packet as suspect nodes.
After collecting enough evidence to make decisions, the
source nodes can derive the geographical position of suspect
nodes based on the location-binding ID technique introduced
in Section 2.2. As future work, we are going to develop
more considerate decision and response mechanisms for the
source nodes or the base station, based on the information
from
source
to base
station u1
u2
u3
u4
u5
u6
malicious node
uncompromised node threatened area
Fig. 6. Identification of suspect nodes.
(alert packets) provided by the intrusion detection scheme
proposed in this paper.
Energy-efficient link-layer jamming attacks have been
proved to be applicable in sensor networks [13,8]. If we sup-
pose that malicious nodes have the potential to launch link-
layer jamming attacks to indirectly cause packet loss between
neighboring normal nodes, then we have to consider more
about the threatened areas of suspect nodes. This is another
area for our future work.
5. Simulation study
In this section, we evaluate the performance of our scheme
through simulations that assume a more realistic scenario. In
this scenario, we consider not only packet loss due to malicious
dropping but also packet loss due to poor channel conditions.
Our detection scheme is always working during each simulation
run so that we can detect the attacks as soon as they happen.
This simulation scenario uses a field size of 2000 ×2000 m2
where 400 nodes are uniformly distributed. One stationary sink
and one stationary source sit on opposite sides of the field, with
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
8
randomly create the number of continuous hops. In the
discovery process, they obtain routes with a specific number
of hops in each direction via probability schemes. In data
transmission, a packet is sent via a source node using a
selected process after an event is produced. The source node
transfers the event packet to the subsequent hop node. The
next hop node, which is referred to as the intermediate node,
receives the event packet and neighboring node, which is
referred to as the monitor node. The monitor node, which is
the neighboring node responsible for detecting the possibility
of a selective forwarding attack, resends the event packet to
the destination node and sends an alarm message to its
neighboring nodes for the location attacker.
The advantages of a lightweight defense scheme are that a
share key is not required between nodes because the malicious
node is assessed via neighboring nodes. Because one node is
active in each RC, the energy will be efficient. The
disadvantages are that a GPS is required to achieve the
locations of nodes, which increases the cost of the networks.
In addition, a change in topology can impact the performance
of the defense scheme.
Tran Hoang and Eui-Nam [51] proposed an approach against
selective forwarding attacks that consists of a lightweight
detection mechanism. The detection is a centralized cluster,
which utilized the two-hop neighborhood node information
and overhearing technique. It is dependent on the broadcast
nature of sensor communication and the high density of
sensors. Each sensor node is provided with a detection module
that is constructed on an application layer. Sensor node sets
routing rules and two-hop neighbor knowledge to generate an
alert packet. Hoang and Nam suggested that the two routing
rules make the monitoring system more suitable. Thus, the
first rule is to determine if the destination node forwards the
packet along the path to the sink. It generates an alert packet
with the malicious factor α to the sender/source node. The
second rule governs that the monitor node waits and detects
the packet that was already forwarded along the path to the
sink. It verifies the two-hop neighbor knowledge to assess
whether the destination node is on the right path to the sink. If
not, it generates an alert packet with the malicious factor β to
the sender/source node.
Fig 10: Illustration of monitor node-Redrawn [51]
The detection module is responsible for passively detecting a
selective forwarding attack in its neighboring sensor node.
The malicious counter is defined as the threshold of abnormal
activity in a sensor node, which could not skip. When the
malicious counter crossed the threshold X, it revoked the
malicious node from its neighbor list. The authors have
assumed that the neighboring node should be recognized. The
neighboring node must be secure and confidential in the
deployment time. The network has a static topology and uses
key management to prevent any outside attacks. The selection
of one type of network topology prevents the scheme from
working with other topologies.
Young Ki Kim etal, [52] proposed the scheme CADE for
identifying malicious nodes. This detection scheme does not
require time synchronization to detect selective forwarding. It
transfers a cumulative acknowledgement to the base station.
Thus, authentication occurs between the base station and a
node. The detection scheme also provides security against
sinkhole attacks. CADE consists three phases: topology
construction (the authors suggested the use of SEEM topology
construction) and route selection, data transmission, and
detection process. The scheme will not work if the topology is
changed.
Huijuan Deng etal, [53] proposed a scheme for secure data
transmission and detecting a selective forwarding attack. They
used watermark technology to detect malicious nodes. Prior to
employing a watermark technique, they used a trust value to
determine a source path for message forwarding. The trust
value involves weighting the credit of each sensor node. The
author notes an error rate of 10% and detection accuracy
greater than 95%. They assumed that the base station is
always trustworthy and cannot be comprised by the adversary,
which renders the scheme inappropriate for real wireless
sensor networks. Every node has a trust value. At the
beginning of network initializing, all nodes should have the
same trust value. Huijuan Deng et al. utilized the watermark
technique to calculate the packet loss. Data transmission
begins when an optimal routing path is confirmed. The base
station creates a κ bits binary sequence as the original
watermark message. Therefore, a watermark message is part
of the packets. A base station compares the extract watermark
to the original watermark to detect a selective forwarding
attack. The simulation results reveal a channel error rate of
10% and detection accuracy greater than 95%.
Kaplantizs etal, [9] presented a scheme based on support
vector machines (SVMs) and sliding windows. The SVM is a
class machine-learning algorithm [54]. The scheme is a
centralized intrusion detection system. It detects two types of
attacks: a selective forwarding attack and a black hole attack.
The authors assumed that a sensor node does not spend more
energy while adding security features. They used anomaly
detection as the basis of IDS and indicated the detection of
these attacks with high accuracy without a drain of energy.
Chanatip et al. [55] have proposed a lightweight scheme.
They referred to it as a traffic monitor-based selective
forwarding attack detection scheme. They used Extra Monitor
(EM) to eavesdrop and monitor all traffic when transferring
data between nodes. They also employed RSSI to detect a
sinkhole attack. The value of RSSI is that four EM nodes can
be arranged to establish the positions of all sensor nodes, of
which the base station position should be (0,0). Chanatip et al.
have assumed that the network is static when sensor nodes are
deployed; thus, any change in the type of topology will
immediately affect their approach. They assumed that the
attackers could capture and damage the nodes. Therefore, all
sensor nodes must protect or use tamper robust hardware.
These assumptions have caused the detection scheme to drain
the energy of the sensor nodes and contribute to the high cost.
5. SECURITY BENCHMARKS
In the last few years, studies of wireless sensor networks have
increased. They can be employed in an extensive range of
applications, such as weather, military objectives tracking,
and patient monitoring. Therefore, these sensor networks need
protection from illegitimate users and attackers. Wireless
sensor network applications need protection against
adversaries who modify, inject, and eavesdrop packets. Some
techniques used to prevent DoS attacks include pushback,
payment for network resources, robust authentication and
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
9
traffic identification [23]. Wood and Stankovic suggested that
encryption and authentication techniques could warn network
administrators of attacks [29]. Perrig et al. [17] proposed
Security Protocols for Sensor Networks (SPINS), which
employ two symmetric keys based on security: SNEP and
µTESLA. SNEP provides confidentiality, data authentication
and data freshness, whereas µTESLA provides an
authentication broadcast using a one-way key chain.
Security benchmarks are used to prevent all types of DoS
attacks on wireless sensor networks. The following security
benchmarks should be included in sensor networks:
5.1 Authentication Protocols
Security is a controversial issue in WSN. Therefore, many
protocols serve WSN threats. The Sensor Network Encryption
Protocol (SNEP) provides confidentiality of the network.
SPINS are designed to provide confidentiality and
authentication by combining SNEP and µTESLA; thus, it is
an efficient broadcast authentication based on a one-way hash
chain [17].
5.2 Encryption
The majority of wireless sensor networks operate in an open
area or risky location, which make them susceptible to
network attacks. Eavesdropping or adding messages to a
network are significant to a WSN [56]. A WSN should adopt
key methods of protection, such as message authentication
codes, symmetric key encryption and public key cryptography
[14]. Law, Dulman, Etalle, and Havinga presented some
cryptographic algorithms in sensor networks, in which RC5
and TEA have a capability for the nodes [57].
5.3 Data Partitioning
The technique of partitioning involves the separation of the
data in networks into several parts. Deng J. [58] provides a
resolve to ensure the attacker cannot take part on data. Data
can make it as several packets to ensure that each packet
sends to a different route. Hence, the attacker attempts to get
all packets from the network, which requires access to the
entire network. It is a perfect solution but the energy drain is
exorbitant [6].
5.4 Secure Data Aggregation
The transmission of data in wireless sensor network has
significantly increased. As a consequence, data traffic is the
critical issue in WSNs. The cost of network traffic is very
high so to decrease it, wireless sensor nodes collect
measurements prior to transferring data to the base station
[14]. In wireless sensor network architecture, aggregation is
performed in many locations in the network [59].
5.5 Cryptography
Symmetric key cryptography is a key that is employed in
cryptography solutions in wireless sensor networks. It is
suitable and can be rapidly implemented [6]. Cryptography is
used to prevent some security attacks. In wireless sensor
networks, applications require protection against injection,
eavesdropping, and modification of data; therefore,
cryptography is a typical defense.
5.6 Shared Keys
The field of key management focuses on wireless sensor
networks. WSN is considered to be a single feature due to
size, mobility and power constraints [14]. Four types of keys
exist in key management: a global key, a pairwise key node, a
pairwise key group, and an individual key. These types of
keys give some solution to put the adversary a way from the
networks.
Table1: Selective Forwarding Attacks Detection Scheme Analysis
Technique/Features
Detection
Approach
Prevention
Approach
Efficiency
Reliabilit
y
Scalability
Energy
Consump
tion
Approach
Accuracy
Approach
Nature
Routing
Protocol
Acknowledg
ement
Based
Neighbor
Monitor
Karlof and Wagner
Χ
Moderate
High
N/A
Distributed
-
Χ
Χ
Yu and Xiao
Χ
Moderate
Χ
Χ
High
95%
Distributed
DD,
PSFQ
Χ
Yu and Xiao
(CHEMAS)
Χ
Moderate
Χ
Χ
High
95%
Distributed
DD,
PSFQ
Χ
Hung-Min Sun et al
Χ
Low
Χ
Χ
High
N/A
Centralized
-
Χ
Χ
Wang Xin-Sheng et
al
Χ
Low
High
N/A
Distributed
OPA_uv
wts
Χ
Tran Hai and Eui
Huh (Two-Hops)
Χ
Moderate
Χ
Χ
Moderate
90%
Distributed
-
Χ
Young Kim et al
(CADE)
Χ
Low
Χ
Χ
High
N/A
Centralized
SEEM
Χ
Huijuan Dong et al
(Watermark)
Χ
Moderate
Χ
Χ
Moderate
95%
Distributed
-
Χ
Χ
Support Vector
Machine (SVM)
Χ
Low
High
80%
Centralized
MTE
Χ
Chanatip and
Ruttikorn
Χ
Low
High
N/A
Centralized
-
Χ
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
10
6. CONCLUSION
Security issues in wireless sensor networks are critical. The
significant energy constraints and deployment of sensor nodes
in an unattended environment have contributed to the
vulnerability of wireless sensor networks. Components that
are designed without security can easily become an area for
attack. In recent years, the security of WSNs has become
increasingly concerning. The use of wireless sensor networks
is increasingly employed in environment, commercial, health
and military applications. The survey addressed the security
laps of WSNs on the network layer, particularly selective
forwarding. This paper contains a survey from 2003 to 2014.
Significant constraints and hazards of physical attacks are
extensively addressed at the network layer. The survey
includes a benchmark for the comparison of existing
approaches for handling the security of selective forwarding
attacks. Its challenges and future directions are
comprehensively highlighted in this paper. This survey will
help researchers to understand attacks on the network layer.
Due to a lack of security, many security attacks to prevent the
smooth functioning of wireless sensor networks; examples
include denial of sleep and homing. Table 1 summarizes the
approaches for the detection of various selective forwarding
attacks. Recommendations for future studies include an
investigation of the efficiency, scalability, and energy
consumption. Some detection approaches require
characteristics that may not be appropriate in real scenarios.
7. REFERENCES
[1] I. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirci,
“Wirelss sensor networks: A survey,” Computer Networks,
38(4):393-422, 2002.
[2] Karlof, C. and Wagner, D., “Secure routing in wireless sensor
networks: Attacks and countermeasures”, Elsevier’s Ad Hoc
Network Journal, Special Issue on Sensor Network Applications
and Protocols, September 2003, pp. 293-315.
[3] T. Zhu, Z. Zhong, T. He, and Z.-L. Zhang, "Energy-
synchronized computing for sustainable sensor networks," Ad
Hoc Networks, vol. 11, pp. 1392-1404, 2013.
[4] S. H. Lee, S. Lee, H. Song, and H. S. Lee, "Wireless sensor
network design for tactical military applications: remote large-
scale environments," in Military Communications Conference,
2009. MILCOM 2009. IEEE, 2009, pp. 1-7.
[5] A. Razaque and K. M. Elleithy, "Energy-Efficient Boarder Node
Medium Access Control Protocol for Wireless Sensor
Networks," Sensors, vol. 14, pp. 5074-5117, 2014.
[6] David Martins, and Herve Guyennet, “Wireless Sensor Network
Attacks and Security Mechanisms: A Short Survey”, 2010
IEEE.
[7] A. Perrig, J. Stankovic, and D. Wagner, “Security in Wireless
Sensor Networks”, Communications of the ACM, 47(6):53 57,
June 2004.
[8] Fengyun Li, Guiran Chang and Fuxiang Gao, Lan Yao,“A
Novel Cooperation Mechanism to Enforce Security in Wireless
Sensor Networks” 2011 Fifth Interantion Conference on
Genetic and Evolutionary Computing. IEEE computer society.
[9] Sophia Kaplantzis, Alistair Shilton, Nallasamy Mani, Y. Ahmet
ekercio glu,”Detecting Selective Forwarding Attacks in
Wireless Sensor Networks using Support Vector Machines”,
Intelligent Sensors, Sensor Networks and Information ,3rd
International Conference,pp.335 340, IEEE 2007.
[10] K. Loannis and T. Dimitriou, “ Toward Intrusion Detection in
Wireless Sensor Networks”, 13th Euopean Wireless Conference,
April 2007, pp. 1-7.
[11] Al-Sakib Khan Pathan, Hyung-Woo Lee, Choong Sean Hong,
“Security in Wireless Sensor Networks: Issues and Challenges”,
Proc. ICACT 2006, Volume 1, 20-22 Feb, 2006, pp. 1043-1048.
[12] Culler, D. E and Hong, W., “Wireless Sensor Networks”,
Communication of the ACM, Val. 47, No. 6, June 2004, pp. 30-
33.
[13] John Paul Walters, Zhengqiang Liang, Weisong Shi, Vipin
Chaudhary, “Wirelss Sensor Network Security: A survey.”
Chapter 17, Security in Distributed, Grid, and Pevasive
Computing Yang Xiao,(Eds.) pp. Auerbach Publications, CRC
Press 2006.
[14] Abhishek Jain, Kamal Kant, and M. R. Tripathy, “Security
Solutions for Wireless Sensor Networks”, to appear in IEEE
ICACCT 2012.
[15] Mayank Saraogi, “Security in Wireless Sensor Networks”,
University of Tennessee, Knoxville.
[16] J. Hill, R. Szewczyk, A. Woo, S. Hollar, D.E. Culler, K. Pister,
“System architectuer directions for networked sensors,” in
Proceedings of the 9th International Conference on Architictural
Support for Programming Languages and Operating System,
New York, ACM Press, 2000,pp. 93-104.
[17] A. Perrig, R. Szewczyk, Victorwen, D.E. Culler, J.D. Tygar,
“SPINS: Security protocols for sensor networks,” Wirelsss
Networks, Vol.8, No. 5, pp. 521-534, September 2002.
[18] S. Slijepcevic, M. Potkonjak, V. Tsiatsis, S. Zimbeck, M.B.
Srivastava, “On communication security in wirelss adhoc sensor
networks,” in Proceeding of 11th IEEE International Workshop
on Enabling Technologies: Infrastucture for Collaborative
Enterprisis (WETICE’02), 2002, pp. 139-144.
[19] L. Bysani and A. Turuk, “A Survey On Selective Forwarding
Attack in Wireless Sensor Networks”, IEEE 2011.
[20] S. H. Lee, S. Lee, H. Song, and H. S. Lee, "Wireless sensor
network design for tactical military applications: remote large-
scale environments," in Military Communications Conference,
2009. MILCOM 2009. IEEE, 2009, pp. 1-7.
[21] J.A. Stankovic et al, “Real-time communication and
coordination in embedded sensor networks,” Proceeding of the
IEEE, Vol. 91, No. 7, pp. 1002-1022, July 2003.
[22] A. Blilat, A. Bouayad, N. Chaoui, and M. Elghazi, “Wireless
Sensor Network: Security Challenges”, Computer
Communication. 2012, IEEE.
[23] David R. Raymond and Scott F. Midkiff,(2008) “Denial-of-
Service in Wireless Sensor Networks: Attacks and Defenses,”
IEEE Pervasive Computing, val. 7, no. 1, 2008, pp. 74-81.
[24] V. Thiruppathy Kesavan and S. Radhakrishnan, “Secret Key
Cryptography Based Security Approach for Wireless Sensor
Networks”, International Conference on Recent Advances in
Computing and Software Systems, 2012 IEEE.
[25] O. Arazi, H. Qi, D. Rose, “A public key Cryptographic Method
for DoS mitigation in WSN”, IEEE 2007.
[26] Asif Habib, “Sensor Network Security Issues at Network
Layer,” IEEE 2008.
[27] G. Padmavathi and D. Shanmugapriya,”A survey of Attacks,
Security Mechanisms and Challenges in Wireless Sensor
Networks” International Journal of Computer Science and
Information Security, IJCSIS 2009.
[28] Annie Jenniefer and John Raybin Jose,”Techniques for
Identifying Denial of Service Attack in Wireless Sensor
Network: a Survey” International Journal of Advanced Research
in Computer and Communication Engineering , IJARCCE 2014.
[29] A. D. Wood and J. A. Stankovic, “Denial of Service in sensor
networks,” IEEE Computer, 35(10): 54-62, 2002.
[30] Mohit Saxena,”Security in Wireless Sensor Networks A layer
based classifications”, Purdue University, West Lafayette, IN
47907.
[31] David R. Raymond and Randy C. Marchany,“Effects of Denial-
of-Sleep Attacks on Wireless Sensor Networks MAC
Protocols,” IEEE Transactions on Vehicular Technology, val.
58, no. 1, 2009.
[32] W. Ye, J. Heidemann, and D. Estrin, “Medium Access Control
with Coordinated Adaptive Sleeping for Wireless Sensor
Networks,” IEEE/ACM Trans. Netw., val. 12, no. 3, pp. 493-
506, Jun. 2004.
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
11
[33] T. VanDam and K. Langendoen, “An Addaptive Energy-
Efficient MAC Protocol for Wireless Sensor Networks,” in
Proc. 1st ACM Int. Conf. Embeded Netw. Sensor Syst., Nov.
2003, pp. 171-180.
[34] J. Polastre, J. Hill, and D. Culler, “Versatile Low Power Media
Access for Wireless Sensor Networks,” in Proc. 2nd ACM Int.
Conf. Embeded Netw. Sensor Syst., Nov. 2004, pp. 95-107.
[35] M. Brownfield, K. Mehrjoo, A. Fayez, and N. Davis, “Wireless
Sensor Networks Energy Adaptive MAC Protocol,” in Proc.
IEEE Consum. Commun. Netw. Conf., Jan. 2006, pp. 778-782.
[36] P. Mohanty, S. Panigrahi, N. Sarma, and S. Satpathy, “Security
Issues in Wireless Sensor Network Data Gathering Protocols: A
Survey”,journal of Theoritical and Applied Information
Technology.
[37] A. Razaque and K. M. Elleithy, "Energy-Efficient Boarder Node
Medium Access Control Protocol for Wireless Sensor
Networks," Sensors, vol. 14, pp. 5074-5117, 2014.
[38] C.-T. Hsueh, C.-Y. Wen, and Y.-C. Ouyang, "A secure scheme
for power exhausting attacks in wireless sensor networks," in
Ubiquitous and Future Networks (ICUFN), 2011 Third
International Conference on, 2011, pp. 258-263.
[39] Bo Yu and Bin Xiao, “Detecting Selective Forwarding Attacks
in Wireless Sensor Networks”, In Parallel and Distributed
Processing Symposiun, 2007. ISSNIP 2006, 20th International,
page 8 pp., 2006.
[40] Pandey, A., Tripathi, R.C., “A Survey on Wireless Sensor
Networks Security”, International Journal Computing
Application, IJCA, pp.43-49, 2010.
[41] Nagi, E. C. H., Liu, J. and Lyu, M. R., “Am Efficient Intruder
Detection Algorithm Against Sinkhole Attacks in Wireless
Sensor Networks”, Computer Communication. 6 May, 2007.
[42] M. M. Patel and A. Aggarwal,”Security Attacks in Wireless
Sensor Networks: A Survey” International Conference on
Intelligent Systems and Signal Processing(ISSP) IEEE 2013.
[43] J. R. Douceur,(2002) “The Sybil Attack,” in 1st International
Workshop on Peer-to-Peer Systems(IPTPS”02).
[44] Newsome, J., Shi, E., Song, D, and Perrig, A, “The Sybil attack
in sensor networks: analysis & defenses”, Proc. of the third
international
Symposium on!Information processing in sensor networks,
ACM, 2004, pp. 259 268.
[45] Yih-Chun Hu, Adrian Perrig, and David B. Johnson,”Packet
leashes: A defense against wormhole attacks in wireless sensor
networks”, IEEE infocom April 2003.
[46] N. Shanti, Lagnesan and K. Ramar, “Study of Different Attack
On Multicast Mobile Ad-Hoc Network”.
[47] Di Wu, Gang Hu, and Gang Ni,”Research and Improve on
Sensor Routing Protocols in Wireless Sensor Networks”, IEEE
2008.
[48] Bin Xiao, Bo Yu, and Chuanshan Gao, “CHEMAS: Identify
Suspect Nodes in Selective Forwarding Attacks”, In Parallel and
Distributed Processing Symposiun, 2007.
[49] Wang Xin-Sheng, Zhan Yong-Xhao, Xiong Shu-ming and
Wang Liang-min, “Lightweight Defense Scheme Against
Selective Forwarding Attacks in Wireless Sensor Networks”,
pages 226-232, IEEE 2009.
[50] Hung-Min Sun, Chien-Min Chen and Ying-Chu Hsiao, “An
Efficient Countermeasure to the Selective Forwarding Attack in
Wireless Sensor Networks. pages 1-4, IEEE 2007.
[51] Tran Hoang Hai and Eui-Nam Huh, “Detecting Selective
Forwarding Attacks in Wireless Sensor Networks Using Two-
hops Neighbor Knowledge” Seventh IEEE Internation
Symposium on Network Computing and Applications, 2008,
pp.325-331.
[52] Young Ki Kim, Hwaseong Lee, Kwantae Cho, and Dong Hoon
Lee, “CADE: Cumulative Acknowledgement based Detection of
Selective Forwarding Attacks in Wireless Sensor Networks”
Third Internation Conference on Convergence and Hybird
Information Technology, 2008, pp.416-422.
[53] Huijuan Deng, Xingming Sun, Baowei Wang, Yuanfu Cao,
“Selective Forwarding Attack Detection using Watermark in
Wireless Sensor Networks”, International Colloquium on
Computing, Communications Control, and Management (2009
ISECS), pp. 109-113.
[54] C. Cortes and V. Vapnik, “Support Vector Networks”, Machine
Learning, vol. 20, no. 3, pp. 273-297, 1995.
[55] Chanatip Tumrongwittayapak and Ruttikorn Varakulsiripunth,
“Detecting Sinkhole Attack and Selective Forwarding Attack in
Wireless Sensor Networks”, ICICS 2009.
[56] Kalpana Sharma. M K Ghose, “Wireless Sensor Networks: An
Overview on its Security Threats”, IJCA Special Issue on
Mobile Ad-hoc Networks 2010.
[57] Y. W. Law, S. Dulman, S. Etalle, P. Havinga,”Accessing
security-critical energy efficient sensor networks” University of
Twente, EA Enshede, Netherlands.
[58] Deng J, Han R, and Mishra S. “Countermeasures against
Traffic Analysis Attacks in Wireless Sensor Networks”, IEEE,
2005, pp. 113-126.
[59] Pathan, A. S. K., Hyung-Woo Lee, and Choong Seon Hong
“Security in Wireless Sensor Networks: Issues and Challenges”
Advanced Communication Technology (ICACT), 2006.
[60] Yan-Xiao Li, Lian-Qin and Qian-Liang, “Research on Wireless
Sensor Network Security”, In Proceedings of the International
Conference on Computing and Security, 2010 IEEE.
[61] Kalpana Sharma. M K Ghose, Deepak Kumar, Raja Peeyush
Kumar Singh, Vikas Kumar Pandey, “A comparative Study of
Various Security Approaches Used in Wireless Sensor
Networks”, In IJAST, Vol 7, April 2010.
[62] V. Kannan and S. Ahmed, “A Resource Perspective To Wireless
Sensor Network Security”, International Conference on
Innovative Mobile and Internet Service in Ubiquitous
Computing, IEEE 2011.
!
!
!
!
!
!
!
International Journal of Computer Applications (0975 8887)
Volume *– No.*, ___________ 2015
12
Mr. Naser Alajmi is pursuing towards
his Ph.D., Department of Computer
Science and Engineering at the
University of Bridgeport, Bridgeport,
CT. Naser’s interests are in Wireless Sensor Network (WSN),
Wireless Sensor Network Security, and Network Security.!
!
Dr. Elleithy is the Associate Vice
President of Graduate Studies and
Research at the University of
Bridgeport. He is a professor of
Computer Science and Engineering. He
has research interests are in the areas of wireless sensor
networks, mobile communications, network security, quantum
computing, and formal approaches for design and verification.
He has published more than three hundred research papers in
international journals and conferences in his areas of
expertise. Dr. Elleithy has more than 25 years of teaching
experience. His teaching evaluations are distinguished in all
the universities he joined. He supervised hundreds of senior
projects, MS theses and Ph.D. dissertations. He supervised
several Ph.D. students. He developed and introduced many
new undergraduate/graduate courses. He also developed new
teaching / research laboratories in his area of expertise.
Dr. Elleithy is the editor or co-editor for 12 books by
Springer. He is a member of technical program committees of
many international conferences as recognition of his research
qualifications. He served as a guest editor for several
International Journals. He was the chairman for the
International Conference on Industrial Electronics,
Technology & Automation, IETA 2001, 19-21 December
2001, Cairo Egypt. Also, he is the General Chair of the
2005-2013 International Joint Conferences on Computer,
Information, and Systems Sciences, and Engineering virtual
conferences. !
... For this, the key word is " localization " , so any malicious attack that can be addressed using localization-based techniques (i.e., position verification) can be a valid target for future research. Relevant examples in this context are the selective forwarding attack [68] or the Hello flood attack [69]. (b) Using directional antenna-based localization mechanisms to detect security attacks on other localization schemes. ...
Full-text available
Article
Being often deployed in remote or hostile environments, wireless sensor networks are vulnerable to various types of security attacks. A possible solution to reduce the security risks is to use directional antennas instead of omnidirectional ones or in conjunction with them. Due to their increased complexity, higher costs and larger sizes, directional antennas are not traditionally used in wireless sensor networks, but recent technology trends may support this method. This paper surveys existing state of the art approaches in the field, offering a broad perspective of the future use of directional antennas in mitigating security risks, together with new challenges and open research issues.
Chapter
Evolving technologies involve numerous IoT-enabled smart devices that are connected 24-7 to the internet. Existing surveys propose there are 6 billion devices on the internet and it will increase to 20 billion devices within a few years. Energy conservation, capacity, and computational speed plays an essential part in these smart devices, and they are vulnerable to a wide range of security attack challenges. Major concerns still lurk around the IoT ecosystem due to security threats. Major IoT security concerns are Denial of service(DoS), Sensitive Data Exposure, Unauthorized Device Access, etc. The main motivation of this chapter is to brief all the security issues existing in the internet of things (IoT) along with an analysis of the privacy issues. The chapter mainly focuses on the security loopholes arising from the information exchange technologies used in internet of things and discusses IoT security solutions based on machine learning techniques including supervised learning, unsupervised learning, and reinforcement learning.
Article
In present scenario, the fastest and cheapest way of communicating worldwide is via networks. Wireless Sensor Networks (WSNs) have become extensive part of many areas like traffic surveillance, defense, asset tracking, healthcares, structural monitoring of building and bridges and environmental monitoring of air, water and soil. The emergence of WSNs in today's technological world has lead to the need of secured and safeguarded transmission of data over networks. WSNs have issues like low memory and limited battery availability, so conventional security establishments are not effective here. A number of attacks are possible over WSNs like black hole, wormhole and selective forwarding attack. Selective forwarding attack is a special case of black hole attack where compromised nodes drop packets selectively. This leads to degradation of network performance. In this paper we will review some important attacks over WSNs with possible ways to detect and defend selective forwarding attack.
Full-text available
Article
Wireless Sensor Network (WSN) is an emerging technology that shows great promise for various futuristic applications both for mass public and military. The sensing technology combined with processing power and wireless communication makes it lucrative for being exploited in abundance in future. The inclusion of wireless communication technology also incurs various types of security threats. The intent of this paper is to investigate the security related issues and challenges in wireless sensor networks. We identify the security threats, review proposed security mechanisms for wireless sensor networks. We also discuss the holistic view of security for ensuring layered and robust security in wireless sensor networks.
Chapter
Wireless sensor networks (WSNs) are networks of tiny sensing devices spread over a large geographic area, and can be used to collect and process environmental data like temperature, humidity, light conditions, seismic activities, and images of the environment. This data can be used to detect certain events and to trigger activities.
Article
Thesupport-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data.High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
Article
Wireless sensor networks are usually deployed for gathering data from unattended or hostile environment. Several application specific sensor network data gathering protocols have been proposed in research literatures. However, most of the proposed algorithms have given little attention to the related security issues. In this paper we have explored general security threats in wireless sensor network and made an extensive study to categorize available data gathering protocols and analyze possible security threats on them.
Article
Unless their developers take security into account at design time, sensor networks and the protocols they depend on will remain vulnerable to denial of service attacks. We identify denial of service weaknesses and solutions for sensor network devices and analyze two network protocols.
Article
The support-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data. High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
Conference Paper
Security is important for many sensor network applications. A particularly harmful attack against sensor and ad hoc networks is known as the Sybil attack [6], where a node illegitimately claims multiple identities. This paper systematically analyzes the threat posed by the Sybil attack to wireless sensor networks. We demonstrate that the attack can be exceedingly detrimental to many important functions of the sensor network such as routing, resource allocation, misbehavior detection, etc. We establish a classification of different types of the Sybil attack, which enables us to better understand the threats posed by each type, and better design countermeasures against each type. We then propose several novel techniques to defend against the Sybil attack, and analyze their effectiveness quantitatively.
Conference Paper
Security in Wireless Sensor Networks (WSNs) is especially challenging and quite different from traditional network security mechanisms. There are two major reasons. Firstly, there are severe constraints on these devices namely their minimal energy, computational and communicational capabilities. Secondly, there is an additional risk of physical attacks such as node capture and tampering. Moreover, cryptography based techniques alone are insufficient to secure WSNs [1]. Hence, intrusion detection techniques must be designed to detect the attacks. Further, these techniques should be lightweight because of resource-constrained nature of WSNs [2]. In this paper, we present a new approach of robust and lightweight solution for detecting the Sinkhole attack and the Selective Forwarding attack based on Received Signal Strength Indicator (RSSI) readings of messages. The proposed solution needs collaboration of some Extra Monitor (EM) node apart from the ordinary nodes. We use RSSI value from four EM nodes to determine the position of all sensor nodes which the Base Station (BS) is origin position (0,0). Later, we use this information as weight from the BS. Another functions of EM nodes are eavesdropper and monitor all traffics, in order to detect the Selective Forwarding attack in the network. Our solution is lightweight in the sense that monitor nodes were not loaded any ordinary nodes or BS and not cause a communication overhead.
Conference Paper
Wireless Sensor Network (WSN) is being emerged as a prevailing technology in future due to its wide range of applications in military and civilian domains. These networks are easily prone to security attacks, since once deployed these networks are unattended and unprotected. Some of the inherent features like limited battery and low memory make sensor networks infeasible to use conventional security solutions. There are lot of attacks on these networks which can be classified as routing attacks and data traffic attacks. Some of the data attacks in sensor nodes are wormhole, jamming, selective forwarding, sinkhole and Sybil attack. In this paper, we discussed about all these attacks and some of the mitigation schemes to defend these attacks.