Analysis of Complex Networks for Security Issues using Attack Graph

Full text

2019 International Conference on Computer Communication and Informatics (ICCCI -2019), Jan. 23 – 25, 2019, Coimbatore, INDIA
Analysis of Complex Networks for Security Issues
using Attack Graph
Abstract- Organizations perform security analysis for
assessing network health and safe-guarding their growing
networks through Vulnerability Assessments (AKA VA Scans).
The output of VA scans is reports on individual hosts and its
vulnerabilities, which, are of little use as the origin of the attack
can’t be located from these. Attack Graphs, generated without an
in-depth analysis of the VA reports, are used to fill in these gaps,
but only provide cursory information. This study presents an
effective model of depicting the devices and the data flow that
efficiently identifies the weakest nodes along with the concerned
vulnerability’s origin.
The complexity of the attach graph using
MulVal has been greatly reduced using the proposed approach of
using the risk and CVSS base score as evaluation criteria. This
makes it easier for the user to interpret the attack graphs and
thus reduce the time taken needed to identify the attack paths
and where the attack originates from.
Keywords- Network Vulnerabilities, Vulnerability
Assessment, Attack Graph, Attack Graph Generation Tools.
I. I
NTRODUCTION
Today’s information is stored and processed in different
electronic forms through fleet of computing devices and its
networks. On the rise of information exchange through these
resources, current establishments face massive challenges in
securing the information. Firewalls have been deployed
widely and extensively to block unauthorized access to
systems from all but a few, well defined access ports.
However, these devices are unable to uproot the diverse kinds
of security threats that are being seen now days, nor detect
attacks when they transpire [1].
As technology advances, the security challenges are
increasing exponentially, which means there is a significant
increase in information usage and also reveals that computing
networks and its resources will be considerably attacked to
compromise the security of the information stored. With
trending technologies, offering wide variety of services in
helping an individual or an organization to store and process
their information, relying heavily on the computing
environment, only marks the importance of network security
to grow and strengthen [2]. A great example is the
proliferation of Cloud Computing or delivery of services
through Clouds. This delivery is done through vast
interconnection of computer networks where ARP (Address
Resolution Protocol) spoofing has become a significant threat
against this emerging technology [3], even though strong
encryption techniques, suggested by W. Diffie and M.
Hellman, is in place [4], and varieties of encryption techniques
are available [24].
Though, Hunt & Zeadally [5] believe that many security
controls and tools are employed, from the perimeter level to
endpoint level of the organization to address the security
problems, still the networked infrastructures routinely come
under attacks which are often sophisticated enough to combine
multiple vulnerabilities to bypass the security controls.
Furthermore, what is often seen is that organizations find
it difficult to quantify the risks posed from internal network.
When it is about analysing the enterprise security, one must
think of multi-staging and multi host attacks. Based on
Collin’s [6] statement the situation warrants such an approach
which at the very beginning, will analyse the network
configuration and identify the security weaknesses; so, the
network graphs are to be denoted with the attack paths by
simulating multi stage and multi host attack processes [6].
The intent of this research is to quantitatively asses the
attacks performed on the computing networks. The network
attack graphs will prove handy in visualizing the attack pattern
of multi-stage network/hosts attacks in the form of graphs.
Also, this research discusses existing systems and some of
trends hackers follow to compromise networks.
Furthermore, the research will move ahead deriving the
proposed system mined from vulnerability trends. In parallel,
the effectiveness of this approach is measured with projected
increase in attacks and how this model can defy hackers with
dynamically emerging system.
978-1-5386-8260-9/19/$31.00 ©2019 IEEE
Fernaz Narin Nur, Fahad Faisal
Department of Computer Science and Engineering
Daffodil International University, Bangladesh
narin@daffodilvarsity.edu.bd, fahad.cse@diu.edu.au
Tanvirali Musa, Kheng Cher Yeo, Sami Azam,
Bharanidharan Shanmugam, Asif Karim,
Friso De Boer
College of Engineering, IT and Environment
Charles Darwin University, NT, Australia
musatanvir@gmail.com,

2019 International Conference on Computer Communication and Informatics (ICCCI -2019), Jan. 23 – 25, 2019, Coimbatore, INDIA
II.
L
ITERATURE REVIEW
A. Introduction to attack graphs
Along with hosts on the network, vulnerabilities are also
increasing proportionately; it is evident that the process of
evaluating vulnerabilities needs to be automated. While
evaluating the security of computer networks, consideration
need to be given to the identified isolated vulnerabilities [7].
When it comes to large-scale networks, it contains numerous
platforms and multiple software packages employed with
several modes of connectivity. Inevitably, these networks have
vulnerabilities which cannot even be noticed by the system
administrator [8]. Automatic generation of Attack Graphs
through symbolic model checking algorithms have also been
proposed to make the task easier [9]. Attack graph systems
employ sophisticated techniques concentrating on the
individual exploits which has the potential to be part of the
attack path [10].
A probabilistic approach to explore attack graphs can
also be used to find out the intention of the attack and the
probable attack paths [25]. Applying the mechanisms of the
attack graphs one can answer the questions on “How an
attacker can break into the network, is there any detectable
path?” [11].
B. Network attack graphs
A network attack graph represents a collection of
probable exploitable scenarios on a given computer networks.
Each scenario shows the steps followed by an attacker to
achieve his goal which can range from an administrative
access, database access, disruption of services to even spying.
In a professionally constructed network model, an attack graph
can produce an eagle-eye view for every scenario which can
lead to a security breach [12].
C. Attack graph tools
This section briefly discusses some of the common attack
graph generation tools:
1. Mulval
An open source logic-based tool used to generate attack
graphs. MulVAL stands for Multi stage Multi host
Vulnerability Analysis, authored by Xinming Ou. Basically,
the generated attack graph has attack-step nodes. Nodes are of
three types, represented in oval (attack state), diamond
(privilege nodes) and rectangular (configuration nodes) shapes
[13] and it is a command line interface with O (n
2
) ~O (n
3
)
complexity. Input files, which are submitted to this tool, are of
(.P) format and there are adapters to generate this file. These
adapters help in creating (.P) files of the reports generated
from the VA sources like Nessus and OpenVAS. The output
of these scanners is of .Nessus/Oval/XML format. The VA
report will be altered to (.P) format with the help of adapters
present in MulVAL tool; and finally the attack graph gets
generated according to the logic present in logic-execution-
engine [14]. MulVAL's framework is an integration of five
parts which includes rules of interaction, logical-execution-
engine, security policies, database (analytical), attack path and
unauthorized access.
Rules of interactions are the points which refer to
statements from Data Log. The configuration information
submitted to database and the rules in the database can
simulate the behavior of attacker on the network.
2. Topological Vulnerability Analysis (TVA)
TVA is another tool to generate the attack graphs. This
tool has capacity to analyze network vulnerabilities
automatically and dig out weaknesses to generate the attack
graph. A state transition diagram is established according to
the attack conditions and procedures, providing network
vulnerability analysis scalable to any size of the network.
3. Net SPA
Net SPA stands for Network Security Planning
Architecture, the attack graphs are used to model the
adversaries and the impact by providing counter measures.
The attack graph generated is termed as Multiple Prerequisite
Graph (MP Graph). It delivers a network model devised
through firewall rules and network vulnerability scans. It has
the capacity to find out the most effective attack path on the
given network topology which directly helps in providing an
effective solution to long term threats. The software uses a
host, running services and given network information to model
an attack graph which can show an attacker’s view on
infiltrating the network.Net SPA can generate analytical
suggestions on the attack graph on how to remediate the most
severe vulnerabilities in the network [15].
Net SPA also helps in identifying the critical hosts where
the vulnerability of that particular host becomes a key node to
be under stack compared to other hosts (node). Thus, Net SPA
greatly aids administrators in identifying the critical host first
and patch it up immediately before any causalities caused by
the attacker. Net SPA’s limitation is that the graph has many
loops which make it harder for network administrators to
understand and manage things effectively.
D. Tools selection
With reference to the comparison matrix of both
vulnerability scanning tools and attack graph tools, the
following section justifies the selecting of tools for the test
environment:
1. Attacker’s activity
This paper deals with probabilistic ways adopted by an
attacker to fulfill their intent in breaking into the systems by

2019 International Conference on Computer Communication and Informatics (ICCCI -2019), Jan. 23 – 25, 2019, Coimbatore, INDIA
compromising the security. The normal procedure adopted by
any attackers will be carried out in phases as shown below in
figure 1. Attacker will pass through different stages and the
success rate of the attack can only be high when data gathered
at each phase is precise and accurate.
Fig. 1. Attack Phases [13, 16]
E. Findings about attack graph generation tools
1. Comparison matrix – attack graph tools
Table 1: Comparison Matrix - Attack graph tools
Reconnaissance is the stage where intruder would gain as
much information about the network. Details of target network
are learnt at this phase and the IP address and its network
connectivity is understood.
In the next phase, the attacker now tries to understand the
weakness of the entire system and of individual nodes. The
attacker uses a vulnerability scanner and looks for open ports,
open services, application exploits, and loopholes in data
transit. In the following phase (Gaining Access) the host is
compromised to either extract information of value from that
node, or to use that node\host to further launch attacks on
other targets. The technique is applied following the probable
path-remote exploitation followed by execution of code to
exploit the weakness in the host.
Once the attacker gains access, the next phase kicks
in, that is to maintain this access. Attackers may decide how
deep they want to get in, but this phase can increase the
attacker's vulnerability to detection with every passing minute
[16].
The final phase of covering tracks simply means the
attacker completes all steps necessary to eradicate all
semblance of detection. This phase will not be consulted in
detail in this work as this paper deals with predicting
attacker’s intent by analyzing the vulnerabilities identified
with respect to the hosts of the network.
In a real-time scenario referring to the experimental network
topology on how an attacker plans attack. The attacker
initiates by performing a reconnaissance over the network, by
using most widely used open-source tool like Nmap to identify
the hosts that are part of the network along with the type of
services running on those hosts. Next the second phase works
on the information acquired by the first phase. Every
identified vulnerability has an assigned reference number
which will be unique and gets listed into National
Vulnerability Database (NVD) as well as in MITRE system
[16]. The database contains complete information about the
identified vulnerability besides loads of other useful
information.
The following vulnerabilities have been reported in the
vulnerability scan performed on the example. The below
vulnerabilities are well-known and are easily exploitable.
Summary of the vulnerabilities are given below:
• CVE-2008-4835 is known to affect SMB (Server Message
Block) service through memory corruption vulnerability
which may allow attacker to execute malicious code or
attacker can also carry out denial of services against the
remote hosts.
• CVE-2008-4250 is identified with triggering buffer
overrun issues in the “server” service which lets an
attacker to execute arbitrary code in the remote host with
the system privileges [17].
• CVE-2014-6321 is related to weakness in processing the
packets over a secure channel (SChannel) allowing
attacker to craft the packets on their way to the server.
• CVE-2012-0152 is known to show up when there is a
RDP service enabled on the vulnerable system and
unauthenticated user can leverage this vulnerability to
execute malicious code by sending RDP packets.
Considering the above weak link, it is clear there are two
types of vulnerabilities which can be exploited at different
levels. First is the remotely exploited vulnerability which
works over a network and exploit the machine without any
prior access to the vulnerable machine; the other is a local
exploit requiring access to the machine prior taking over that
very host. Post exploitation will escalate the privilege levels to
administrator.
2. Target environment and vulnerability correlation
It is vital for any prediction attack graphs to have collected
information about the systems and associated vulnerabilities.
Since the graph is purely dependent on the number of hosts

2019 International Conference on Computer Communication and Informatics (ICCCI -2019), Jan. 23 – 25, 2019, Coimbatore, INDIA
and its vulnerabilities it is vital to perform a comprehensive
scan of every host. If anyone of the prime host’s information
is missing or incomplete, the created graph would not be
effective. Other important feature is runtime configuration as
it requires HACL (host access control list), which has the
information of attacker whereabouts, running network services
and user accounts [16]. Another characteristic is attacker’s
logs which get generated during the process. Every attacker
uses their own tactics and every step of the approach will be
clearly documented.
F. Methods on reducing attack graphs complexity
The complexity of the attack graph is basically
determined by two important factors. Number of Hosts N(h)
and Number of identified vulnerabilities present in the NVD
database number of vulnerability N(v). For instance, consider
a network with n number of hosts and after scanning an attack,
action have been performed on the n hosts and is
approximately represented as:
F(N(h), N(v)) = N(h)N(v)F(N(h)-1,N(v)) = N(h)N(v)(N(h)-
1)N(v)F(N(h)-2,N(v))= N(v)
N(h)
N(h)! [21].
This shows that the approach faced a combinational explosion
with respect to complexity. Therefore it is more suitable to
smaller networks, but not applicable unless there is a
modification for large networks [19].
In 2009, researchers [19] described about Model Checking
which was in use to enumerate the attack chains to link initial
access points to the goal of the attacker. Due to explicit
enumeration of attacker’s state, these families of approaches
are always growing exponentially in proportion to the size of
the network. Monotonic logic helped attack graph’s
complexity subside to polynomial from exponential. The
complexity was further reduced while having quadratic
number of hosts. It is also possible to bring the complexity
down by grouping of networks into single domain where
connectivity among the hosts is not restricted and this domain
has tight security protection rules already in place. With this
kind of topology, complexity will reduce to linear considering
single domain; generally, the complexity swells to quadratic
depending on the count of the protected domains (as the count
will be domain number but not host hence it will be a lot lesser
than expected). Such graphs can be produced from a mere
hundred to tens of thousands of hosts within minutes but not
with visualizations. Attempts were also made to measure the
network security risk in combination of individual
vulnerabilities and its relevant metrics. Converting the attack
graphs and vulnerability score to Bayesian networks for better
computation of cumulative probability has been proposed by
Frigault et al. [20] in his paper, explained a better approach on
recognizing the cycles which are existing in attack graphs..
Singhal’s work was quite meaningful research in
improving visualizations of network security architecture. For
any environment the preliminary point is to quantify the attack
surface and its impacts because it is the factor to control risk
posed to the computer networks described in his research [21].
Due to explicit enumeration of attack states, attack graphs
become considerably convoluted. With the work in monotonic
graph generation, complexity for the same graph reduces to
polynomial from exponential [22]. Alhomidi and Reed [23],
proposed a methodology to explore the graph using genetic
algorithm where each attack path is an attack scenario from its
source to attacker’s goal. This evolves to be a natural way to
generate maximum number of possible attack paths which
again makes the graph a lot more complex gradually.
III.
P
ROPOSED
APPROACH
–
(RISK
AND
CVSS
BASE
SCORE
AS
EVALUATION
CRITERIA)
Vulnerability scans were performed using Nessus and the
output of the scan can be extracted in multiple formats like
Nessus dB, csv, html, pdf, .Nessus. MulVAL and Nessus
complement each other, MulVAL have utilities which
supports and converts the Nessus file formats to MulVAL
readable files. Then these readable files are processed for
graph generations. The process has been depicted in figure 2.
Hence before processing the vulnerability report from
Nessus scanner to MulVAL’s framework, it is possible to
analyze and figure out for any false positives, vulnerabilities
which are outdated and also to identify those vulnerabilities
where CVE-IDs have been registered but do not demonstrate
any effectiveness. Thus considering all these factors the
current research works in a direction where the output of
vulnerability assessments are thoroughly evaluated before
generating the attack graphs leading to improvement of
complexity. MulVAL identifies the vulnerabilities based on
the CVE-IDs but the output of the scan is not precisely
evaluated, hence there is a need to verify the output generated
considering Risk and CVSS score as the factors.
Nessus vulnerability scanner provides with the facility to
export a filtered vulnerability report based on user
requirement. Normally in this approach evaluation of Risk
along with Common Vulnerability Scoring System (CVSS)
base score have been used. Vulnerability with CVSS base
score of 10 -7 were considered the first vulnerabilities on the
network to be addressed as these vulnerabilities will have
exploits available.
The reason for selecting this range is that the
vulnerabilities within this range can be remotely exploited,
meaning these vulnerabilities provide attacker a gateway for
successful exploitation.
So once these gets identified and dealt with, and assigned
the highest priority levels, the gap for the attacker gets closed.
Depending on the Risk Factor and associated CVSS Score,
vulnerability is defined.

2019 International Conference on Computer Communication and Informatics (ICCCI -2019), Jan. 23 – 25, 2019, Coimbatore, INDIA
Fig. 2. A Flowchart for the Proposed Approach
1) Risk Factor - This helps while segregation of
vulnerabilities according to its risk level (critical,
high, medium, low, informational).
2) CVSS - Common Vulnerability Scoring System
(CVSS), it is an open framework. Any
software/hardware/firmware vulnerabilities can be a
threat to the entire organization and it is quite
difficult to mitigate. CVSS provides a way to capture
the characteristics of the vulnerability and assign a
score which is a numerical value indicating its
severity. This numerical score is translated into a
qualitative representation which is as follows.
• Critical (Risk) Vulnerability – 10.0 (CVSS
Base Score)
• High (Risk) Vulnerability – 9.0 -7.0 (CVSS
Base score)
• Medium (Risk) Vulnerability – 6.9 – 4.0
(CVSS Base Score)
Rest is not suitable to the current proposed approach.
IV.
R
ESULTS
AND
DISCUSSIONS
The attack graph generated was quite comprehensive in
providing the information of the attacks. The graph is better
than the previously generated attack graphs based on the
system generated report. Every tool/software result contains
some degree of false positives, reports are to be thoroughly
analyzed based on vulnerability exploitable factors and then
an attack graph generation would be something which has real
effect. Hence the graph seems to be simple compared to the
first attack graph. It is also observed that the attack loops
drastically dwindle. There are few other factors which are
helping the cause, like, those which have a CVE-ID but cannot
be exploited.
It is mandatory for one to understand though there is a
vulnerability registered with CVE-ID, risk posed by the
vulnerability should also be considered along with availability
of exploits. If any of the mentioned factors associated to a
vulnerability is missing, it cannot be exploited, which means
the attack path generated in supporting these vulnerabilities
considered to be “true negative”- because vulnerability was
detected but still couldn’t be exploited. Hence instances like
these add to the complexity of the graph and complicate it,
rendering it hardly readable. With respect to the above
generated graph, the presented paths are the potential paths of
the attack for breaking into the network.
It is also understood from the analysis that vulnerabilities
which can be remotely exploitable are the prime concern to
any networks. Since they can be accessed across the network
and in case of being hosted through internet, the risk of threat
is top notch. The final attack graph is one such graph with
special concentration on remotely exploitable vulnerability
family. If these vulnerabilities are rectified, it will then
basically end the road for the attacker to sneak into the
network, leading to zero attack paths.
A. Findings
It is clearly depicted in Table II the major difference in the
count of the vulnerabilities alongside the number of nodes.
Hopefully now due to this approach the graph has only those
attack paths which have potential to become attacker’s
probable paths because this is after adopting the new approach
of only processing precise data which can generate better
reliable graph. The Nessus report is thoroughly evaluated
based on vulnerability risk and exploitable features which
were not done earlier. This evaluation helps identifying the
false positives and those vulnerabilities which cannot create
any impact.
In previous approach Nessus was used to generate
information and that same data were processed into MulVAL
framework. There was no evaluation of the data which were
being generated by Nessus hence the graph was misleading,
imperfect and above all too complex to be understood
properly.
Hence there was a necessity to evaluate the Nessus output
which has possible number of vulnerabilities that cannot be
served to be potential enough to create an impact. Those
vulnerabilities have been identified and excluded which

2019 International Conference on Computer Communication and Informatics (ICCCI -2019), Jan. 23 – 25, 2019, Coimbatore, INDIA
directly helped in improving the complexity of the attack
graph.
Table 2: Comparison table between two Attack graphs
Complexity Factors Attack Graph Attack Graph
(Post Evaluation
Approach)
No. of hosts 6 4
Total no. of
vulnerabilities
10 5 (6th one is an
outlier)
No. of nodes 100 53
No. of attack hoops 38 9
No. of vulnerabilities
with "Risk"=None
2 0
No. of vulnerabilities
without exploits available
5 0
V.
C
ONCLUSIONS
AND
FUTURE
WORK
The aim of this research is to analyze the security of
networks using attack graph concepts and reduce the
complexity of attack graph. However, even though this tool is
slightly complex but in general provides good foundation for
research work with respect to attack graphs, despite the issue
of spending some extra hours filtering out the false positives.
Future research could be working with other open source
vulnerability scanners and incorporating the attack graph into
open source scanners like NMAP. Other ways of reducing the
complexity of the attack graphs can also be explored. This
would enable the network security administrators to have clear
idea of the attack and where it originates from.
REFERENCES
[1]. M. Bennet, S.S., M. Deepika, N. Nanthini, S. Bhuvaneshwari &
M. Priyanka, "A Memory Efficient Hardware Based Pattern
Matching And Protein Alignment Schemes For Highly Complex
Databases," International Journal on Smart Sensing and
Intelligent Systems, 2017. 10(4): p. 101-122.
[2]. I.Kotenko, A.M.S., "Attack Graph Based Evaluation of Network
Security," Communications and Multimedia Security, New
York, USA: Springer-Verlag, October 2006. 216-227.
[3]. V.D.S. Vijayarangam, "Detecting Ip Based Attack On Cloud
Server Using Passive Ip Traceback," International Journal on
Smart Sensing and Intelligent Systems, 2017. 10(4): p. 136-146.
[4]. X. Elvis, Kheng Cher Yeo, A. Sami, S. Bharanidharan,
"Performance analysis of various encryption techniques in
communication network," Asian Journal of Information
Technology, 2017. 16(1): p. 125130.
[5]. R. Hunt, A.S.Z., "Network Forensics: An Analysis of
Techniques, Tools, and Trends. Computer," December 2012.
45(12): p. 36-43.
[6]. M.P. Collins, "Graph-based analysis in network security," p.
1333-1337.
[7]. V.N.L. Franqueira, "Finding multi-step attacks in computer
networks using heuristic search and mobile ambients,"
University of Twente, 2009.
[8]. E. Cole, "Network security bible." John Wiley & Sons, 2011.
768.
[9]. C. Wang, N.D., and H. Yang, "Generation and Analysis of
Attack Graphs. Procedia Engineering," 2012. 29: p. 4053-4057.
[10]. C. Phillips, A.L.P.S., "A graph-based system for network-
vulnerability analysis," Digests in Proceedings of the 1998
workshop on New security paradigms, 1998: p. 71-79.
[11]. S. Yi, Y.P., Q. Xiong, T. Wang, Z. Dai, H. Gao, J. Xu, J. Wang
and L. Xu, "Overview on attack graph generation and
visualization technology," International Conference on Anti-
Counterfeiting, Security and Identification (ASID), Shanghai,
2013: p. 1-6.
[12]. S. Jajodia, S.N., and B. O’Berry, "Topological analysis of
network attack vulnerability," Managing Cyber Threats,
Springer, 2005: p. 247-266.
[13]. X. Ou, S.G., and A. W. Appel, "MulVAL: A Logic-based
Network Security Analyzer," USENIX security, 2005.
[14]. X. Ou, W.F.B. & M. A. McQueen, "A scalable approach to
attack graph generation," Digests 13th ACM conference on
Computer and communications security, 2006: p. 336-345.
[15]. M.T.A.A.N.Z. Heywood, "VEA-bility security metric: A
network security analysis tool," Digests Third International
Conference on Availability, Reliability and Security, 2008: p.
950-957.
[16]. R. Baloch, "Ethical Hacking and Penetration Testing Guide,"
CRC Press, 2014.
[17]. O.M. Sheyner, "Scenario graphs and attack graphs," Air Force
Research Laboratory, 2004.
[18]. L.-H. Hsu and C.-K. Lin, "Graph theory and interconnection
networks," 2008: CRC press.
[19]. S. Noel, A.S.J., "Managing attack graph complexity through
visual hierarchical aggregation," [Digests in Proceedings of the
2004 ACM workshop on Visualization and data mining for
computer security, October 2004: p. 109-118.
[20]. M. Frigault, L.W., A.Singhal, and S. Jajodia, "Measuring
network security using dynamic bayesian network," Digests 4th
ACM workshop on Quality of protection, October 2008: p. 23-
30.
[21]. I.
Singhal, A.X.O., "Techniques for enterprise network security metrics,"
Digests 5th Annual Workshop on Cyber Security and Information
Intelligence Research: Cyber Security and Information Intelligence
Challenges and Strategies, April 2009. ACM: p. 25.
[22]. P. Ammann, D.W., and S. Kaushik, "Scalable, graph-based
network vulnerability analysis," Digests 9th ACM Conference
on Computer and Communications Security, 2002: p. 217-224.
[23]. M. Alhomidi, A.M.R., "Attack Graph-Based Risk Assessment
And Optimisation Approach," School of Computer Science and
Electronic Engineering University of Essex, Colchester, UK
International Journal of Network Security & Its Applications
(IJNSA), 2014. 6(3).
[24]. A.VMota, S. Azam, B. Shanmugam, K. C. Yeo & K.
Kannoorpatti, “Comparative analysis of different techniques of
encryption for secured data transmission”, IEEE International
Conference on Power, Control, Signals and Instrumentation
Engineering, ICPCSI 2017.
[25]. Gao, N., He, Y. & Ling, B. Wuhan Univ. J. Nat. Sci. (2018) 23:
171. https://doi.org/10.1007/s11859-018-1307-0 “Exploring
Attack Graphs for Security Risk Assessment: A Probabilistic
Approach.