ArticlePDF Available

Abstract and Figures

In this article, we present a model of cyber attacks which can be used to represent a cyber attack in an intuitive and concise way. With ever-increasing popularities of online services, we have seen a growing number of cyber attacks targeted towards large online service providers as well as individuals and the IoT devices. To mitigate these attacks, there is a strong urge to understand their different aspects. Creating a model is a widely used method towards this goal. Unfortunately, the number of models for cyber attacks is pretty low and even the existing models are not comprehensive. In this paper, we aim to fill this gap by presenting a comprehensive cyber attack model. We have used this model to represent a wide range of cyber attacks and shown its applicability and usefulness. We believe that our model will be a useful tool for the formal analysis of cyber attacks.
Content may be subject to copyright.
MODELLING CYBER ATTACKS
Farida Chowdhury1 and Md Sadek Ferdous2
1Department of Computer Science & Engineering, Shahjalal University of Science and
Technology, Sylhet, Bangladesh
farida-cse@sust.edu
2Electronics and Computer Science, University of Southampton, Southampton, UK
S.Ferdous@soton.ac.uk
ABSTRACT
In this article, we present a model of cyber attacks which can be used to represent a cyber attack in an
intuitive and concise way. With ever-increasing popularities of online services, we have seen a growing
number of cyber attacks targeted towards large online service providers as well as individuals and the
IoT devices. To mitigate these attacks, there is a strong urge to understand their different aspects.
Creating a model is a widely used method towards this goal. Unfortunately, the number of models for
cyber attacks is pretty low and even the existing models are not comprehensive. In this paper, we aim to
fill this gap by presenting a comprehensive cyber attack model. We have used this model to represent a
wide range of cyber attacks and shown its applicability and usefulness. We believe that our model will be
a useful tool for the formal analysis of cyber attacks.
KEYWORDS
Cyber Attacks, Modelling, Security.
1. INTRODUCTION
With the ever-increasing demand for online services, more and more financial transactions,
consisting of highly sensitive financial and personal information, are carried out online.
Furthermore, we are heading towards a direction where a plethora of IoT (Internet-of-Things)
devices will get connected to the Internet. These IoT devices comprise of a wide variety of
hardware ranging from small devices envisioned to be used within a smart home environment to
large Cyber-Physical devices to be used for protecting national infrastructures. All these
scenarios present a lucrative economic incentive for the attackers to carry out cyber attacks.
This is evident from large scale attacks against a number of large online service providers such
as Amazon, Ebay, Yahoo, Sony and so on [1, 2, 3, 4, 5]. In a recent cyber attack, a huge number
of IoT devices have been compromised to launch a large scale Distributed Denial of Service
attack [4].
To mitigate these attacks, there is a strong urge to understand their different aspects. Towards
this goal, a popular approach among the researchers is to categorise and analyse cyber attacks
using a taxonomy. This has resulted in a number of taxonomies created using different criteria
[6, 7, 8, 9]. Even though very popular, a taxonomy can barely be used to represent a cyber
attack in a concise way and to understand its different characteristics.
Creating a comprehensive model would be a much more effective approach in order to
understand and represent a cyber attack. Unfortunately, the number of models for cyber attacks
is pretty low and even the existing models fail to embody all the required properties of a cyber
attack. In this paper, we aim to feel this gap by presenting a comprehensive model of a cyber
attack that captures a wide range of properties associated with a cyber attack.
Contribution. The major contributions of this article are:
International Journal of Network Security & Its Applications (IJNSA) Vol.9, No.4, July 2017
DOI: 10.5121/ijnsa.2017.9402
A comprehensive model of a cyber attack encoding its different characteristics.
A taxonomy of cyber attacks based on the motivations of an attacker.
Modelling a variety of major cyber attacks according to the presented model and the
taxonomy.
Showcasing the applicability and the usefulness of the presented model by illustrating
two examples.
Structure. The paper is structured as follows. The related work is presented in Section 2. In
Section 3, we present the model of a cyber attack. Next, a taxonomy of different cyber attacks is
presented in Section 4. Then, we model different cyber attacks utilising our model in Section 5,
Section 6 and Section 7. We discuss the usefulness of our model by presenting two examples in
Section 8 and finally, we conclude in Section 9 with a direction of future work as well.
2. RELATED WORK
Understanding cyber attacks from different perspectives is crucial in order to mitigate them.
Towards this vein, a popular research topic is to create, present and analyse different cyber
attacks using a taxonomy based on a range of criteria. There are a number of research papers
which either present a novel taxonomy based on new criteria or survey and analyse the existing
taxonomies to identify their strengths and weaknesses as well as to identify the gaps in them.
These works provide a solid foundation towards understanding different cyber attacks which is
essential to model them. That is why, at first, we analyse a few works on the taxonomy of cyber
attacks in different domains.
In [6], the authors have presented a dimension based taxonomy. The authors have utilised four
different dimensions where the first dimension consists of different attack vectors. In the second
dimension, the authors have considered the attack targets to create a taxonomy whereas the third
dimension considers different vulnerabilities. Finally, in the fourth dimension, attack payloads
are considered. Moreover, the authors have also analysed and compared a few other similar
taxonomies.
Supervisory Control and Data Acquisition (SCADA) systems are an integrated component of
any crucial critical infrastructure as well as cyber-physical systems. In recent years, SCADA
systems have been increasingly targeted for different cyber attacks. In [7], authors have
presented a taxonomy of cyber attacks in SCADA systems. Their main motivation in presenting
the taxonomy is to highlight common traits among the attacks and to identify unique challenges
for securing such systems. In a similar vein, the authors in [8] have presented another taxonomy
of different cyber attacks based on a different set of criteria. Similarly, in [9], the authors have
presented a taxonomy of network attacks as well as a taxonomy of attack tools. In addition, a
comprehensive survey of different attack and defence tools and systems have been presented.
There are a few other works presenting different taxonomies of attacks. They are briefly
discussed below:
An analysis and comparison of different taxonomies within the domain of social
engineering has been presented in [10].
The authors in [11] have presented a taxonomy of a wide range of attack methods in
Peer-to-Peer networks.
A taxonomy of threats in Cloud-of-Things has been presented in [12].
Next, we explore a few works which focus on modelling (mathematically or analytically)
different threats, security attacks and other related issues. A formal way for modelling
information security attacks has been presented in [13]. The model utilises attack trees to
represent a security attack. This attack tree is then used to identify attack patterns. The authors
argue that modelling based on attack trees and attack patterns will be helpful to identify
common yet recurring attack traits which could be utilised to mitigate such attacks.
In [14], a formal language, called Correlated Attack Modeling Language (CAML), has been
presented to model a multistep cyber attack scenario. CAML supports a modular functionality
where each module can represent an attack inference and therefore, multiple modules can be
linked together to represent multistep attack scenarios. In addition, CAML is equipped with a
library of predicates which are used to describe different system properties, states and events.
Their motivation is similar to the motivation of this paper. However, unlike their work, our
focus is on the representation of the attack itself.
Petri nets have been utilised to model cyber-physical attacks within the domain of smart-grid in
[15]. The authors argue that petri nets are more expressive in comparison to attack trees to
represent any attack. However, modelling a large scale cyber-physical system requires a
significant manual input to create the petri net. To address this limitation, the authors have
proposed a hierarchical method to create large petri net by combining a number of small petri
nets. A petri net can provide an excellent visual representation of attacks, however, their main
shortcoming is that they do not encode different significant properties of an attack.
There are other works, as presented in [16, 17, 18, 19], which discuss and present a threat model
in lifelogging, mathematical representation of identity and trust issues. Even though they are not
strictly related to the scope of current paper, we have drawn motivations on how to model an
attack from these works.
In essence, there is not any work presenting a comprehensive model of a cyber attack. To the
best of our knowledge, this article presents the first attempt to model a cyber attack
comprehensively by encoding its different properties.
3. ATTACK MODELLING
Let us assume that denotes the set of attackers while denotes the set of victims. The set of
cyber attacks is denoted using .
A victim can be a single person or an organisation. We denote the set of persons and
organisations as and respectively and define as follows.

Furthermore, we denote the set of systems as . We assume that every system is either owned or
operated by a person or an organisation. Every system has many processes which are different
computing programs running in the system [20]. We denote the set of processes as . Every
system is connected to another system using the network, consisting of different routers,
bridges, switches, hubs and so on. Without specifying too much granularities regarding these
components of a network, we consider it as a single entity. We use the notation to denote the
set of networks. Finally, every system, in reality different programs in the system, handles data
which are also transmitted between different systems using a network. We denote the set of data
with .
According to our model, every attack is originated from an attacker (where ) and is
aimed towards a single target or a set of targets. The target can be data, a process, a system or a
network. We denote the set of targets with which is defined in the following way:

Via the target, each attack attempts to victimise a person or an organisation, generally labelled
as the . We model this relationship using the following notation:

where, and  The notation represents the direction of the attack whereas
the notation is used to point out the respective victim.
Every cyber attack is launched using a channel. We consider three types of channels: visual,
network and hybrid channel. A visual channel allows an attacker to visually inspect a victim or a
system, collect sensitive information and then launch an attack by physically accessing a
system. A network channel, on the other hand, allows an attacker to inspect a victim or a system
remotely in order to collect sensitive information and then launch an attack over a network.
Finally, a hybrid channel allows an attacker to visually inspect a victim or a system in order to
collect sensitive information and then launch an attack targeting the system or the victim
remotely over a communication network. We use the notation  to denote the visual channel,
 to denote the network channel and  to denote the hybrid channel. Combining these
three channels, we define the set of channels (denoted using ) in the following way:

Every computing system that is connected to the Internet leverages a conceptual model of layers
in order to communicate with another system. In our model, we utilise the TCP/IP Protocol
model, also known as the DARPA model [21]. According to the TCP/IP protocol model, there
are four layers: Network Interface layer (denoted as ), Internet layer (denoted as ),
Transport layer (denoted as ) and Application Layer (denoted as ). Each cyber attack
usually targets a specific layer to launch a successful attack. However, there might be some
attacks which might target multiple layers. Combining these four layers, we define the set of
layers (denoted using ) in the following way:

Each attack has an associated probability of being successful. We denote the probability of an
attack with . In essence, the probability of an attack determines its severity. If an attack
has a higher probability of being successful, it is considered to be more severe than another
attack having a lower probability. In our model, we consider three different types of severity:
and . The severity of an attack is defined as a function denoted as 
and is defined as: 
To concretise the severity of an attack, we utilise two different thresholds of probability, and
, where denotes a high probability threshold and denotes a relatively low probability
threshold in the threshold spectrum. Therefore, the severity of an attack is concretised in the
following way for any :
 


Since, the probability thresholds effectively can be enumerated using different concrete values
(e.g.  and so on) in different situations, we have restricted ourselves from assigning a
concrete numerical value for the thresholds.
We differentiate between two types of attacks: active and passive [8]. An active attack (denoted
as ) enables an attacker to modify, misconfigure or disrupt a target (e.g. modifying a
process, system or data; disrupting a communication channel and so on) whereas a passive
attack (denoted as ) allows an attacker to observe a target without modifying it. With
these two types, we define the types (denoted as ) of an attack as follows:

Finally, we define an attack () as the following tuple consisting of the attack
relation, the channel, the layer it utilises and its associated severity along with the type:

where,  and .
4. ATTACK TAXONOMY BASED ON MOTIVATIONS
An attacker initiates an attack with a concrete motivation. Here, we have identified three
different types of motivations: footprinting, launching and trace removal. Unlike any previous
taxonomies, the taxonomy presented here is based on these motivations. Next, we describe each
of these motivations.
Footprinting. Footprinting, also known as information gathering [9], can be defined as the
systematic use of tools and techniques by an attacker to create a complete security profile of a
victim. The security profile consists of information with respect to the target, such as process,
layer, network and the system, of the victim. The collected information is used to identify any
vulnerability within the target of the victim.
Launching/Compromisation/Gaining access. After gathering enough information about the
target and identifying vulnerabilities within them, the attacker aims to compromise the target by
launching more severe attacks which exploit the identified vulnerabilities. The main motivation
is to get hold of credentials or to compromise the target victim for gaining access to the system.
Once the attacker gains access to the system, he/she can exploit the system anyway possible.
The attacker can steal sensitive information, install malicious software such as a key-logger or
rootkit or even abuse the system to attack another system belonging to another victim.
Trace removal. After abusing the access to carry out malicious activities, the attacker attempts
to modify the system in order to eradicate any trace (history) of the attacker compromising and
accessing the system. This step is carried out to ensure that any sign of the system being
compromised remains unnoticed.
Next, we present a taxonomy of attacks by classifying them based on these motivations (Figure
1). In the next section, we present the attacks belonging to a particular category and model each
corresponding attack using the model presented in Section 3.
Figure 1: Taxonomy of attacks based on motivations.
5. FOOTPRINTING
There are a wide range of footprinting attacks which can be categorised according to their
underlying mechanism. Based on these mechanisms, a taxonomy of footprinting attacks is
presented in Figure 2. Each of these attacks is presented below according to their mechanism
and modelled using our mathematical model. It is to be noted that the attacks presented here are
indirect in nature, meaning that none of these can be used to launch an attack with any
devastating implication.
5.1. Social Engineering
Humans are often regarded as the weakest link in any information security system [22]. The
social engineering is the process of exploiting the weakest link, the people, in the system with
illegitimate motivations. It can be defined as an attacker’s use of multitude methods, such as
personal interviewing techniques, research skills and trickery/deception, to communicate,
deceive and consequently, extract sensitive information regarding a victim from the victim or
from the people who are close to the victim such as the victim's employees, partners or
customers [23]. It is considered as one of the most primitives yet one of the most successful
ways to gather unauthorised information which can be leveraged at a later stage of an attack [24,
25, 26]. There are several mediums by which an attacker initiate a social engineering attack
such as via telephone, via an email message, a television commercial, a web-based mechanism
or countless other mediums which might provoke human reactions.
Figure 2: Taxonomy of footprinting attacks.
There are several attacks that can be categorised under the umbrella of social engineering. Some
of these attacks can be leveraged for the footprinting process whereas the others can be directly
exploited to carry out more direct attacks. Here, we only describe those attacks which are used
for footprinting. The attacks are presented and modelled below [27, 10].
Phishing. Phishing is a kind of social engineering attack which is defined as follows [27].
Phishing is the attempt to acquire sensitive information or to make somebody act in a desired
way by masquerading as a trustworthy entity in an electronic communication medium. They are
usually targeted at large groups of people. Phishing attacks can be performed over almost any
channel, from physical presence of the attacker to websites, social networks or even cloud
services.
There are different modes of phishing attacks: Deceptive Phishing, Spear Phishing, Whaling,
Pharming, Dropbox Phishing, Google docs phishing and so on [27].
Phishing is denoted as  and modelled in the following way:

The explanation for modelling this attack in the above manner is as follows:
A phishing attack targets critical user information (data) which can be used to
compromise a system. That is why the target is and .
Such an attack is carried out via the Internet and hence, the modelled channel is .
Footprinting
Social engineering
Phishing
Dumpster diving
Vishing
Reverse social
engineering
Scanning
Ping sweep
TCP scan
UDP scan
OS Identification
Sniffing
Similarly, the attack occurs through an application interface (e.g. via email, website,
etc.) and therefore, the layer is modelled with .
The severity of the attack depends on what information is captured. For example, if only
innocuous information such as a username or date of birth is captured, it can hardly be
abused to compromise a system and hence, it has a  severity. On the other hand, if
the attacker can capture both username and password or any credit card information, it
might have severe consequences and that is why it can be modelled as .
Finally, this is a passive attack as this does not actively compromise a system, rather
information captured via this attack is used to launch an active attack. Hence, it is
modelled as .
Dumpster diving. The dumpster diving is the process of scavenging through the dumpster with
the hope to find sensitive information [27, 28]. Without considering the implication, documents
are sometimes discarded in the dumpster which might contain several pieces of crucial
information regarding employees, memos and even printed/hand-written sensitive information,
such as a password. The main motivation of the attacker is to find such crucial information to
carry out more direct attacks, e.g. compromising a system or a specific user account, in the
subsequent step. We denote this attack with  and model in the following way:

The explanation is similar to what presented before and hence, it has been skipped for brevity.
Vishing. Vishing, also known as Voice Phishing, is the method of utilising a rogue Interactive
Voice Response (IVR) system to simulate a legitimate IVR system of an important institution
(e.g. a bank) in a convincing manner [10, 29]. The ultimate motivation is to trick the victim to
release sensitive information. The vishing attack is denoted with  and modelled in the
following way:

Reverse social engineering. In the traditional mode of any social engineering attack, an
attacker initiates the interaction by the process described above. However, in the reverse social
engineering attack, the attacker presents a problematic scenario to the victim as well as
impersonates someone that the victim trusts in order to address the problem [27, 30]. This
would allow the attacker to gain the trust of the victim which could be abused to collect
sensitive information to be leveraged at the subsequent step of the attack.
We denoted this attack with  and model as follows:

5.2. Scanning
All the footprinting methods discussed above require the attacker to interact with the victim to
collect sensitive information. However, information with respect to the network and the system
of the victim, such as employee names and phone numbers, IP address ranges, DNS servers, and
mail servers, also represents a valuable source of information. Interestingly, such information
can be collected via remote methods without any direct interaction with the victim. Scanning is
the process to facilitate the collection of such information. There are a variety of scanning tools
and techniques available, some of which are listed next.
Ping sweep: The basic and the most primitive scanning technique is Ping sweep. It is
based on sending an automated ping request on a range of IP addresses and network
blocks to determine if the target systems are alive [31, 32, 33].
TCP scan: TCP scan is the process of probing and determining open TCP ports which
are associated with different network services that the attacker can exploit [34]. There
are different ways it can be carried out. For example, the scanning process can probe for
normal TCP connections (TCP Connect Scan) or employ advanced stealth scans that
probe for half-open connections to prevent them from being logged (TCP SYN Scan or
TCP FIN scans).
UDP Scan: In this process, a UDP packet is sent to the target port of the target machine
[34]. Most machines will respond with an ICMP “destination port unreachable”
message, indicating that no service is active on that port. However, if no message is
received, an attacker can deduce the port is open and a service is utilising the port. It
should be noted that the UDP scanning process is not as reliable as the TCP scan as
UDP is a connectionless protocol. Therefore, the accuracy of this method depends on
which system and network resources have been utilised.
OS Identification: Once an attacker identifies the ports and the corresponding services
running in the target machine using the scanning methods described above, the attacker,
then, tries to determine the type of Operating System (OS) within the target system.
Different OS have different responses with respect to different queries. The attacker
will match those queries with a predetermined QUERY-REPLY profile to determine the
target OS.
We denote all the scanning attacks with  and model as follows:

5.3. Sniffing
Sniffing is a method by which an attacker can compromise the security of a network in a passive
fashion [42]. To initiate this attack, an attacker captures and analyses all network packets to
retrieve some useful information. For this, an attacker utilises a tool called Sniffer which can
capture packets in a network and analyse them to identify sensitive information, such as
authentication information consisting of usernames and passwords. We denote this attack using
 and model it in the following way:

6. LAUNCHING
At this step, it is assumed that the attacker has been successful to gather a wide range of crucial
information utilising different footprinting methods. Next, the attacker aims to launching
different  attacks abusing the gathered information. There are different launching
attacks available at the disposal of the attacker. The taxonomy of these launching attacks is
presented in Figure 3. Next, each of these attacks is briefly discussed and modelled according to
our model.
6.1. Account scan
The simplest form of attacks for an attacker to launch is account scanning. For this, the attacker
attempts to break in to all identified active network services (e.g. web service, FTP service, etc.)
by checking:
an account with no password,
an account with the password same as the username, or “password”,
a default account that is shipped with the product/software,
an anonymous FTP account,
rlogin/rsh/rexec ports that may support less trusted logins.
We denote this attack with  and model it as follows:

Figure 3: Taxonomy of launching attacks.
6.2. Social Engineering
Here, we present the social engineering attacks which can be utilised for launching direct
attacks. One example of a successful and common attack method is to simply call the helpdesk
of an organisation and say something like this: “Hi, this is Mr. X, the senior director of the
organisation. I have to present something to the CEO, but I cannot log into server XYZ to get
my notes. Would you please reset my password now according to my choice? I have to be in this
meeting in 2 minutes.” Nowadays, most corporations should have a policy for their helpdesk
operators not to reset password as requested. However, an unsuspecting and ill-trained operator
might simply reset the password as requested. Once, the password is reset, the attacker can
easily log in to the account and do whatever he likes. Next, we present some of the most widely-
used such social engineering methods.
Waterholing. It refers to a targeted attack in which a website, supposedly to be of interest to the
victim, is compromised and then malicious contents are injected by the attacker [27, 35]. Once
the victim visits the website, the malicious contents are loaded and executed, thus enabling the
user to compromise and then ultimately take control of an online account or even the system of
the attacker. We model this attack in the following way where  represents the waterholing
attack: 
Baiting. Baiting is the process of luring the victim by drawing his attention using an object [27,
36]. The object could be a malware/trojan horse infected storage medium (e.g. a USB drive)
which is intentionally left in a place that could be easily found by the victim. To increase the
curiosity of the victim, the attacker often labels the object with tempting labels such as
confidential. Out of curiosity, once the victim picks up the object and inserts into his system, the
system gets infected with the malware/trojan allowing the attacker to have full access to the
system. The attack method (denoted as ) is modelled in the following way:

6.3. Advanced Persistent Threat
Advanced Persistent Threat (APT) refers to a long-term attack method employed by a group of
attackers targeting specific organisations, governmental institutions, commercial enterprises in
order to infiltrate the system for monetary or espionage purposes [27, 37, 38]. This is unlike
other traditional attack methods where an attack is launched on a one-time basis. In APT, the
attackers study, monitor and attack the target systems for a long period of time with a stay-low
and slow approach so that the attacks remain unnoticed. Also, the attackers in APT are usually
very well resourced and well organised. US National Institute of Standards and Technology
(NIST) has formulated a comprehensive definition of APT combining its different
characteristics [39]. The characteristics from this definition is presented below:
The advanced persistent threat: (i) pursues its objectives repeatedly over an
extended period of time; (ii) adapts to defenders' efforts to resist it; and (iii) is
determined to maintain the level of interaction needed to execute its objectives.”
We model APT, denoted as  in the following way:

6.4. Password Cracking
If the attacker fails to compromise the target system using the account scanning or any of the
social engineering methods, he might adopt a more direct exploitation approach to acquire the
credentials (passwords) associated with the target systems. One of the most effective methods
with this regard is the process of password cracking which can be carried out in one of the
following ways.
Guessing attack. The simplest approach for password cracking is to guess the password. Many
unaware users do not comprehend the necessity to maintain a secure difficult-to-guess password
and hence, they often use passwords which can be easily guessed. A few examples of such
easily guessable passwords are [40]:
To use the word "password" as the password.
The same password as the username.
The real names as the passwords.
The name of the children, spouse, pet, or car model as the passwords.
Birthdays and birth places as passwords.
Favourite colours, foods and places as passwords.
Understandably, this approach is more effective if the attacker personally knows the victim and
has the knowledge of such information which is susceptible to password guessing.
Dictionary attack. In this attack, the attacker utilises a program or a script that tries different
possible combinations of words in a dictionary along with some additional special characters
(such as '\#', '\$', '\_' and so on) often used in the beginning or in the end of a password [40]. The
attacker usually possesses a copy of the English dictionary as well as foreign language
dictionaries for this purpose. In addition, additional dictionary-like databases containing names
and lists of common passwords are often used.
Brute-force attack. The brute-force attack, often resorted as the last step for the password
cracking, requires an attacker to try all possible combinations of characters within the length of
the password. A short 4-letter password consisting of lower-case letters can be cracked in just a
few minutes. However, a 7-character long password consisting of either upper or lower case
letters would take 267 (8,031,810,176) guesses. A combination of alpha-numerical characters
along with case-sensitivity and special characters would increase the complexity significantly
that it might be impossible to crack within a reasonable period of time.
All these attacks are represented using a combined model in the following way where
 denotes the generic password cracking attack:

6.5. Observation/Shoulder surfing
Another primitive yet successful method to accumulate the password is by observation. One of
the traditional problems in password-based security is that passwords must be long and difficult
to guess. However, such passwords are often difficult to remember. Therefore, users often write
them down somewhere. An attacker can often search a person's work site/desk in order to find
passwords written on little pieces of paper usually hidden under the keyboard. An attacker can
also train themselves to launch the shoulder surfing attack where they look over the victim's
shoulder while the victim types in his password at the screen or keyboard [27]. This attack is
denoted as  and modelled as follows:

6.6. Port re-direction
Port re-direction can be defined as the process to direct network traffic destined for one port and
redirect it to another host into another port [41, 8]. To gain access to a system, the attacker has
to compromise the system. However, there might arise a situation where an intervening entity
such as a firewall blocks any direct access to a target system. Resourceful attackers can find
their way around these obstacles using the port redirection attack to gain access to any system
behind a firewall. For this, the attacker listens to certain ports of a trusted host and forwards
certain raw packets to a specified secondary target which might be behind a network firewall.
We model this attack in the following way:

6.7. Spoofing
A spoofing attack allows an attacker to falsify the identity of a trusted and authorised user so
that the attacker can gain access to the system or the services [43, 44, 45]. It has different
modes. In an IP spoofing attack, the attacker spoofs the source address of an IP packet in order
to simulate that the packet has originated from a trusted computer and thereby, compromising
any IP based authentication mechanism. In a DNS (Domain Name Service) spoofing attack, the
attacker compromises a DNS server and modifies the DNS entry in such a way that a domain
name is rerouted to an IP address belonging to a server controlled by the attacker. In an ARP
(Address Resolution Protocol) spoofing attack, the attacker spoofs the association between
MAC (Media Access Control) and IP address in a LAN in such a way that all traffic targeted for
a trusted node are routed to the attacker node. These spoofing attacks are generally used to steal
or alter sensitive information regarding the victim. However, these attacks are used to facilitate
other attacks (e.g. Man-in-the-Middle and Denial-of-Service attacks) which will be described
below. We denote Spoofing attack using  and model it in the following way:

6.8. Session hijacking
Session hijacking is the process of taking over a connection which has either been established or
is in the process of being set up [46]. In most cases, this would be connections for web
applications, however, connections could belong to other protocols such as FTP, SMTP and so
on. There are many advanced techniques to launch this attack. We present the most well-used
session hijacking techniques below.
HTTP Session. The HTTP Session hijacking is targeted towards a web application [47, 48, 49].
Once a user is successfully logged in a web application, a session is created with a unique
session identifier. Then, a session token is issued to the user containing the session identifier,
indicating that the user logged in. This identifier is passed in different ways, such as via HTTP
cookies, URL rewriting or hidden fields and is used in all subsequent interactions between the
user and the web application. Once the session is established, the attacker can hijack the HTTP
session by simply getting hold of this identifier which can be used to impersonate the user. This
attack is represented using  and modelled in the following way:

TCP Session. A TCP session hijacking technique exploits one of the key features of the TCP/IP
protocol [50]. Using different approaches, an attacker can insert malicious TCP packets into an
already-established TCP session, thereby enabling commands to be executed on the remote
host. The most effective time to hijack the session is after a session has been established
between a server and a client and these entities trust each other. There are different ways to
launch this attack. For example, an attacker can spoof IP packets (as described above) to insert
packets containing malicious commands into the session. Another variant of a TCP session
attack leverages a special form of attack called Man-in-the-Middle (MITM) [51]. In an MITM
attack, the attacker places himself, using the ARP spoof attack, within an established connection
between two entities in such a way that every packets transmitted between these entities go via
the attacker, thus enabling to insert falsified packets to hijack a session. We model the TCP
session attack in the following way where  denotes the attack:

6.9. Remote table modification
In a remote table modification attack, similar to an MITM attack, an attacker will try to modify
the routing table of the target host in such a way that all packets flow through a system he
controls. Preferably, an attacker will try to maliciously modify the routing tables remotely by
targeting the Open Shortest Path First (OSPF) or the Border Gateway Protocol (BGP) which are
used by most ISPs for exchanging route information with each other [52, 53]. A local version of
this attack might try to spoof ICMP (Internet Control Message Protocol) packets so that the
target host is tricked to route packets via the attacker's host. This works as many OSs have a
default configuration which accepts ICMP redirects [51]. We model this attack in the following
way where  denotes the attack:

6.10. Unexpected input/Buffer overflow
To interact with most of network and web-based applications, users are expected to provide
different types of inputs in the forms of mouse click, keyboard typing or multi-modal touch.
Handling these user inputs without proper care can introduce vulnerabilities which an attacker
can exploit to launch different types of attacks. Many of these vulnerabilities occur due to a
mistake in coding, lack of experience in writing secure code as well as undocumented anomaly.
One of the most notorious attacks exploiting the unexpected input is the Buffer Overflow attack
which is widely used by the attackers. In a buffer overflow attack, an attacker crashes or gains
control of a specific program of a target host by overflowing the buffer of the program [6, 54,
55]. A common practice among the programmers is to allocate an arbitrary number of bytes for
a buffer within a program which is often utilised to store user inputs. If the size of user input is
larger than the allocated bytes for the buffer, a situation of buffer overflow occurs. When this
happens, the program might crash. However, a resourceful attacker might input a carefully
crafted data which includes malicious code. This triggers to overflow the buffer in such a way
that the flow of the target program is diverted and then the malicious program is executed, thus
allowing the attacker to compromise the program and ultimately the system. We model this
attack (denoted by ) in the following way:

6.11. Malicious programs: virus/worm/trojan horse
One major tool employed by the attacker is to exploit malicious programs known as viruses,
worms or trojan horses. RFC 1135 defines a computer virus and worm in the following way
[56]:
Virus: “A virus is a piece of code that inserts itself into a host, including operating
systems, to propagate. It cannot run independently. It requires that its host program be
run to activate it.”
Worm: “A worm is a program that can run independently, will consume the resources
of its host from within in order to maintain itself, and can propagate a complete
working version of itself on to other machines.”
On the other hand, a Trojan Horse (or simple trojan) is disguised as a benign program which,
once executed, can collect sensitive information from the target system [57]. To be most
effective, an attacker normally combines a trojan with a virus/worm. The trojan helps the
attacker to retrieve sensitive required information from the target system whereas the
virus/worm helps to deliver that information through the network to the attacker.
Even though a virus, worm and trojan horse are different in nature and in their functionalities,
we group them together under the category of malicious programs for simplicity. Then, we
model an attack, denoted with , involving a malicious program in the following way:

6.12. Denial-of-Service (DoS)
A Denial-of-Service (DoS) attack is a special class of attacks in which the attacker does not
attempt to gain access to the target system. Instead, the main motivation of launching this attack
is to disrupt or crush the target system so that legitimate users, networks, systems, or other
resources are denied to avail the services offered by the target system [58, 59]. Within this
motivation, the attack may target to overload the process/service, the computing and storage
resources or even attempt to forcefully shut down the part of the service/system. There are many
ways these can be achieved. Next, we present a few ways.
Bandwidth consumption: The most insidious form of DoS attacks is the bandwidth-
consumption attack. In this attack, attackers will attempt to consume all available
bandwidth to a particular network so that the target system becomes unreachable from
other systems.
Resource starvation: A resource starvation attack mostly focuses on consuming
resources, such as CPU utilisation, memory and storage quotas, of the target host. An
attacker generally is authorised to consume a certain amount of such resources.
However, he abuses the authorisation in order to consume additional resources in such
as way other users cannot use them anymore and thereby denying access to the system.
Routing and DNS attacks: A routing-based DoS attack is based on the idea that the
attacker manipulates the routing table enabling the attacker to route the traffic of a
victim to the attacker's system or to a black hole which is a network that does not exist
and thus denying the victim to access the requested service. On the other hand, in a
DNS-based DoS attack, an attacker compromises a DNS server to cache bogus DNS
information so that traffic towards the target system is routed to the attacker's system
and thereby denying other users to access services offered by the target system.
With tools readily available over the Internet, the attacker needs to possess little skills. That is
why DoS attacks are currently on the rise. Also, as more and more traditional as well as
innovative services are offered online with increasing popularities, the business motivations
attached to these services carry a massive monetary value. If such services are disrupted, it often
causes a significant amount of monetary loss to the corresponding business organisations. This
has attracted the attackers to carry out DoS attacks with the sole purpose of causing significant
monetary damages. The motivation of carrying out such an attack often involves scenarios
where an organisation would like to inflict monetary damages to other competitive
organisations. Moreover, there are personal as well as political vendettas that would drive an
attacker to carry out such an attack. We denote a DoS attack using  and model it in the
following way:

6.13. Distributed Denial-of-Service (DDoS)
A DoS attack is usually generated from a single source which, in reality, can cause any
significant damage as the source also has limited resources to carry out such an attack. This has
motivated an attacker to create a new form of attack called Distributed Denial-of-Service
(DDoS). A DDoS attack enables the attacker to launch a DoS attack targeted towards a victim
from a huge number of different sources [60, 61]. For this, the attacker needs to compromise as
many systems as possible by leveraging the attack methods presented before. Such a
compromised host is often referred to as a Zombie. Then, the attacker installs a specific DDoS
tool which remains dormant and preserves a connection with the attacker. Upon receiving a
signal from the attacker, the DDoS tool becomes active and participates in the DDoS attack
along with other compromised zombies.
In recent years, there has been a steady rise on the number of DDoS attacks, targeted towards
large online service providers or different countries [62]. With the prediction of a large number
of IoT (Internet-of-Things) devices connected to the Internet in near future, it is feared that
many of these IoT devices will be exploited to launch even larger type of DDoS attacks which
might be difficult to contain. In fact, we have already seen a DDoS attack involving IoT devices
[62].
We denote a Distributed DoS attack using  and model it in the following way:

7. TRACE REMOVAL
As a final step, the attacker tries to hide his activities just to ensure that the user/administrator of
the system cannot trace any source of attack back to the attacker. For this, the attacker may
subvert the logging/registry system to remove the captured logs or history of illicit activities
[63]. We represent this attack using  and model it in the following way:

8. APPLICATION
In this section we present a couple of applications of our model in order to illustrate the
applicability and usefulness of the model. The first application illustrates how the model can be
leveraged to create different types of categorisation and is presented in Section 8.1. On the other
hand, the second application sketches how the model can be extended for other scenarios and is
presented in Section 8.2.
8.1. Dimension-based categorisation
In Section 4, we have presented a taxonomy of cyber attacks based on the motivations of the
attacker. Interestingly, the model presented here can be utilised to create different taxonomies,
representing different categorisation, based on different aspects within the model. We exemplify
a few such taxonomies below. Other categorisations can be easily created following these
examples and hence, have been omitted for brevity.
Taxonomy based on channels. We can define a function called  which,
given a subset of different channels, returns a set of attacks which can be launched using those
channels. 
where, and .
We can use this function to classify attacks based on channels. A few examples follow:
The attacks which can be launched using the visual channel can be classified as:

The attacks which can be launched using either the visual or the network channel can be
classified as:

Here, indicates that there are other attacks belonging to this class which has been
omitted for brevity.
Taxonomy based on layers. Similarly, we can define a function called  which,
given a subset of different layers, returns a set of attacks which can be launched in those layers.

where, and .
We can use this function to classify attacks based on layers. A few examples follow:
The attacks which can be launched in the application layer () can be classified as:

Like before, indicates that there are other attacks belonging to this class which has
been omitted for brevity.
The attacks which can be launched using either in the internet layer () or in the
transport layer () can be classified as:

Likewise, indicates that there are other attacks belonging to this class which has
been omitted for brevity.
8.2. Modelling incident handling
An incident handling mechanism is a strategic process for any organisation to prepare itself with
a series of steps in case a security incident in the form of cyber attacks occurs. It is a crucial
strategy for any organisation to mitigate risks associated with cyber attacks. It consists of four
phases [64, 65]:
Preparation: In this phase, an organisation attempts to minimise the occurrence
likelihood of any security incident by taking proactive measures such as deploying
firewalls within the organisation’s network, malware protection, access control
mechanisms, real-time network monitoring and so on.
Attack detection and analysis: Even with the deployment of an array of strong
protective measures, there is always a probability for any cyber attack to occur. In this
phase, the organisation aims to detect and analyse such an attack.
Incident response. This phase aims to address and contain the identified cyber attack
incident so that the associated risk can be minimised as much as possible by adopting a
set of reactive approaches such as shutting down the infected system, changing the
password of compromised account/system and so on.
Post-incident. Finally, once the incident has been contained, it is required to reflect
upon the newly identified attack so that protective measures can be fed back to the
preparation phase. This is to ensure that the similar attack can be prevented at the initial
stage in future.
We model the incident handling mechanism using as the following:

Here,
 symbolises the preparation phase,
 symbolises the attack detection and analysis phase,
 symbolises the incident response phase,
 symbolises the post-incident phase,
the symbol denotes the sequence of operation and the indicates the
feedback loop.
Within the scope of this article, we focus on the  phase where the attack is detected and
analysed. We can intuitively model this phase using the model of cyber attacks introduced in
this article as follows:


This essentially denotes that all attacks identified in this phase (denoted with 
) are required to be detected and analysed.
9. CONCLUSION
In this article we have presented a novel mathematical model of cyber attacks. Using this model,
any cyber attack can be represented in a concise and intuitive way. The model encodes different
essential properties of a cyber attack. For example, the model expresses the victim, the target
entity for an attack, the channel and the layer used to launch the attack, the probable severity of
the attack and the type of the attack. Then, we have introduced a novel attack taxonomy based
on the motivations from the perspectives of attackers. We have also modelled each single
identified attack using our model. Finally, we have showcased the applicability of the presented
model in two scenarios.
The principle motivation of modelling cyber attacks according to our model is to prepare a solid
foundation for a formal analysis of attacks within a system, organisation and network. Within
this larger picture, modelling an attack is just one single component. The formal analysis will
additionally need to consider how an attack can successfully exploit any vulnerability, the
threats associated with each attack and how such attacks can be mitigated. In this article, we
have just shown how the incident handling mechanism can be modelled and how just one phase
(attack detection and analysis) can be represented using our attack model. However, we have
not explored the ways to model, represent and analyse other aspects of incident handling. The
presented model can be utilised to model other aspects of incident handling. It will also be
interesting to explore how the model can be combined with other existing attack formalising
approaches which explore petri nets and attack trees.
REFERENCES
[1] Ben Quinn and Charles Arthur. PlayStation Network hackers access data of 77 million users.
https://www.theguardian.com/technology/2011/apr/26/playstation-network-hackers-data. 26
April, 2011. Accessed: 14 February, 2017.
[2] Samuel Gibbs. Ebay urges users to reset passwords after cyberattack.
https://www.theguardian.com/technology/2014/may/21/ebay-urges-users-to-reset-passwords-
after-cyberattack. 21 May, 2014. Accessed: 14 February, 2017.
[3] Andrea Peterson. The Sony Pictures hack, explained.
https://www.washingtonpost.com/news/the-switch/wp/2014/12/18/the-sony-pictures-hack-
explained/?utm_term=.7d8221627572. 18 December, 2014. Accessed: 14 February, 2017.
[4] Kif Leswing. A massive cyberattack knocked out major websites across the internet.
http://uk.businessinsider.com/amazon-spotify-twitter-github-and-etsy-down-in-apparent-dns-
attack-2016-10. 21 October, 2016. Accessed: 14 February, 2017.
[5] Cara McGoogan. Yahoo hack: What you need to know about the biggest data breach in
history. http://www.telegraph.co.uk/technology/2016/12/15/yahoo-hack-need-know-biggest-
data-breach-history/. 15 December, 2016. Accessed: 14 February, 2017.
[6] Simon Hansman and Ray Hunt. A taxonomy of network and computer attacks. Computers &
Security, 24(1):3143, 2005.
[7] Bonnie Zhu, Anthony Joseph, and Shankar Sastry. A taxonomy of cyber attacks on SCADA
systems. In Internet of things (iThings/CPSCom), 2011, pages 380388. IEEE, 2011.
[8] M Uma and Ganapathi Padmavathi. A survey on various cyber attacks and their classification.
IJ Network Security, 15(5):390396, 2013.
[9] Nazrul Hoque, Monowar H Bhuyan, Ram Charan Baishya, DK Bhattacharyya, and Jugal K
Kalita. Network attacks: Taxonomy, tools and systems. Journal of Network and Computer
Applications, 40:307324, 2014.
[10] F Mohd Foozy, Rabiah Ahmad, MF Abdollah, R Yusof, and MZ Mas'ud. Generic taxonomy of
social engineering attack. 2011.
[11] Md. Sadek Ferdous, Farida Chowdhury, and Md. Moniruzzaman. A taxonomy of attack
methods on peer-to-peer network. In ICCIIS’ 07, p. 132138. 2007.
[12] Md Sadek Ferdous, Raid Hussein, Madini Alassafi, Abdulrahman Alharthi, Robert Walters, and
Gary Wills. Threat taxonomy for Cloud of Things. In Internet of Things and Big Data
Analysis: Recent Trends and Challenges, volume 1, pages 149191. United Scholars
Publications, USA, 2016.
[13] Andrew P Moore, Robert J Ellison, and Richard C Linger. Attack modeling for information
security and survivability. Technical report, DTIC Document, 2001.
[14] Steven Cheung, Ulf Lindqvist, and Martin W Fong. Modeling multistep cyber attacks for
scenario recognition. In DARPA information survivability conference and exposition, 2003.
Proceedings, volume 1, pages 284292. IEEE, 2003.
[15] Thomas M Chen, Juan Carlos Sanchez-Aarnoutse, and John Buford. Petri net modeling of
cyber-physical attacks on smart grid. IEEE Transactions on Smart Grid, 2(4):741749, 2011.
[16] Md. Sadek Ferdous, Gethin Norman, and Ron Poet. Mathematical modelling of identity,
identity management and other related topics”. In, SIN ’14, pages 9:9–9:16. ACM, 2014.
[17] Md Sadek Ferdous, Gethin Norman, Audun Jøsang, and Ron Poet. Mathematical modelling of
trust issues in federated identity management. In IFIP International Conference on Trust
Management, pages 1329. Springer International Publishing, 2015.
[18] Md Sadek Ferdous, Soumyadeb Chowdhury, and Joemon M Jose. Privacy threat model in
lifelogging. In UBICOMP 2016, pages 576581. ACM, 2016.
[19] M. S. Ferdous and R. Poet. Formalising identity management protocols. In 2016 14th Annual
Conference on Privacy, Security and Trust (PST), pages 137146. IEEE, Dec 2016.
[20] Terry Fleury, Himanshu Khurana, and Von Welch. “Towards a taxonomy of ‘attacks against
energy control systems. In International Conference on Critical Infrastructure Protection, pages
7185. Springer, 2008.
[21] TCP/IP Protocol Architecture. https://technet.microsoft.com/en-gb/library/cc958821.aspx.
Accessed: 5 March, 2017.
[22] Bruce Schneier. Secrets and lies: digital security in a networked world. John Wiley & Sons,
2011.
[23] Markus Huber, Stewart Kowalski, Marcus Nohlberg, and Simon Tjoa. Towards automating
social engineering using social networking sites. In CSE’09, volume 3, pages 117124. 2009.
[24] Gregory L Orgill, Gordon W Romney, Michael G Bailey, and Paul M Orgill. The urgency for
effective user privacy-education to counter social engineering attacks on secure computer
systems. In Proceedings of the 5th conference on Information technology education, pages
177181. ACM, 2004.
[25] Tim Thornburgh. Social engineering: the dark art. In Proceedings of the 1st annual conference
on Information security curriculum development, pages 133135. ACM, 2004.
[26] Enkhbold Nyamsuren and Ho-Jin Choi. Preventing social engineering in ubiquitous
environment. In Future Generation Communication and Networking (FGCN 2007), volume 2,
pages 573577. IEEE, 2007.
[27] Katharina Krombholz, Heidelinde Hobel, Markus Huber, and Edgar Weippl. Advanced social
engineering attacks. Journal of Information Security and applications, 22:113122, 2015.
[28] Sarah Granger. Social Engineering Fundamentals, Part I: Hacker Tactics.
https://www.symantec.com/connect/articles/social-engineering-fundamentals-part-i-hacker-
tactics. December, 2001. Accessed: 7 March, 2017.
[29] Hanan Sandouka, Andrea J Cullen, and Ian Mann. Social engineering detection using neural
networks. International Conference on CyberWorlds, 2009. CW’09. IEEE, 2009.
[30] Danesh Irani, Marco Balduzzi, Davide Balzarotti, Engin Kirda, and Calton Pu. Reverse social
engineering attacks in online social networks. In International Conference on Detection of
Intrusions and Malware, and Vulnerability Assessment, pages 5574. Springer, 2011.
[31] Lawrence Teo. Port scans and ping sweeps explained. Linux Journal, 2000(80es):2, 2000.
[32] Vladimir Gorodetski and Igor Kotenko. Attacks against computer network: Formal grammar-
based framework and simulation tool. In International Workshop on Recent Advances in
Intrusion Detection, pages 219238. Springer, 2002.
[33] Ofir Arkin. Network scanning techniques. Publicom Communications Solutions, 1999.
[34] Marco De Vivo, Eddy Carrasco, Germinal Isern, and Gabriela O de Vivo. A review of port
scanning techniques. ACM SIGCOMM Computer Communication Review, 29(2):4148, 1999.
[35] Will Gragido. Lions at the Watering Hole: The VOHO Affair. http://blogs.rsa.com/lions-at-
the-watering-hole-the-voho-affair/. July, 2012. Accessed: 21 March, 2017.
[36] Steve Stasiukonis. Social engineering, the USB way. Dark Reading, 7, 2006.
[37] Colin Tankard. Advanced persistent threats and how to monitor and deter them. Network
security, 2011(8):1619, 2011.
[38] Ping Chen, Lieven Desmet, and Christophe Huygens. A study on advanced persistent threats.
In IFIP International Conference on Communications and Multimedia Security, pages 6372.
Springer, 2014.
[39] Joint Task Force Transformation Initiative et al. Managing Information Security Risk:
Organization, Mission, and Information System View. 2011.
[40] Will Mitchell. Password cracking.
http://web.cs.du.edu/~mitchell/forensics/information/pass_crack.html. Accessed: 21 Feb, 2017.
[41] Xin Sun, Ruben Torres, and Sanjay Rao. DDoS attacks by subverting membership management
in p2p systems. In 3rd IEEE NPSec Workshop 2007, pages 16. IEEE, 2007.
[42] Common Types of Network Attacks. https://technet.microsoft.com/en-
us/library/cc959354.aspx. Accessed: 21 March, 2017.
[43] Edward W Felten, Dirk Balfanz, Drew Dean, and Dan S Wallach. Web spoofing: An internet
con game. Software World, 28(2):68, 1997.
[44] Matthew Tanase. IP spoofing: an introduction. Security Focus, 11, 2003.
[45] Spoofing Attack: IP, DNS & ARP. https://www.veracode.com/security/spoofing-attack.
Accessed: 21 February, 2017.
[46] Martin Johns. Session hijacking attacks. Encyclopedia of Cryptography and Security, pages
11891190, 2011.
[47] Mitja Kolšek. “Session fixation vulnerability in web-based applications. Acros Security, 2002.
[48] Nick Nikiforakis, Wannes Meert, Yves Younan, Martin Johns, and Wouter Joosen.
Sessionshield: Lightweight protection against session hijacking. In International Symposium
on Engineering Secure Software and Systems, pages 87100. Springer, 2011.
[49] Session Hijacking Cheat Sheet. http://resources.infosecinstitute.com/session-hijacking-cheat-
sheet. January, 2015. Accessed: 15 April, 2017.
[50] Shray Kapoor. Session hijacking exploiting tcp, udp and http sessions.
www.infosecwriters.com/text_resources/pdf/SKapoor_SessionHijacking.pdf. 2006.
[51] Alberto Ornaghi and Marco Valleri. Man in the middle attacks. In Blackhat Conference
Europe, 2003.
[52] Alex Kirshon, Dima Gonikman, and Gabi Nakibly. Owning the routing table - new OSPF
attacks. BlackHat Briefings and Trainings USA+ 2011, pages 118, 2011.
[53] Ola Nordström and Constantinos Dovrolis. Beware of bgp attacks. ACM SIGCOMM
Computer Communication Review, 34(2):18, 2004.
[54] Crispan Cowan, Calton Pu, Dave Maier, Jonathan Walpole, Peat Bakke, Steve Beattie, Aaron
Grier, Perry Wagle, Qian Zhang, and Heather Hinton. Stackguard: automatic adaptive detection
and prevention of buffer-overflow attacks. In Usenix Security, volume 98, pages 6378, 1998.
[55] Cowan, Crispin, F. Wagle, Calton Pu, Steve Beattie, and Jonathan Walpole. Buffer overflows:
Attacks and defenses for the vulnerability of the decade”. In DARPA Information Survivability
Conference and Exposition, 2000. DISCEX'00. Proceedings, vol. 2, pp. 119-129. IEEE, 2000.
[56] Joyce K Reynolds. The helminthiasis of the internet (rfc 1135), December 1989, 1989.
[57] What is a Trojan Virus? Definition. https://usa.kaspersky.com/internet-security-
center/threats/trojans#.WNwsIPnyvIU. Accessed: 17 April, 2017.
[58] Jelena Mirkovic, Sven Dietrich, David Dittrich, and Peter Reiher. Internet denial of service:
Attack and defense mechanisms (radia perlman computer networking and security). 2004.
[59] Christoph L Schuba, Ivan V Krsul, Markus G Kuhn, Eugene H Spafford, Aurobindo Sundaram,
and Diego Zamboni. Analysis of a denial of service attack on tcp. In Security and Privacy,
1997. Proceedings., 1997 IEEE Symposium on, pages 208223. IEEE, 1997.
[60] Felix Lau, Stuart H Rubin, Michael H Smith, and Ljiljana Trajkovic. Distributed denial of
service attacks. In IEEE International Conference on Systems, Man, and Cybernetics, pages
22752280. 2000.
[61] Jelena Mirkovic, Janice Martin, and Peter Reiher. A taxonomy of ddos attacks and ddos defense
mechanisms. ACM SIGCOMM Computer Communication Review, 34(2):3953, 2004.
[62] David Bisson. The 5 Most Significant DDoS Attacks of 2016. https://www.tripwire.com/state-
of-security/security-data-protection/cyber-security/5-significant-ddos-attacks-2016/. 29
November, 2016. Accessed: 17 April, 2017.
[63] How to Cover Your Tracks & Leave No Trace Behind on the Target System. https://null-
byte.wonderhowto.com/how-to/hack-like-pro-cover-your-tracks-leave-no-trace-behind-target-
system-0148123/. August, 2013. Accessed: 17 April, 2017.
[64] Nurul Hidayah Ab Rahman and Kim-Kwang Raymond Choo. A survey of information security
incident handling in the cloud. Computers & Security, 49:4569, 2015.
[65] Bernd Grobauer and Thomas Schreck. Towards incident handling in the cloud: challenges and
approaches. In Proceedings of the 2010 ACM workshop on Cloud computing security
workshop, pages 7786. ACM, 2010.
Authors
Farida Chowdhury is an Assistant Professor at the department of Computer Science and Engineering of
the Shahjalal University of Science & Technology, Bangladesh. She received her PhD degree at the
Institute of Computing Science and Mathematics at the University of Stirling, Scotland. Her research
interests focus on Network security, P2P networks, Pervasive computing, Next-Generation Wireless
Networks, Wireless Sensor Networks, Internet of Things and Social Networks.
Md Sadek Ferdous is a Research Fellow at the Electronics and Computer Science of the University of
Southampton. He holds a PhD in the area of Mobile Identity Management at the School of Computing
Science of the University of Glasgow in 2015. His research interest includes Network Security,
Distributed Ledger, Identity Management, Trust Management and Privacy Enhancing Technologies.
... Currently, the devices are too much up to date. Unfortunately, the number of cyber-attack models is reasonably low, and even the previous models break to represent all the required properties of a cyber-attack [5]. So, we need to think of different models which will be simple to comprehend. ...
... The standard traditional cyber-attack model cannot attack any private network, computer or server [5] [7] [10] because the private network is not directly connected to the internet. However, this model can be capable of striking any private network through a company's employee. ...
... However, it can be noticed that there are very few studies on mathematical modelling of social engineering that could help organizations to make an assessment of the risk of such an attack. In one of these studies, phishing, which is the most common social engineering attack, is described as following [5]: ...
... Referring to critical infrastructure, Baig and Zeadally described the following model of risk mitigation E i [10]: (5) The model described in equation (5) can be used to show the capacity of an organization to mitigate social engineering attacks, but also resides on malicious events that are ongoing or already took place. ...
Article
Full-text available
Social engineering is a very common type of malicious activity conducted on cyberspace that targets both individuals and companies in order to gain access to information or systems. It is part of the broader domain of cybersecurity and the first step to mitigate this type of attack is to know its attack vectors. This way, the risk of becoming a victim of this type of attack can be reduced by technical means, proper security culture and procedural solutions – if organizations are referred to.
... Attack is any cybercriminal action, try to access confidential data the security of facts owned via an organization using any system that designed to detect [13]. Cyber-attacks from different perceptions are critical in order to moderate them [14]. There are several forms of attacks; however, the most commonplace security attacks are described. ...
Article
Full-text available
The main purpose of this study is to analyses the targeted cyber-attacks of Pakistani banks that happened or targeted in 2018 and the solution to control the crimes. The aim of the study is to obtain further information on the impact of cybercrime on the Pakistani banking sector. This study examines the important contribution that raising awareness of the security of information about the relationship between cybercrime and organised services can make. The impact of cybercrime on the Organization's activities will be examined by deepening the moderating effects of raising awareness of cyber security. Cybercrime has a undesirable impact on the organization's performance, but knowledge of cyber security weakens the negative impact of cybercrime on the organization's performance. This study focuses on the banking sector and therefore cannot be extended to other sectors. In addition, comprehensive relative studies in other areas with different cultural contexts will contribute to the validation of the research results. Awareness about the security of information weakens the negative effect of cybercrime on performance; therefore, it is important for banks, security, human resource supervisors, training to raise awareness of employees about cybercrime. Cyber-attacks, threats, vulnerabilities, Security attacks and challenges combination of these topics has led to a new study in the field of cybercrime. This study also improves the understanding of the role of employees in combating the effect of cybercrime on organizational performance.
Conference Paper
Full-text available
With the absence of physical evidence, the concept of trust plays a crucial role in the proliferation and popularisation of online services. In fact, trust is the inherent quality that binds together all involved entities and provides the underlying confidence that allows them to interact in an online setting. The concept of Federated Identity Management (FIM) has been introduced with the aim of allowing users to access online services in a secure and privacy-friendly way and has gained considerable popularities in recent years. Being a technology targeted for online services, FIM is also bound by a set of trust requirements. Even though there have been numerous studies on the mathematical representation, modelling and analysis of trust issues in online services, a comprehensive study focusing on the mathematical modelling and analysis of trust issues in FIM is still absent. In this paper we aim to address this issue by presenting a mathematical framework to model trust issues in FIM. We show how our framework can help to represent complex trust issues in a convenient way and how it can be used to analyse and calculate trust among different entities qualitatively as well as quantitatively. © IFIP International Federation for Information Processing 2015.
Conference Paper
Full-text available
There exist disparate sets of definitions with different se-mantics on different topics of Identity Management which often lead to misunderstanding. A few efforts can be found compiling several related vocabularies into a single place to build up a set of definitions based on a common semantic. However, these efforts are not comprehensive and are only textual in nature. In essence, a mathematical model of iden-tity and identity management covering all its aspects is still missing. In this paper we build up a mathematical model of different core topics covering a wide range of vocabular-ies related to Identity Management. At first we build up a mathematical model of Digital Identity. Then we use the model to analyse different aspects of Identity Management. Finally, we discuss three applications to illustrate the ap-plicability of our approach. Being based on mathematical foundations, the approach can be used to build up a solid understanding on different topics of Identity Management.
Article
Full-text available
Social engineering has emerged as a serious threat in virtual communities and is an effective means to attack information systems. The services used by today's knowledge workers prepare the ground for sophisticated social engineering attacks. The growing trend towards BYOD (bring your own device) policies and the use of online communication and collaboration tools in private and business environments aggravate the problem. In globally acting companies, teams are no longer geographically co-located, but staffed just-in-time. The decrease in personal interaction combined with a plethora of tools used for communication (e-mail, IM, Skype, Dropbox, LinkedIn, Lync, etc.) create new attack vectors for social engineering attacks. Recent attacks on companies such as the New York Times and RSA have shown that targeted spear-phishing attacks are an effective, evolutionary step of social engineering attacks. Combined with zero-day-exploits, they become a dangerous weapon that is often used by advanced persistent threats. This paper provides a taxonomy of well-known social engineering attacks as well as a comprehensive overview of advanced social engineering attacks on the knowledge worker.
Article
The role of computers and the Internet in modern society is well recognized. Recent developments in the fields of networking and cyberspace have greatly benefited mankind, but the rapid growth of cyberspace has also contributed to unethical practices by individuals who are bent on using the technology to exploit others. Such exploitation of cyberspace for the purpose of accessing unauthorized or secure information, spying, disabling of networks and stealing both data and money is termed as cyber attack. Such attacks have been increasing in number and complexity over the past few years. There has been a dearth of knowledge about these attacks which has rendered many individuals/agencies/organizations vulnerable to these attacks.[7] Hence there is a need to have comprehensive understanding of cyber attacks and its classification. The purpose of this survey is to do a comprehensive study of these attacks in order to create awareness about the various types of attacks and their mode of action so that appropriate defense measures can be initiated against such attacks.
Conference Paper
In this paper we present the formalisation of three well-known Identity Management protocols - SAML, OpenID and OAuth. The formalisation consists of two steps: formal specification using HLPSL (High-Level Protocol Specification Language) and formal verification using a state-of-the-art verification tool for security protocols called AVISPA (Automated Validation of Internet Security Protocols and Applications). The existing formalisation initiatives using AVISPA are based on SAML and OpenID, leaving OAuth entirely, even though OAuth is one of the most widely-used Internet protocols. Furthermore, the motivation of the existing initiatives was to identify any weakness. In this paper, we have taken an opposite approach as we are keen to present how to model these protocols correctly. Moreover, our formalisation is based on a model of identity and also captures the authentication mechanism; both of these are missing in the existing works.
Conference Paper
The lifelogging activity enables a user, the lifelogger, to passively capture multimodal records from a first-person perspective and ultimately create a visual diary encompassing every possible aspect of her life with unprecedented details. In recent years it has gained popularity among different groups of users. However, the possibility of ubiquitous presence of lifelogging devices especially in private spheres has raised serious concerns with respect to personal privacy. Different practitioners and active researchers in the field of lifelogging have analysed the issue of privacy in lifelogging and proposed different mitigation strategies. However, none of the existing works has considered a well-defined privacy threat model in the domain of lifelogging. Without a proper threat model, any analysis and discussion of privacy threats in lifelogging remains incomplete. In this paper we aim to fill in this gap by introducing a first-ever privacy threat model identifying several threats with respect to lifelogging. We believe that the introduced threat model will be an essential tool and will act as the basis for any further research within this domain.
Conference Paper
A recent class of threats, known as Advanced Persistent Threats (APTs), has drawn increasing attention from researchers, primarily from the industrial security sector. APTs are cyber attacks executed by sophisticated and well-resourced adversaries targeting specific information in high-profile companies and governments, usually in a long term campaign involving different steps. To a significant extent, the academic community has neglected the specificity of these threats and as such an objective approach to the APT issue is lacking. In this paper, we present the results of a comprehensive study on APT, characterizing its distinguishing characteristics and attack model, and analyzing techniques commonly seen in APT attacks. We also enumerate some non-conventional countermeasures that can help to mitigate APTs, hereby highlighting the directions for future research.
Article
Incident handling strategy is one key strategy to mitigate risks to the confidentiality, integrity and availability (CIA) of organisation assets, as well as minimising loss (e.g. financial, reputational and legal) particularly as organisations move to the cloud. In this paper, we surveyed existing incident handling and digital forensic literature with the aims of contributing to the knowledge gap(s) in handling incidents in the cloud environment. 139 English language publications between January 2009 and May 2014 were located by searching various sources including the websites of standard bodies (e.g. National Institute of Standards and Technology) and academic databases (e.g. Google Scholar, IEEEXplore, ACM Digital Library, Springer and ScienceDirect). We then propose a conceptual cloud incident handling model that brings together incident handling, digital forensic and the Capability Maturity Model for Services to more effectively handle incidents for organisations using the cloud. A discussion of open research issues concludes this survey.
Article
Open Shortest Path First (OSPF) is the most popular interior gateway routing protocol on the Internet. Most of the known OSPF attacks are based on falsifying the link state advertisement (LSA) of an attacker-controlled router. These attacks may create serious damage if the attacker-controlled router is strategically located. However, these attacks can only falsify a small portion of the routing domain's topology; hence their effect is usually limited. More powerful attacks are the ones that affect LSAs of other routers not controlled by the attacker. However, these attacks usually triggerthfight-back" mechanism by the victim router which advertises a correcting LSA, making the attacks' effect non-persistent. In this work we present new attacks that exploit design vulnerabilities in the protocol specification. These new attacks can affect the LSAs of routers not controlled by the attacker whileevadifight-back". These attacks afford an attacker a greater power to persistently falsify large portions of the routing domain's topology. This allows an attacker to effectively own the routing tables of the routers in the AS without actually owning the routers themselves. This may be utilized to induce routing loops, network cuts or longer routes in order to facilitate DoS of the routing domain or to gain access to information flows which otherwise the attacker had no access to. The main implication of this work is the new recognition that by controlling a single router the attacker can control the entire routing domain.