Nandita Sengupta

Designing of Cryptography Based Security System for Cloud Computing

In todays world, demand of cloud computing is increasing day by day because of its many advantages, like sharing hardware, software and absence of fear of losing data. As number of users is increasing for cloud computing, threat for protecting confidential data in cloud is also increasing. This has led the computer scientists and researchers to think for finding robust security system for cloud computing. Detection, correction and prevention provide the complete solution for the security of cloud computing. Cryptography is one of the solutions which can prevent the intruders to enter into clouds for hacking information. In our paper, efficient hybrid cryptography system, Hybrid Vigenere Caesar Cipher Encryption (HVCCE) is proposed which will prevent the cloud infrastructure in three main places, in client location, in the network and in server. This cryptographic security system is designed in such a way so that computation time for decryption of cypher text messages for the hackers will be more compared to any single cryptographic system.

Evaluation of Rough Set Theory Based Network Traffic Data Classifier Using Different Discretization Method

In information security, intrusion detection is a challenging task for which designing of an efficient classifier is most important. In the paper, network traffic data is classified using rough set theory where discretization of data is a necessary preprocessing step. Different discretization methods are available and selection of one has great impact on classification accuracy, time complexity and system adaptability. Three discretization methods are applied on continuous KDD network data namely, rough set exploration system (RSES), supervised and unsupervised discretization methods to evaluate the classifier accuracy. It has been observed that supervised discretization yields best accuracy for rough set classification and provides system adaptability.

Designing encryption and IDS for cloud security

Usage of data is increasing exponentially with advancement of network and internet technologies. Cloud is becoming popular to save such enormous data along with facilities of using infrastructure, software and platform. Though cloud has many advantages but security is the biggest challenging factor in cloud. Cloud security includes the security in cloud client side, server side and when data is in transmission. Data is most vulnerable when data is in server as well as when data is in transmission medium. In this paper, two tier architecture of security system is proposed for cloud data. In the first tier, encryption is provided to secure data when the same is in medium. Encrypted data will be stored in server. In the second tier, Support Vector Machine based intrusion detection system is designed for cloud server to detect intruder.

Dynamic network traffic data classification for intrusion detection using genetic algorithm

Intrusion Detection System (IDS) classifies network traffic data either (anomaly( or (normal( to protect computer systems from different types of attacks. In this paper, data mining concepts and genetic algorithm have been applied to classify online traffic data efficiently by developing a rule based lazy classifier. The proposed method updates the rule set dynamically to accommodate the changing pattern in the traffic data in order to attain highest classification accuracy and at the same time maintaining consistency. The classifier is able to detect variants of common network traffic data patterns or modified existing security attacks based on the knowledge gained from its existing training data set with significant classification accuracy.

Comparison of Different Rule Calculation Method for Rough Set Theory

In the paper, information table for network traffic data is discretized by supervised learning using WEKA and from this discretized data, three different algorithms for calculation of rules are applied, genetic algorithm, covering algorithm and LEM2 algorithm. Rough set classification through three different table rule set are done and accuracies are observed. It has been concluded that out of these three methods of rule calculation, classification through table rule set using covering algorithm yields best accuracy. In fourth table of classification, rules have been calculated through reduct generation but in that case, rough set classification produces same total accuracy of classification through covering algorithm. RSES software is used for rule set calculation, reduct generation and rough set classification.

Comparison of Performance for Intrusion Detection System Using Different Rules of Classification

Classification is very important for designing intrusion detection system that classifies network traffic data. Broadly traffic data is classified as normal or anomaly. In the work classification performance using rules obtained by different methods are applied on network traffic and compared. Classifier is built based on rules of decision table, conjunctive rule, OneR, PART, JRip, NNge, ZeroR, BayesNet, Ridor from WEKA and using rough set theory. Classification performance is compared applying on KDD data set where the whole data set is divided into training and test data set. Rules are formed using training data set by different rule generation methods and later applied on test data set to calculate accuracy of classifiers.

Security and Privacy at Cloud System

This chapter is focused to provide security mechanism for complete cloud system by implementing encryption and intrusion detection system. Hybrid encryption is applied on data at cloud client level so that data in medium will be safe as well as data will be stored in cloud server in safe mode. Data in server will be accessible only to the authorized users which have the decryption key. Computation for decryption becomes challenging and difficult in case of hybrid encryption. The second phase of security will be applied in cloud server by implementing intrusion detection system which will detect the anomaly traffic towards server and block the unauthorized and unauthenticated traffic. Dimension reduction techniques are also focused in this chapter to make the efficient intrusion detection system.

Machine Learning Based Medicinal Care in Cloud

The cloud computing is the emerging technology with benefits like reduced investment and maintenance cost, increased scalability, availability and reliability. With the characteristics of ubiquitous access, on-demand access, pay-per-use service and resiliency, the cloud computing is applied everywhere. Due to non-availability of specialists in primary health centers in developing and under developing countries, patients are not treated properly. In this paper, treatment model for treating the patients in primary health center has been suggested. This model is built on machine learning system, where IaaS, SaaS, PaaS based cloud computing contribute a lot. Data like symptoms and prescribed medicines for different diseases are collected by different hospitals and various research centers. Same data will be stored in cloud and for similar types of symptoms; medicines can be prescribed whenever required. Thus, in our suggested treatment model, machine learning will help in treating the patients with stored knowledge of specialists through cloud computing.

Comparison of Supervised Learning and Reinforcement Learning in Intrusion Domain

In modern world use of network is increasing exponentially. Network security needs attention of computer science researchers. Intrusion Detection System is software / hardware which detects intruder in the network or host system. Classification plays an important role in Intrusion Detection System. Detection of anomaly or normal traffic is main working philosophy for such type of system. For detection of online traffic, learning of the system is required. In our paper, performance of Supervised Learning and Reinforcement Learning is compared in Intrusion Domain. NSL-KDD data is considered for our work. In that dataset for each object 41 conditional attributes and one decision class attribute are mentioned. Out of 41 attributes, 7 attributes are discrete and 34 attributes are continuous. Using feature ranking method, number of discrete attributes are reduced and these reduced number of attributes are used for classification in Supervised Learning. Some Supervised Learning like CS-MC4, Decision List, ID3, Naive Bayes, C4.5, Rnd Tree are applied on this data set and compared this classification result with classification accuracy derived from Reinforcement Learning combined with Rough Set Theory classifier.

Reduction of Data Size in Intrusion Domain Using Modified Simulated Annealing Fuzzy Clustering Algorithm

Network security is becoming an important issue as the size and application of the network is exponentially increasing worldwide. Performance of Intrusion Detection System (IDS) is greatly depends on the size of data and a systematic approach to handling such data. In the paper, modified simulated annealing fuzzy clustering (SAFC) algorithm has been proposed using the concept of Rough set theory that removes randomness of the SAFC algorithm and applied on intrusion domain for data size reduction. The reduced data set increases classification accuracy in detecting network data set as ‘anomaly’ or ‘normal’ compared to the original data set. Davies-Bouldin (DB) validity Index is evaluated to measure the performance of the proposed IDS.