Survey of network anomaly detection using markov chain

International Journal of Computer Science, Engineering and Information Technology (IJCSEIT), Vol. 4,No. 1, February 2014
DOI : 10.5121/ijcseit.2014.4105 49
SURVEY OF NETWORK ANOMALY DETECTION
USING MARKOV CHAIN
Brindasri S1
and Saravanan k2
1
M.E(CSE), Department of Computer Engineering, Erode Sengunthar Engineering
college, Anna University(Chennai)
2
Assistant Professor, Department of Computer Engineering, Erode Sengunthar
Engineering college, Anna University(Chennai)
ABSTRACT
Recently an internet threat has been increased. Our motive is detect the intrusion in the network in concise.
The real time issue such as DoS attack in banking, companies, industries and organization have been
increased significantly IDS has been used in both server and host side. The major challenge is to effectively
predict the periods of threats and protect the server from the unauthorized user. In this study, a novel
probabilistic approach is proposed effectively to detect the network intrusions. It uses a Markov chain for
probabilistic modelling of abnormal events in network systems. The degree of abnormality of the incoming
data is performed on the basis of the network states.
KEYWORDS
Anomaly detection, K-Mean Algorithm, Data Clustering, ID3 Decision tree, Markovian chain.
1. INTRODUCTION
[1]Intrusion detection is referred as scanning, it used to scan and detect any intrusion occur in
the system. It is a technology developed to assess the security of a computer system. Intrusions is
the activities that violate the security policy in the system. The main security goals are
confidentiality, integrity, availability, Possession of a computer or network[2]. IDS purpose is
used to identify and report unauthorized user or unapproved network activities in the system
There are 2approches are Anomaly and misuse detection. [2] Anomaly it attacks from outside the
organization and misuse it attacks from within the organization. SBIDS also known as misuse
detection that is signatures of known attacks is already stored and according to that it match the
attacks which is already present in the stored data. It will signal an intrusion if a match is found.
[3] The main drawback in this detection is that it cannot detect any new attacks whose signatures
are unknown. This means that an IDS using misuse detection will only detect known attacks or
attacks that are similar enough to a known attack to match its signature.[3] ABIDS is the Novelty
detection is the identification of new or unknown attack in the testing process.
The current process in intrusion detection is to combine both host based and network based
information to develop hybrid systems. Intrusion Detection System it combine both HIDS and
NIDS to identify the attacks effectively. So In hybrid system both kinds of IDS is used to detect
the attack simultaneously in the network.

50
Host based Intrusion Detection System (HIDS) is used to analyse the intrusions by analyzing
system calls, application logs, and data modifications. It is used on personal computer to provide
secure to the system host based Intrusion detection system used.
Network Intrusion Detection System (NIDS) are placed at strategic points within a network to
monitor traffic from all devices on the network.
2. RELATED WORK
2.1 TAXONOMY OF ANOMALY BASED INTRUSION DETECTION SYSTEM
IDS aim at detecting attacks against information system. It is difficult to provide a secure
information system and to prevent it for lifetime[2]. Maintenance of such system is difficult and
costly. IDS have been used for this purpose. Two types of IDS are there. They are Signature
Based IDS(SBIDS) and Anomaly Based IDS(ABIDS). ABIDS have been used to detect intrusion
in a network efficiently.
2.2 A CLUSTERING-BASED METHOD FOR UNSUPERVISED INTRUSION DETECTIONS
Detection of intrusion attacks is an important issue in network security. A novel method [4] is
proposed to compute the cluster radius threshold. The data classification is performed by an
improved nearest neighbor (INN) method. A powerful clustering-based method is presented for
the unsupervised intrusion detection (CBUID). The clustering based methods for the intrusion
detection and prevention in the network by classify the states Low accuracy detection compare
with existing system.
The parameter such as detection rate (DR), false alarm rate (FR) and detection rate for unknown
attack types to measure performance of the intrusion detection methods. False Alarm Rate
defined as the number of 'normal' patterns classified as attacks (False Positive) divided by
the total number of 'normal' patterns. The detection rate is the ratio of the detected attack
records based on the attacks in the original record.
The outlier factor of cluster is performed on the basis of the local deviation of a given data point.
It used to compute the cluster radius threshold. The data classification has been performed by an
improved nearest neighbour method. A powerful clustering based method has been used for the
unsupervised intrusion detection.
2.3 AN EFFICIENT INTRUSIONS DETECTION BASED ON DECISION TREE CLASSIFIER
USING FEATURE REDUCTION
Large computational value has always been a restraint in processing huge network intrusion data
[5]. Dataset was available (KDD data set) for the purpose of intrusion detection. It contains 42
attributes and the classes are categorized into five main classes (one normal class and four main
intrusion classes. Here the performance is based on correctly classified instances, incorrectly
classified instances, kappa statistic, mean absolute error, root mean squared error, relative
absolute error, root relative squared error and time .the comparison performed for 41 and 11
attributes. The four classifier models on the dataset were built and tested by means of 10-fold
cross-validation, due to the method of information gain feature reduction it enable better accuracy
of 99% and true positive rate is high with a value of one and false positive rate is low value is
zero.j48 is better than other three classifier.

51
2.4 INTRUSION DETECTION SYSTEMS USING DECISION TREES AND SUPPORT VECTOR
MACHINES
Investigation and evaluation of intrusion detection mechanism by the tree data mining techniques
[6] with the Support Vector Machines (SVM) was found. The Intrusion detection with Decision
trees and SVM were tested with benchmark 1998 DARPA Intrusion Detection data set. The work
enable the accuracy of the attack detection is low.
The Classification algorithm is inductively learned to construct a model from the pre-classified
data set. Each data item is defined by values of the attributes. Classification may be viewed as
mapping from a set of attributes to a particular class. The Decision tree classifies the given data
item using the values of its attributes. The decision tree is initially constructed from a set of pre-
classified data. The main approach is to select the attributes, which best divides the data items
into their classes. The main problem here is deciding the attribute, which will best partition the
data into various classes. The ID3 algorithm uses the information theoretic approach to solve this
problem. Information theory uses the concept of entropy, which measures the impurity of a data
items.
SVMs are able to handle only binary class classification problems. They divided the data into the
two classes of “Normal” and “Attack” patterns, where the Attack is collection of four classes of
attacks (Probe, DOS, U2R, and R2L). The classifier is learned from the training data and it is
used on the test data to classify the data into normal or attack patterns. This process is repeated
for all classes. The training time and testing times are also less for decision tree compared to the
SVM.
2.5 AN IMPLEMENTATION OF INTRUSION DETECTION SYSTEM USING GENETIC
ALGORITHM
Intrusion Detection Systems has become a powerful component in day today life in network
security. Intrusion Detection System (IDS), by applying genetic algorithm (GA)[7] to efficiently
detect various types of network intrusions. GA uses evolution theory to information evolution in
order to filter the traffic data and thus reduce the complexity. KDD benchmark dataset is used to
measure the performance of the system and obtained reasonable detection rate.
Here both Genetic Algorithms (GAs) and Genetic Programming (GP) is used for detecting
intrusion detection in the system. Some uses GA which is derived from the classification rules.
GA is used to select required features and to determine the optimal and minimal parameters
perform the operation in which different AI methods were used to derive acquisition of rules.
Problems are fidelity problem, resource usage problem, reliability problem. GA is used for
solving various problems such as it include three impact factors that vitally impact on the
effectiveness of the algorithm and application process. They are: i) the fitness function ii) the
representation of individuals iii) the GA parameters- Crossover and mutation occurs in rest of the
population which becomes the population of new generation. The process runs until the
generation size comes down to 1 (one). And it obtain 0.9500 as the detection rate.
2.6 EVALUATION OF FUZZY K-MEANS AND K-MEANS CLUSTERING ALGORITHMS
Clustering used to detect possible attacks in one of the branches of unsupervised learning[8].
Fuzzy sets reduces spurious alarms and intrusion detection. It uses KDDCUP99 data that are

52
collected based on DARPA innovation in 1998 for IDS designers that are used in several
investigation to find attacks and intrusion.
These data are simulated in seven weeks for intrusion detection, KDDCUP99 data have 41
properties which are divided in to four parts Fundamental Properties, Content, Traffic property
based on time, Traffic property based on the host: K-means algorithm acts better than fuzzy k-
means in 66% of DOS attacks.
By comparing the average distance between the sum of data points belong to one cluster from
center of that cluster ,Fuzzy K-Means and K-Means algorithm, K-Means clustering algorithm
give better result but when choosing different mode of study Fuzzy K-Means gives better result.
2.7 A MARKOV CHAIN MODEL OF TEMPORAL BEHAVIOR FOR ANOMALY DETECTION
A Markov chain model is a discrete time series stochastic process specifies the random variable
changes at discrete points in time [11]. Let y(t) be a random variable representing the state of a
system at time t, where t=0,1,2.. This process satisfying the following two assumptions:
• The probability distribution of the previous state at time t+1 depends on the current state
t, and it include dependent of time.
• A state transition from t to t+1 is independent of time
The Markov chain model is learned from historic data of the system’s normal behaviour. The
behaviour is analyzed to infer the probability of the process. A low probability indicates the
abnormal activities and it clearly separates normal from intrusive activities. Capturing activities in
network by network traffic data and audit data. Here it uses audit data from a UNIX Based Host
Machine. The audit records contains a variety of information including the event type, user ID, IP
address, and attacks. Here it extracted the event type characteristics of an audit event. The
learning and inference algorithm of the Markov model for intrusion detection were implemented
using c++. By using any threshold value it detects the attacks , it distinguishes normal activities
from the attack activities with the 0% false alarm rate and the 100% detection rate.
2.9 COMPARISON TABLE FOR VARIOUS ALGORITHM AND TECHNIQUE
Table 1 shows comparison of various algorithm and technique used to find out intrusion
detection.

53
Table 1 Comparison for various algorithm and technique
3. K-MEANS CLUSTERING ALGORITHM
K-Means is an unsupervised algorithm which is used for clustering. K-Means follows simple and
easy way to classify given dataset. Here k centers take one for each cluster. These centers should
be placed in a cunning way because of different location causes different result. So, the better
choice is to place them as much as possible far away from each other. The next step is to take
each point belonging to a given data set and associate it to the nearest center. When no point is
pending, the first step is completed and an early group age is done. At this point it needs to re-
calculate k new centroids as barycenter of the clusters resulting from the previous step. After
these k new centroids, a new binding has to be done between the same data set points and the
nearest new center. A loop has been generated. As a result of this loop it produces the k centers.
The location of the k center is based on the changing of location.
3.1 K-MEANS ALGORITHM
The k-means algorithm is used for partitioning; each cluster is represented on the basis of mean
value.
Method
Step 1:Choose k objects from D as the initial cluster centers
Step 2:Repeat Step 3
Step 3: Assign each object to the cluster to which the object is the most similar, based on the
mean value of the objects in the cluster.
Algorithm
(or)
Techniques
Advantage Disadvantage
K-Means Simplest algorithm to solve clustering
problem
It generates high false
positive and negative
Fuzzy K-Means Easy to find out the type of DoS attack
which have occur
It creates bottle neck
problem
Data Mining It helps to prevent hacked IP address
It identify false alarm generators
It creates bottle neck
problem
HOP-COUNT It used to detect whether hacking
attempt is made
It does not identify type
of attack
Genetic
Algorithm
(GA )
It used to filter the data
Reduced the complexity
Measure the fitness of the process
False positive rate is low
It need to improve more
statistical analysis
Complex equation are
needed

54
Step 4: Update the cluster means, i.e., calculate the mean value of the objects for each cluster.
Step 5: Until no change
Fig. 1 Clustering a set of objects based on the k-means method
4. MARKOV CHAIN MODEL
A Markov chain is a continous process that predicts a change in the future using analysis on
transitional characteristics from one state to another state. It describes a system with state
transition probability and provides a powerful method for analyzing the operation of a system
composed of finite states.
A Markov chain is a sequence of random variables X1, X2, X3, ... with the Markov property,
namely that, given the present state, the future and past states are independent. Formally,
The possible values of Xi form a countable S called the state space of the chain. Markov chains
are often described by a directed graph, where the edges are labeled by the probabilities of going
from one state to the other states.
Fig. 2 Flow diagram of markov chain model
5. CONCLUSION
The existing system is implemented using k-means algorithm with Markov chain model. The
system detects the attacks such as NIDS and HIDS. The rate of detection and rate of prevention
measure is low. In future to improve the detection performance the work might be implemented
by Eigen value and threshold values to detect the intrusion in network.
REFERENCES
[1] Poonam Dabas, Rashmi Chaudhary(2013), ”Survey Of Network Intrusion Detection Using K-Mean
Algorithm”, International Journal Of Advanced Research In Computer Science And Software
Engineering, Volume:3, Issue: 3,ISSN: 2277,pp. 30-35.

55
[2] Manasi Gyanchandani, Rana.J.L, Yadav.R.N(2012), ”Taxonomy of Anomaly Based Intrusion
Detection System: A Review”, International Journal of Scientific and Research Publications,
Volume:2, Issue:12,1 ISSN 2250-3153.
[3] Jyothsna.V, Rama Prasad.V.V, Munivara Prasad.K(2011), ”A Review of Anomaly based Intrusion
Detection Systems”, International Journal of Computer Applications (0975 – 8887)Volume:28 No:7,
pp: 283–304.
[4] Jiang, S., Song, X., Wang, H., Han, J., & Li, Q. (2006), ” A clustering-based method for unsupervised
intrusion detections”, Pattern Recognitions Letter, 27(7), pp.802–810.
[5] Yogendra Kumar Jain , Upendra, ” An Efficient Intrusions Detection Based On Decision Tree
Classifier Using Feature Reductions”, International Journal Of Scientific And Research
Publications,Volume 2,Issue 1, January 2012 ISSN 2250-3153.
[6] Peddabachigari.S, Abraham.A & Thomas(2004), ”Intrusion detection systems using decision trees
and support vector machines”, International Journal of Applied Science and Computations, pp. 118-
134.
[7] 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
& Its Applications (IJNSA), Volume:4, pp.109-120.
[8] Farhad Soleimanian Gharehchopogh, Neda Jabbari, Zeinab Ghaffari Azar(2012), ”Evaluation of
Fuzzy K-Means And K-Means Clustering Algorithms In Intrusion Detection Systems”, International
Journal Of Scientific & Technology Research Volume:1, Issue 11, ISSN 2277-8616 66, pp. 283–304.
[9] Doob, John Wiley & Sons, Dreger, Kreibich.C, Paxson.V, & Sommer.R (2005), ”Enhancing the
accuracy of network-based intrusion detection with host-based context”, Lecture Notes in Computer
Science, 3548,pp. 206–221.
[10] Mukkamala.S, Janoski.G & Sung.A.H(2002), ” Intrusion detection using neural networks and support
vector machines”, In Proceedings of IEEE International Joint Conference On Neural Networks , pp.
1702–1707.
[11] Nong Ye(2000), ”A Markov Chain Model of Temporal Behavior for Anomaly Detection”,
Proceedings of the 2000 IEEE Workshop on Information Assurance and Security United States
Military Academy, West Point,pp. 171-174.

Survey of network anomaly detection using markov chain

More Related Content

What's hot (12)

Viewers also liked (9)

Similar to Survey of network anomaly detection using markov chain (20)

Recently uploaded (20)

Survey of network anomaly detection using markov chain