Current Studies On Intrusion Detection System, Genetic Algorithm And Fuzzy Logic

International Journal of Distributed and Parallel Systems (IJDPS) Vol.4, No.2, March 2013

CURRENT STUDIES ON INTRUSION
DETECTION SYSTEM, GENETIC
ALGORITHM AND FUZZY LOGIC
Mostaque Md. Morshedur Hassan

LCB College, Maligaon, Guwahati, Assam, India.

mostaq786@gmail.com

ABSTRACT
Nowadays Intrusion Detection System (IDS) which is increasingly a key element of system security is used
to identify the malicious activities in a computer system or network. There are different approaches being
employed in intrusion detection systems, but unluckily each of the technique so far is not entirely ideal. The
prediction process may produce false alarms in many anomaly based intrusion detection systems. With the
concept of fuzzy logic, the false alarm rate in establishing intrusive activities can be reduced. A set of
efficient fuzzy rules can be used to define the normal and abnormal behaviors in a computer network.
Therefore some strategy is needed for best promising security to monitor the anomalous behavior in
computer network. In this paper I present a few research papers regarding the foundations of intrusion
detection systems, the methodologies and good fuzzy classifiers using genetic algorithm which are the focus
of current development efforts and the solution of the problem of Intrusion Detection System to offer a real-
world view of intrusion detection. Ultimately, a discussion of the upcoming technologies and various
methodologies which promise to improve the capability of computer systems to detect intrusions is offered.

KEYWORDS

Intrusion Detection System (IDS), Anomaly based intrusion detection, Genetic algorithm, Fuzzy logic.

1. INTRODUCTION

Intrusion detection is the process of monitoring the events ([1], [2], [3]) occurring in a computer
system or network and analyzing them for signs of probable incidents, which are violations or
forthcoming threats of violation of computer security strategies, adequate used policies, or usual
security practices. Intrusive events to computer networks are expanding because of the liking of
adopting the internet and local area networks [4] and new automated hacking tools and strategy.
Computer systems are evolving to be more and more exposed to attack, due to its wide spread
network connectivity.

Currently, networked computer systems play an ever more major role in our society and its
economy. They have become the targets of a wide array of malicious threats that invariably turn
into real intrusions. This is the reason computer security has become a vital concern for network
practitioner. Too often, intrusions cause disaster inside LANs and the time and cost to renovate
the damage can grow to extreme proportions. Instead of using passive measures to repair and
patch security hole once they have been exploited, it is more efficient to take up a proactive
measure to intrusions.

DOI : 10.5121/ijdps.2013.4204 35


Intrusion Detection Systems (IDS) are primarily focused on identifying probable incidents,
monitoring information about them, tries to stop them, and reporting them to security
administrators [5] in real-time environment, and those that exercise audit data with some delay
(non-real-time). The latter approach would in turn delay the instance of detection. In addition,
organizations apply IDSs for other reasons, such as classifying problems with security policies,
documenting existing attacks, and preventing individuals from violating security policies. IDSs
have become a basic addition to the security infrastructure of almost every organization. A usual
Intrusion Detection System is demonstrated in Figure 1.

NOTE: The arrow lines symbolize the amount of information flowing from one component to another

Figure 1. Very Simple Intrusion Detection System

Intrusion Detection Systems are broadly classified into two types. They are host-based and
network-based intrusion detection systems. Host-based IDS employs audit logs and system calls
as its data source, whereas network-based IDS employs network traffic as its data source. A host-
based IDS consists of an agent on a host which identifies different intrusions by analyzing audit
logs, system calls, file system changes (binaries, password files, etc.), and other related host
activities. In network-based IDS, sensors are placed at strategic position within the network
system to capture all incoming traffic flows and analyze the contents of the individual packets for
intrusive activities such as denial of service attacks, buffer overflow attacks, etc. Each approach
has its own strengths and weaknesses. Some of the attacks can only be detected by host-based or
only by network-based IDS.

The two main techniques used by Intrusion Detection Systems for detecting attacks are Misuse
Detection and Anomaly Detection. In a misuse detection system, also known as signature based
detection system; well known attacks are represented by signatures. A signature is a pattern of
activity which corresponds to intrusion. The IDS identifies intrusions by looking for these
patterns in the data being analyzed. The accuracy of such a system depends on its signature
database. Misuse detection cannot detect novel attacks as well as slight variations of known
attacks.

An anomaly-based intrusion detection system inspects ongoing traffic, malicious activities,
communication, or behavior for irregularities on networks or systems that could specify an attack.
The main principle here is that the attack behavior differs enough from normal user behavior that

36


it cannot be detected by cataloging and identifying the differences involved. By creating supports
of standard behavior, anomaly-based IDS can view when current behaviors move away
statistically from the normal one. This capability gives the anomaly-based IDS ability to detect
new attacks for which the signatures have not been created. The main disadvantage of this
method is that there is no clear cut method for defining normal behavior. Therefore, such type of
IDS can report intrusion, even when the activity is legitimate.

One of the major problems encountered by IDS is large number of false positive alerts that is the
alerts that are mistakenly analyzed normal traffic as security violations. An ideal IDS does not
produce false or inappropriate alarms. In practice, signature based IDS found to produce more
false alarms than expected. This is due to the very general signatures and poor built in verification
tool to authenticate the success of the attack. The large amount of false positives in the alert logs
generates the course of taking corrective action for the true positives, i.e. delayed, successful
attacks, and labor intensive.

My goal is to detect novel attacks by unauthorized users in network traffic. I consider an attack to
be novel if the vulnerability is unknown to the target's owner or administrator, even if the attack
is generally known and patches and detection tests are available. I mostly like to cite four types of
remotely launched attacks: denial of service (DOS), U2R, R2L, and probes. A DoS attack is a
type of attack in which the unauthorized users build a computing or memory resources too busy
or too full to provide reasonable networking requests and hence denying users access to a
machine e.g. ping of death, neptune, back, smurf, apache, UDP storm, mail bomb etc. are all DoS
attacks. A remote to user (U2R) attack is an attack in which a user forwards networking packets
to a machine through the internet, which he/she does not have right of access in order to expose
the machines vulnerabilities and exploit privileges which a local user would have on the computer
e.g. guest, xlock, xnsnoop, sendmail dictionary, phf etc. A R2L attacks are regarded as the
exploitations in which the unauthorized users start off on the system with a normal user account
and tries to misuse vulnerabilities in the system in order to achieve super user access rights e.g.
xterm, perl. A probing is an attack in which the hacker scans a machine or a networking device in
order to determine weaknesses or vulnerabilities that may later be exploited so as to negotiate the
system. This practice is commonly used in data mining e.g. portsweep, saint, mscan, nmap etc.

The Intrusion Detection System (IDS) is also carried out by implementing Genetic Algorithm
(GA) to efficiently identify various types of network intrusions. The genetic algorithm [1] is
applied to achieve a set of classification rules from the support-confidence framework, and
network audit data is employed as fitness function to judge the quality of each rule. The created
rules are then used to classify or detect network intrusions in a real-time framework. Unlike most
available GA-based approaches remained in the system, because of the easy demonstration of
rules and the efficient fitness function, the proposed system is very simple to employ while
presenting the flexibility to either generally detect network intrusions or precisely classify the
types of attacks.

The normal and the abnormal intrusive activities in networked computers are tough to forecast as
the boundaries cannot be well explained. This prediction process may generate false alarms [1] in
many anomaly based intrusion detection systems. However, with the introduction of fuzzy logic,
the false alarm rate in determining intrusive activities can be minimized; a set of fuzzy rules (non-
crisp fuzzy classifiers) can be employed to identify the normal and abnormal behavior in
computer networks, and fuzzy inference logic can be applied over such rules to determine when
an intrusion is in progress. The main problem with this process is to make good fuzzy classifiers
to detect intrusions.

37


It has been shown by Baruah [6] that a fuzzy number [a, b, c] is concluded with reference to a
membership function μ(x) remaining within the range between 0 and 1, a ≤ x ≤ c. Further, he has
extended this definition in the following way. Let μ1(x) and μ2(x) be two functions, 0 ≤ μ2(x) ≤
μ1(x) ≤ 1. He has concluded μ1(x) the fuzzy membership function, and μ2(x) a reference
function, such that (μ1(x) – μ2(x)) is the fuzzy membership value for any x. Finally he has
characterized such a fuzzy number by {x, μ1(x), μ2(x); x ∈ Ω}.

The complement of µx is always counted from the ground level in Zadehian’s theory [9], whereas
it actually counted from the level if it is not as zero that is the surface value is not always zero. If
other than zero, the problem arises and then we have to count the membership value from the
surface for the complement of µx.

In Figure 2, Baruah [6] explained that for a fuzzy number A = [a, b, c], the value of membership
for any x ∈ Ω is specified by μ(x) for a ≤ x ≤ c, and is taken as zero otherwise. For the fuzzy
number AC, the value of membership for any x ∈ Ω is given by (1 - μ(x)) for a ≤ x ≤ c, otherwise,
the value is 1. The main difference is that for AC the membership function holds 1 all over the
place with the reference function being μ(x), whereas for A the membership function is μ(x) with
the reference function being 0 everywhere.

Complement of
1- µ(x) A

µ(x)
A

Figure 2. Extended definition of Fuzzy Set

This proposed system forwarded a definition of complement of an extended fuzzy set in which
the fuzzy reference function is not always taken as zero. The definition of complement of a fuzzy
set forwarded by Baruah ([6], [7]), Neog and Sut [8] could be viewed a particular case of what I
am proposing. I would use Baruah’s definition of the complement of a normal fuzzy set in my
work.

2. INTRUSION DETECTION STRATEGIES

The intrusion detection strategies concerns four primary issues. First is the dataset that is captured
from network communications. The second is Genetic Algorithms (GA) which use mutation,
recombination, and selection applied to a population of individuals in order to evolve iteratively
better and better solutions and a way to generate fuzzy rules to characterize normal and abnormal
behavior of network systems. The third is to generate alerts and reports for malicious traffic
behavior, and the fourth is the maintenance of the ids for observation of placement of sensors, and
qualified trained intrusion analysts so that the latest malicious traffic is being detected.

38


2.1. The Dataset

To implement the algorithm and to evaluate the performance of the system, I propose the standard
datasets employed in KDD Cup 1999 “Computer Network Intrusion Detection” competition.

The KDD 99 intrusion detection datasets depends on the 1998 DARPA proposal, which offers
designers of intrusion detection systems (IDS) with a standard on which to evaluate different
methodologies ([21], [24]). Hence, a simulation is being prepared from a fabricated military
network with three ‘target’ machines running various services and operating systems. They also
applied three extra machines to spoof different IP addresses for generating network traffic.

A connection is a series of TCP packets beginning and ending at some well defined periods,
between which data floods from a source IP address to a target IP address under some well
defined protocol ([21], [22], [24]). It results in 41 features for each connection.

Finally, there remains a sniffer that accounts all network traffic by means of the TCP dump
format [24]. The total simulated period is seven weeks. Normal connections are shaped to outline
that expected in a military network and attacks are categorized into one of four types: User to
Root; Remote to Local; Denial of Service; and Probe.

The KDD 99 intrusion detection benchmark consists of different components [23]:

kddcup.data; kddcup.data_10_percent; kddcup.newtestdata_10_percent_unlabeled;

kddcup.testdata.unlabeled; kddcup.testdata.unlabeled_10_percent; corrected.

I propose to use “kddcup.data_10_percent” as training dataset and “corrected” as testing dataset.
In this case the training set consists of 494,021 records among which 97,280 are normal
connection records, while the test set contains 311,029 records among which 60,593 are normal
connection records. Table 1 shows the intrusion types distribution in the training and the testing
datasets.
Table 1. Intrusion types distribution in datasets

Dataset normal prob Dos u2r r2l Total
Train 97280 e
4107 391458 52 1124 494021
(“kddcup.data_10_percent”)
Test (“corrected”) 60593 4166 229853 228 16189 311029

2.2. Genetic algorithm

2.2.1. Genetic algorithm overview

A Genetic Algorithm (GA) is a programming technique that uses biological evolution as a
problem solving strategy [20]. It is based on Darwinian’s theory of evolution and survival of
fittest to make effective a population of candidate result near a predefined fitness [13].

The proposed GA based intrusion detection system holds two modules where each acts in a
dissimilar stage. In the training stage, a set of classification rules are produced from network audit
data using the GA in an offline background. In the intrusion detection phase, the generated rules
are employed to classify incoming network connections in the real-time environment. Once the
rules are generated, the intrusion detection system becomes simple, experienced and efficient one.

39


GA applies an evolution and natural selection that employs a chromosome-like data structure and
evolve the chromosomes by means of selection, recombination and mutation operators [13]. The
process generally starts with randomly generated population of chromosomes, which signify all
possible solution of a problem that are measured candidate solutions. From each chromosome
different positions are set as bits, characters or numbers. These positions are regarded as genes.
An evaluation function is employed to find the decency of each chromosome according to the
required solution; this function is known as “Fitness Function”. During the process of evaluation
“Crossover” is applied to have natural reproduction and “Mutation” is applied to mutation of
species [13]. For survival and combination the selection of chromosomes is partial towards the
fittest chromosomes.

When I use GA for solving various problems three factors will have crucial impact on the use of
the algorithm and also of the applications [2]. The factors are : i) the fitness function, ii) the
representation of individuals, and iii) the genetic algorithm parameters. The determination of
these factors often depends on implementation of the system.

2.2.2 Fuzzy logic

Zadeh explained that Fuzzy logic [9] is an extension of Boolean logic that is often used for
computer-based complex decision making. While in classical Boolean logic an element can be
either a full member or non-member of a Boolean (sometimes called ”crisp”) set, the membership
of an element to a fuzzy set can be any value within the interval [0, 1], allowing also partial
membership of an element in a set.

A fuzzy expert system consists of three different types of entities: fuzzy sets, fuzzy variables and
fuzzy rules. The membership of a fuzzy variable in a fuzzy set is determined by a function that
produces values within the interval [0, 1]. These functions are called membership functions.
Fuzzy variables are divided into two groups: antecedent variables, that are assigned with the input
data of the fuzzy expert system and consequent variables, that are assigned with the results
computed by the system.

The fuzzy rules determine the link between the antecedent and the consequent fuzzy variables,
and are often defined using natural language linguistic terms. For instance, a fuzzy rule can be ”if
the temperature is cold and the wind is strong then wear warm clothes”, where temperature and
wind are antecedent fuzzy variables, wear is a consequent fuzzy variable and cold, strong and
warm clothes are fuzzy sets.

The process of a fuzzy system has three steps. These steps are Fuzzification, Rule Evaluation, and
Defuzzification. In the fuzzification step, the input crisp values are transformed into degrees of
membership in the fuzzy sets. The degree of membership of each crisp value in each fuzzy set is
determined by plugging the value into the membership function associated with the fuzzy set. In
the rule evaluation step, each fuzzy rule is assigned with a strength value. The strength is
determined by the degrees of memberships of the crisp input values in the fuzzy sets of
antecedent part of the fuzzy rule. The defuzzification stage transposes the fuzzy outputs into crisp
values.

It has been revealed by Baruah [6] that a fuzzy number [a, b, c] can be explained with reference
to a membership function μ(x) remaining between 0 and 1, a ≤ x ≤ c. Further, he has extended
this definition in the following way. Let μ1(x) and μ2(x) be two functions, 0 ≤ μ2(x) ≤ μ1(x) ≤ 1.
He has concluded μ1(x) the fuzzy membership function, and μ2(x) a reference function, such that
(μ1(x) – μ2(x)) is the fuzzy membership value for any x. Finally he has characterized such a
fuzzy number by {x, μ1(x), μ2(x); x ∈ Ω}.

40


The complement of µx is always counted from the ground level in Zadehian’s theory [9], whereas
it actually counted from the level if it is not as zero that is the surface value is not always zero. If
other than zero, the problem arises and then we have to count the membership value from the
surface for the complement of µx. Thus I could conclude the following statement –

Complement of µx = 1 for the entire level

Membership value for the complement of µx = 1- µx

I have forwarded Baruah’s definition of complement of an extended fuzzy set where the fuzzy
reference function is not always taken as zero. The definition of complement of a fuzzy set
recommend by Baruah ([6], [7]), Neog and Sut [8] could be considered a particular case of what I
am giving. I would use Baruah’s definition of the complement of a normal fuzzy set in my
proposed work.

In the two classes’ classification problem, two classes are available where every object should be
classified. These classes are called positive (abnormal) and negative (normal). The data set
employed by the learning algorithms holds a set of objects where each object contains n+1
attributes. The first n attributes identifies the monitored parameters of the object characteristics
and the last attribute identifies the class where the object belongs to the classification attribute.
A fuzzy classifier is a set of two rules for solving the two classes’ classification problem, one for
the normal class and other for the abnormal class, where the conditional part is described by
means of only the monitored parameters and the conclusion part is viewed as an atomic
expression for the classification attribute.

2.2.3 Fitness function

The authors in [1] used the fuzzy confusion matrix to calculate the fitness of a chromosome. In
the fuzzy confusion matrix the fuzzy truth degree of the condition represented by the
chromosome and the fuzzy negation operator are used directly. The fitness of a chromosome for
the abnormal class is evaluated according to the following set of equations:

p
TP = ∑ predicted (class_data i)
i=1

q
TN = ∑1 – predicted (other_class_datai)
i=1

q
FP = ∑ predicted (other_class_data i)
i=1

p
FN = ∑ 1– predicted (class_datai)
i=1

Where,

Sensitivity = TP/(TP+FN)
Specificity = FP/( FP+ TN)

41


Length = 1 – chromosome_length/10

So finally Fitness of a chromosome is calculated as follows –

Fitness = W1 * Sensitivity + W2 * Specificity + W3 * Length

Where,

TP, TN, FP, FN are true positive, true negative, false positive, false negative value for the rule, p
is the number of samples of the evolved class in the training data set, q is the number of samples
of the remaining class in the training data set, predicted is the fuzzy value of the conditional part
of the rule, class_datai is an element of the subset of the training samples of the evolved class,
other_class_datai is an element of the subset of the remaining classes in the training samples,
and W1, W2, W3 are the assigned weights for each rule characteristics respectively.

2.3 Generate alerts and reports

The reports portraits an entire image of the status of the network under observation. It handles all
the output from the system, whether that be an automated response to the suspicious activity, or
which is most common, the notification of some security officer. IDSs should provide facilities
for practitioners to fine-tune thresholds for generating alarms as well as facilities for suppressing
alarms selectively.

Reporting can demonstrate the economic value of the monitoring tools. It can also ease the
burden of monitoring. The IDS should generate reports that help practitioners investigate the
alarms. In addition, the intrusion detection system can assist network practitioners prioritize their
tasks, by assigning priorities to alarms, or providing each alarm to a practitioner for further
investigation.

2.4 IDS Maintenance

2.4.1 Maintenance

IDS maintenance is essential for all IDS technologies because all sorts of threats and prevention
technologies are constantly varying, patches, signatures, and configurations must be modernized
to ensure that the latest malicious traffic is being detected and prevented. We could maintain IDS
from a console using a graphical user interface (GUI), application, or secure web-based interface.
Network administrators could monitor IDS components from the console to make sure they are
operational, validate they are working properly, and carry out vulnerability assessments (VA) and
updates.

2.4.2 Tuning

To be effective in detection policy, IDS must be tuned precisely. Tuning requires varying
different settings to be in conformity with the security guiding principles and objective of the IDS
administrator. Scanning techniques, thresholds, and focus can be regulated to make certain that
anIDS is making out relevant data without overloading the network administrator with warnings
or too many false positives. Tuning is time-consuming, but it must be performed to make sure an
efficient IDS configuration. It is to be noted that tuning must be specific to the IDS product only.

42


2.4.3 Detection Accuracy

The accuracy of intrusion detection system relies on the technique in which it identifies, such as
by the rule set. Signature-based detection detects only simple and recognized attacks, while
anomaly-based detection can detect more types of attacks, but has a higher number of false
positives ratios. Tuning is essential to reduce the number of false positives and to make the data
further functional.

3. CHALLENGES IN IDS

There are number of challenges that impact on organization’s decision to use IDS. In this section
I have described a few challenges that the organizations encounter while installing an intrusion
detection system. These are discussed below –

1. Human intervention - IDS technology itself is experiencing a lot of enhancements. It is
therefore very important for organizations to clearly define their prospect from the IDS
implementation. Till now IDS technology has not achieved a level where it does not
require human interference. Of course today's IDS technology recommends some
automation like reporting the administrator in case of detection of a malicious activity,
avoiding the malicious connection for a configurable period of time, dynamically
changing a router's access control list in order to prevent a malicious connection etc.
Therefore the security administrator must investigate the attack once it is detected and
reported, determine how it occurred, correct the problem and take necessary action to
prevent the occurrence of the same attack in future.

2. Historical analysis - It is still very important factor to monitor the IDS logs regularly to
continue on top of the incidence of events. Monitoring the logs on a daily basis is
necessary to analyze the different type of malicious activities detected by the IDS over a
period of time. Today's IDS has not yet achieved the level where it can provide historical
analysis of the intrusive activities detected over a span of time. This is still a manual
activity.

Hence it is vital for an organization to have a distinct incident handling and response plan
if an intrusion is detected and reported by the IDS. Also, the organization should have
expert security personnel to handle this kind of situation.

3. Deployment - The success of an IDS implementation depends to a large degree on how it
has been deployed. A lot of plan is necessary in the design as well as the implementation
phase. In most cases, it is required to apply a fusion solution of network based and host
based IDS to gain from both cases. In fact one technology complements the other.

However, this decision can differ from one organization to another. A network based IDS
is an instant choice for many organizations because of its capability to monitor multiple
systems and also the truth that it does not need a software to be loaded on a production
system different from host based IDS.

Some organizations implement a hybrid solution. Organizations installing host based IDS
solution needs remember that the host based IDS software is processor and memory
challenging. So it is very important to have sufficient available resources on a system
before establishing a host based sensor on it.

43


4. Sensors - It is important to maintain sensor to manager ratio. There is no strict rule as
such for calculating this ratio. To a large degree it depends upon how many different
types of traffic is monitored by each sensor and in which background. Most of the
organizations deploy a ratio of 10:1, while some organizations maintain 20:1 and some
others go for 15:1.

It is very important to plan the baseline strategy before starting the IDS implementation
and avoid false positives. A poorly configured IDS sensor may post a lot of false positive
ratios to the console and even a ratio of 10:1 or even enough better sensors to the console
ratio can be missing.

5. False positive and negative alarms rate – It is impossible for IDS to be ideal mostly
because network traffic is so complicated. The erroneous results in IDS are divided into
two types: false positives and false negatives. False positives take place when the IDS
erroneously identify a problem with benign traffic. False negatives occur when redundant
traffic is overlooked by the IDS. Both create problems for security administrators or
practitioners and demands that the malicious threats must be detected powerfully. A
greater number of false positives are generally more acceptable but can burden a security
administrator with bulky amounts of data to filter through. However, because it is
unnoticed, false negatives do not provide a security administrator a chance to check the
data. Therefore IDS to be implemented should minimize both false positive and negative
alarms.

6. Signature database - A common policy for IDS in detecting intrusions is to remember
signatures of known attacks. The inherent weak points in relying on signatures are that
the signature patterns must be acknowledged first. New threats are often unrecognizable
by eminent and popular IDS. Signatures can be masked as well. The ongoing event
between new attacks and detection systems has been a challenge. Therefore the signature
database must be updated whenever a different kind of attack is detected and repair for
the same is available.

7. Monitor traffic in large networks - Network Intrusion Detection System (NIDS)
components are spotted throughout a network, but if not placed tactically, many attacks
can altogether avoid NIDS sensors by passing through alternate ways in a network.
Moreover, though many IDS products available in the market are efficient to distinguish
different types of attacks, they may fail to recognize attacks that use many attack sources.
Many IDS cannot cleverly correlate data from numerous sources. Newer IDS
technologies must influence integrated systems to increase an overview of distributed
intrusive activity. Therefore IDS must be able to successfully monitor traffic in a large
network.

4. PRIOR WORKS

In this section, I describe the important and relevant research works of different authors that I
have come across during the literature survey of my proposed work. I illustrate each attack
manner and point to the impact of this attack and its intrusive activities. From an intruder’s point
of view, I analyze each of the attack’s modes, intention, benefits and suitable conditions and try
to find out the solution how to improve the attack by introducing the concept of fuzzy logic-based
technique and genetic algorithm.

44


The normal and abnormal behaviors [1] in networked computers are hard to forecast, as the limits
cannot be explained clearly. This prediction method usually generates fake alarms in many
anomaly based intrusion detection systems.

In [1] the authors introduced the concept of fuzzy logic to reduce the fake alarm rate in
determining intrusive behavior. The set of fuzzy rules is applied to identify the normal and
abnormal behavior in a computer network. The authors proposed a technique to generate fuzzy
rules that are able to detect malicious activities and some specific intrusions. This system
presented a novel approach for the presentation of generated fuzzy rules in classifying different
types of intrusions.

The advantage of their proposed mechanism is that the fuzzy rules are able to detect the malicious
activities.

But they failed to implement the real time network traffic, more attributes for the classification
rules. In determining the fuzzy rules, they used the concept of fuzzy membership function and
reference function, but they said that the membership function and reference function are same. In
reality, these two concepts are totally different concepts. I have forwarded the extended definition
of fuzzy set of Baruah ([6], [7]), Neog and Sut [8].

In [9] Zadeh initiated the idea of fuzzy set theory and it was mainly intended mathematically to
signify uncertainty and vagueness with formalized logical tools for dealing with the vagueness
connected in many real world problems. The membership value to a fuzzy set of an element
describes a function called membership function where the universe of discourse is the domain
and the interval lies in the range [0, 1]. The value 0 means that the element is not a member of the
fuzzy set; the value 1 means that the element is fully a member of the fuzzy set. The values that
remain between 0 and 1 distinguish fuzzy members, which confined to the fuzzy set merely
partially.

But the author gave an explanation that the fuzzy membership value and fuzzy membership
function for the complement of a fuzzy set are same concepts and the surface value is always
counted from the ground level.

Baruah ([6], [7]), Neog and Sut [8] have forwarded an extended definition of fuzzy set which
enables us to define the complement of a fuzzy set. My proposed system agrees with them as this
new definition satisfies all the properties regarding the complement of a fuzzy set.

In [2] Gong, Zulkernine, Abolmaesumi gave an implementation of genetic based approach to
Network Intrusion Detection using genetic algorithm and showed software implementation to
detect the malicious activities. The approach derived a set of classification rules from network
audit data and utilizes a support-confidence framework to judge the quality of each rule. The
generated rules are then used in intrusion detection system to detect and to classify network
intrusions efficiently in a real-time environment.

But, some limitations of their implemented method are observed. First, the generated rules were
partial to the training dataset. Second, though the support-confidence framework is simple to
implement and provides improved accuracy to final rules, it requires the whole training datasets
to be loaded into memory before any computation. For large training datasets, it is neither
efficient nor feasible.

In [12] Hoque, Mukit and Bikas presented an implementation of Intrusion Detection System by
applying the theory of genetic algorithm to efficiently detect various types of network intrusive

45


activities. To apply and measure the efficiency of their system they used the standard KDD 99
intrusion detection benchmark dataset and obtained realistic detection rate. To measure the fitness
of a chromosome they used the standard deviation equation with distance.

But their performance of detection rate was poor and they failed to reduce the false positive rate
in Intrusion Detection System.

5. CONCLUSION

In this paper, I have described an overview of some of the current and past intrusion detection
technologies which are being utilized for the detection of intrusive activities against computer
systems or networks. The different detection challenges that affect the decision policy of the IDS
employed in an organization are clearly outlined. I propose to use the new definition of the
complement of fuzzy sets where the fuzzy membership value and fuzzy membership function for
the complement of a fuzzy set are two different concepts because the surface value is not always
counted from the ground level. This new definition of fuzzy sets can classify efficient rule sets.
This would help in reducing the false alarm rate occurred in intrusion detection system.

ACKNOWLEDGEMENTS

I would like to extend my sincere thanks and gratefulness to H.K. Baruah, Professor, Department
of Statistics, Gauhati University, Guwahati, India for his kind help, moral support and guidance in
preparing this article.

REFERENCES

[1] J. Gomez & D. Dasgupta, (2002) “Evolving Fuzzy Classifiers for Intrusion Detection”, IEEE
Proceedings of the IEEE Workshop on Information Assurance, United States Military Academy,
West Point, NY.
[2] R. H. Gong, M. Zulkernine & P. Abolmaesumi, (2005) “A Software Implementation of a Genetic
Algorithm Based Approach to Network Intrusion Detection”, Proceedings of the Sixth International
Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed
Computing and First ACIS International Workshop on Self-Assembling Wireless Networks.
[3] T. Xia, G. Qu, S. Hariri & M. Yousif, (2005) “An Efficient Network Intrusion Detection Method
Based on Information Theory and Genetic Algorithm”, Proceedings of the 24th IEEE International
Performance Computing and Communications Conference, Phoenix, AZ, USA.
[4] Yao, J. T., S.L. Zhao & L.V. Saxton, (2005) “A Study On Fuzzy Intrusion Detection”, In Proceedings
of the Data Mining, Intrusion Detection, Information Assurance, And Data Networks Security, SPIE,
Vol. 5812, Orlando, Florida, USA , pp. 23-30.
[5] B. Abdullah, I. Abd-alghafar, Gouda I. Salama & A. Abd-alhafez, (2009) “Performance Evaluation of
a Genetic Algorithm Based Approach to Network Intrusion Detection System”, 13th International
Conference on Aerospace Sciences and Aviation Technology (ASAT), May 26-28.
[6] Hemanta K. Baruah, (2011) “Towards Forming A Field Of Fuzzy Sets”, International Journal of
Energy, Information and Communications (IJEIC), Vol. 2, Issue 1, February, pp. 16-20.
[7] Hemanta K. Baruah, (2011) “The Theory of Fuzzy Sets: Beliefs and Realities”, International Journal
of Energy, Information and Communications (IJEIC), Vol. 2, Issue 2, pp. 1-22.
[8] Tridiv Jyoti Neog & Dushmanta Kumar Sut, (2011) “Complement of an Extended Fuzzy Set”,
International Journal of Computer Applications (IJCA), Vol. 29-No.3, September, pp. 39-45.
[9] Zadeh L A, (1965) “Fuzzy Sets”, Information and Control, Vol.8, pp. 338-353.
[10] Y. Dhanalakshmi & Dr. I. Ramesh Babu, (2008) “Intrusion Detection Using Data Mining Along
Fuzzy Logic and Genetic Algorithms”, International Journal of Computer Science and Network
Security (IJCSNS), Vol.8, No.2, February, pp. 27-32.

46


[11] W. Lu & I. Traore, (2004) “Detecting New Forms of Network Intrusion Using Genetic
Programming”, Computational Intelligence, vol. 20, pp. 3, Blackwell Publishing, Malden, pp. 475-
494.
[12] Mohammad Sazzadul Hoque, Md. Abdul Mukit & Md. Abu Naser Bikas, (2012) “An Implementation
of Intrusion Detection System using Genetic Algorithm”, International Journal of Network Security
and Its Applications (IJNSA),Vol.4, No.2, March, pp. 109-120.
[13] W. Li, (2004) “A Genetic Algorithm Approach to Network Intrusion Detection”, SANS Institute,
USA.
[14] A. Sung & S. Mukkamala, (2003) “Identifying important features for intrusion detection using
support vector machines and neural networks”, in Symposium on Applications and the Internet, pp.
209– 216.
[15 J. P. Planquart, “Application of Neural Networks to Intrusion Detection”, SANS Institute Reading
Room.
[16] R. G. Bace, (2000) “Intrusion Detection”, Macmillan Technical Publishing.
[17] S. Kumar & E. Spafford, (1995) “A Software architecture to Support Misuse Intrusion Detection”, in
the 18th National Information Security Conference, pp. 194-204.
[18] K. Ilgun, R. Kemmerer & P. A. Porras, (1995) “State Transition Analysis: A Rule-Based Intrusion
Detection Approach”, IEEE Transaction on Software Engineering, pp. 181-199.
[19] S. Kumar, (1995) “Classification and Detection of Computer Intrusions”, Purdue University.
[20] V. Bobor, (2006) “Efficient Intrusion Detection System Architecture Based on Neural Networks and
Genetic Algorithms”, Department of Computer and Systems Sciences, Stockholm University/Royal
Institute of Technology, KTH/DSV.
[21] KDD-CUP (99) Task Description; http://kdd.ics.uci.edu/databases/kddcup99/task.html
[22] KDD Cup (1999): Tasks; http://www.kdd.org/kddcup/index.php?section=1999&method=task
[23] KDD Cup (1999): Data; http://www.kdd.org/kddcup/index.php?section=1999&method=data
[24] H. G. Kayacık, A. N. Zincir-Heywood & M. I. Heywood, (2005) “Selecting Features for Intrusion
Detection: A Feature Relevance Analysis on KDD 99 Intrusion Detection Datasets”.
[25] G. Folino, C. Pizzuti & G. Spezzano, (2005) “GP Ensemble for Distributed Intrusion Detection
Systems”, ICAPR, pp. 54-62.

Authors

Mostaque Md. Morshedur Hassan

Mostaque Md. Morshedur Hassan is a senior Assistant Professor of Computer Science
in the department of Computer Science and Information Technology at Lalit Chandra
Bharali College (LCBC), Maligaon, Guwahati, Assam, India. He holds his Master of
Computer Application (MCA) degree from Allahabad Agricultural Institute Deemed
University, Allahabad. His area of interests includes Network Security, Intrusion
Detection and Prevention, Wireless Security, Web Security, Fuzzy Logic, and Social
Networking Site.

47

Current Studies On Intrusion Detection System, Genetic Algorithm And Fuzzy Logic

More Related Content

What's hot (16)

Similar to Current Studies On Intrusion Detection System, Genetic Algorithm And Fuzzy Logic (20)

Recently uploaded (20)

Current Studies On Intrusion Detection System, Genetic Algorithm And Fuzzy Logic