Gugulothu Narsimha

Data perturbation and feature selection in preserving privacy

Privacy Preserving plays a vital role; in designing various security-related data mining applications. Protecting sensitive information in data mining has become an important issue. Data distortion or data perturbation is a critical component, widely used to protect sensitive data. Many approaches try to preserve privacy by adding noise or by matrix decomposition methods. In this paper we propose data distortion methods such as singular value decomposition (SVD) and sparsified singular value decomposition (SSVD) technique along with feature selection to reduce feature space. Various privacy metrics have been proposed to measure the difference between original dataset and distorted dataset and degree of privacy protection. Our experimental results use a real world dataset. It shows a feasible solution using sparsified singular value decomposition along with a feature selection, which could better preserve privacy. Extracting accurate information from datasets will make reasonable decisions using data mining algorithms. The mining utility on perturbed data is tested with a well known classifiers such as SVM, ID3 and C4.5.

Design of novel fuzzy distribution function for dimensionality reduction and intrusion detection

Reducing the processing complexity is main challenge when dealing with intrusion detection systems. The processing complexity is reduced and efficiency is increased if we can reduce the number of dimensions so that only the minimum number of dimensions is retained. This work mainly targets on achieving dimensional reduction for intrusion detection using a novel membership function. The membership function is used to cluster the features in iterative incremental manner and obtains a reduced dimensional representation which retains the original distribution of process data. A case study is discussed to explore working of proposed model.

Dynamic TCP-Vegas based on cuckoo search for efficient congestion control for MANET

In the internet congestion control, transmission control protocol (TCP)-Vegas is a source algorithm that provides better performance. However, it has two main issues: first one is rerouting and unable to identify updation of round-trip time (RTT) when changes occurred in the network, these two problems leads to affects the performance of Vegas. These drawbacks persist mostly in the Vegas estimation process of the propagation delay, i.e. BaseRTT. In this paper, we proposed a novel algorithm that uses the cuckoo search optimisation algorithm for selecting the optimal value of BaseRTT. Also, our proposed algorithm has dynamically considered slow start algorithm based on the estimation in real time, the available bandwidth and adjust decrease/ increase rate in congestion avoidance phase for a particular network environment. Simulation results have shown that our proposed algorithm can effectively avoid packet losses and attain the maximum throughput when compared with existing algorithms.

Cross-layer based routing protocol and solution to packet reordering for TCP in MANET

In mobile ad hoc networks (MANETs), link failures and route changes occur most frequently, which may result in packet reordering. Transmission control protocol (TCP) performs poorly in such environment, which misinterprets the reordered packets as lost packets due to congestion. This has motivated us on developing a new protocol towards the packet reordering for improving the performance of TCP in MANETs. Optimal path or route selection is the major concern to improve the energy efficiency and network lifetime. In this paper, trust aware routing protocol for selecting optimal route in MANET is proposed. Based on this protocol, trust value for each node is calculated using direct and indirect trust value. Then the routing cost metric value is calculated and the path with minimum cost metric value is chosen as the best path in the network. After selecting the optimal path, data packet is to be transmitted through the optimal path. During the transmission, the data packet may get dropped or reordered due to congestion or mobility. A cross layer approach between network layer and transport layer to identify the dropped and reordered packets in the network is proposed in this paper. Simulation results are reported, which support this proposal.

Multiplicative Data Perturbation Using Fuzzy Logic in Preserving Privacy

In Data mining is the method of extracting the knowledge from huge amount of data and interesting patterns. With the rapid increase of data storage, cloud and service-based computing, the risk of misuse of data has become a major concern. Protecting sensitive information present in the data is crucial and critical. Data perturbation plays an important role in privacy preserving data mining. The major challenge of privacy preserving is to concentrate on factors to achieve privacy guarantee and data utility. We propose a data perturbation method that perturbs the data using fuzzy logic and random rotation. It also describes aspects of comparable level of quality over perturbed data and original data. The comparisons are illustrated on different multivariate datasets. Experimental study has proved the model is better in achieving privacy guarantee of data, as well as data utility.

Optimized Clustering with Statistical-Based Local Model for Replica Management in DDM over Grid

With the increasing complexity of the forms of the data (unstructured, massive, real-time, and heterogeneous), Distributed Data mining (DDM) approaches encounters a significant problem over grid infrastructure. The paper has identified a challenging problem i.e. an effective replica management which cost maximum resources to process the data from the warehouse in distributed data mining. The proposed system introduces a technique which presents a unique clustering mechanism of the data extracted from the replica (warehouse) and applies a novel statistical-based local model in order to extract the non-repetitive and unique data for accomplishing faster response time during distributed data mining. Powered by optimization using genetic algorithm, the proposed system offers better response time with increasing traffic load as compared to the similar existing technique of distributed data mining.

An approach for intrusion detection using fuzzy feature clustering

This work discusses the approach for intrusion detection and classification by devising a membership function, inspired from [43] and is used in this work to carry the dimensionality reduction of processes present in the training set. The reduced process representation is then used to perform classification and prediction for detecting intrusion. The reduced representation of processes retains the system call distribution same as the initial process representation.

Text mining based approach for intrusion detection

Intrusion detection is classified as NP-Hard in the literature even today. Also supervised learning also termed classification, when performed on high dimensional documents has problem from the noise or outliers, which make the text classification inaccurate and leads to reduced accuracy by classifiers. We discuss the feature reduction methods which we adopted to achieve dimensionality reduction. In the Feature Extraction process, the high dimensional text documents are projected onto their corresponding low dimensional representation in feature space through using algebraic rules and transformations. The objective is to find optimal transformation matrix corresponding the input high dimensional document feature matrix. This objective is achieved in this thesis by using the concept of feature clustering and through clustering the features into a optimal set of clusters by designing a novel fuzzy membership function. The membership function designed retains the original distribution of words in the documents which is the importance of this approach.

Trust-based light weight authentication routing protocol for MANET

In mobile ad hoc network MANET, most of the existing routing technique lags trusted way of communication among the mobile nodes. The control messages are also prone to the external threats. Also, the act of performing authentication of the nodes during each routing process causes increased overhead. Hence, in this paper, we propose a trust-based light weight authentication routing protocol in MANET. Initially, a multipath route discovery technique is utilised that selects the path with maximum packet success ratio as optimal path for data transmission. For each node in the chosen path, global trust value is estimated based on direct and indirect trust values of the node. If the trust value of any node is below threshold value, then it is authenticated using the secret sharing technique. This authentication technique enhances the reliability, redundancy and network lifetime. By simulation results, we show that proposed protocol improves the reliability and security of routing.

OCSLM: Optimized Clustering with Statistical Based Local Model to Leverage Distributed Mining in Grid Architecture

Grid computing offers significant platform of technologies where complete computational potential of resources could be harnessed in order to solve a complex problem. However, applying mining approach over distributed grid is still an open-end problem. After reviewing the existing system, it is found that existing approaches doesn’t emphasized on data diversity, data ambiguity, data dynamicity, etc. which leads to inapplicability of mining techniques on distributed data in grid. Hence, the proposed system introduces Optimized Clustering with Statistical Based local Model (OCSLM) in order to address this problem. A simple and yet cost effective machine-learning based optimization principle is presented which offers the capability to minimize the errors in mined data and finally leads to accumulation of superior quality of mined data. The study outcome was found to offer better sustainability with optimal computational performance when compared to existing clustering algorithms on distributed networking system.