Madhavi Sinha

Genetic optimization for the join ordering problem of database queries

A query optimizer is a core component of any Database Management System. As a database query optimizer might face different voluminous and complex queries, leading to a huge search space of alternative query plans, it should be appropriate to adapt the search strategy to the problem solving technique which handles complex and large data. Genetic Algorithms (GAs) avoid the high cost of optimization and provide flexibility by being independent of the problem specific knowledge. These qualities make them a viable option for solving the query optimization problem. All query optimization algorithms primarily deal with joins. Our study concerns the use of GA for join order optimization. Prior theories indicate that genetic algorithms are apposite for optimizing join expressions and produce solutions of high quality within a reasonable running time. We have implemented the GA technique on RDBMS queries and found that the GA based optimizer performs better for queries involving large number of joins.

Performance analysis of agent based IPSM

In computing, Inter-process communication (IPC) is a set of methods for the exchange of data among multiple threads in one or more processes. Operating system works as an inter process synchronization manager (IPSM). In the present paper agent based inter process synchronization manager is described and its performance is compared with other non agent based IPSM on different operating systems.

State based test case generation using VCL-GA

This paper presents a method for test cases generation from a state machine. This paper investigates variable chromosome length genetic algorithm for automatically generating state based test cases. For efficiency we developed a Genetic algorithm that optimize and select test case which are covering maximum number of states and transitions on distinct chromosome length. Variable chromosome Length Genetic algorithm was requisite because of throttle in representing FSM as a binary chromosome discussed in this research paper and various issues discussed in paper. In this paper, Genetic Algorithm starts with a short chromosome length and retrieves the best solution with respect to optimized test cases covering maximum states and maximum transition of chosen state machine. The best solutions are then moved to the next stage with longer chromosome length. In this paper population size is again varying on the basis of chromosome length. To validate the introduced technique, experiment with the test-data-generation technique has been applied on numerous state machines.

Run-time performance analysis of non-agent based solution for Inter Process Synchronization problem

Inter Process Synchronization is aimed at providing interrupt free and deadlock free well-structured communication among processes. The approach used is generally non-agent oriented approach which may lead to deadlock situation. This paper analyzes the runtime performance of non-agent based solution to IPS Problem. In this paper the testing is carried out and the results are obtained under time intervals T1 to T10 for eleven different parameters. Further based upon the correlation among parameterized variables, only five dependent variables were selected during the test conducted. The analysis is based upon the testing conducted on Window 7, a GUI based operating system. The results motivate the need of agent oriented approach to resolve the IPS problem.

Applying variable chromosome length genetic algorithm for testing dynamism of web application

This paper presents a proficient technique to test all static and dynamic analysis of chosen web application using state machine and variable chromosome length genetic algorithm. We propose a framework to generate user session based state machine model of dynamic behavior of a web application and the technique used to test the correctness of generated state machine is Variable chromosome length genetic algorithm. In this study, the author projected a new reflexive approach User Session based State Machine to test the dynamic content of a web application. This proposed technique will walk through various states of a highly dynamic web application, create a model of the active user session navigational paths and illustrate the dynamic content by means of a Finite State Machine (FSM). We applied Variable chromosome length genetic algorithm(VCL-GA) for generating test cases for proposed framework generated state machine. During test case generation we are refining test cases of generated state machine on progressive refinement manner using proposed VCL-GA. Finally the author depicts comparison of simple GA and variable chromosome length GA. Results shows that variable chromosome length GA gives efficient and better result on large state machine and especially in dynamic behaviors of application results are efficient.

Pattern construction by extracting user specific features in keystroke authentication system

One of the biggest challenges in behavioral biometric systems is to select biometric features for pattern construction due to large domain of users behavior. It is desirable that, behavioral pattern should be as small as possible for low computational cost of the system. But reduced pattern size costs accuracy of the authentication system. In this paper, authors have proposed to select user specific features for pattern construction, which is an obvious choice, as specific features are much stronger than others. This paper also proposes a technique to select user specific feature set for keystroke dynamics system. It is also demonstrated through experiments that, when user specific features are used, pattern size is drastically reduced without compromising authentication accuracy of the system.

Investigation of energy efficient routing parameters & protocols

Routing protocol in ad hoc networks essentially needs a routing mechanism to establish a path from source to destination. The protocols differ in techniques used to discover path and recover in case of failure. Traditional routing protocol such as AODV, DSR, DSDV, OLSR, and AOMDV have been found to perform efficiently in the networks without consideration of battery power or network life time. This paper analyzes the traditional routing protocols to find a batter candidate on which further applied energy efficient techniques to optimize the energy consumption. The performance of protocols adjudge energy consumption versus various parameters such as data insertion rate, mobility, network density, the degree of nodes, number of connections in the networks. The performance analysis implemented in NS-2 simulator and finds the reactive routing protocols AODV & AOMDV perform as compared to proactive routing protocols DSDV & OLSR.

Multi-response metric analysis of Adaptive Fault Tolerant Replication (AFTR) routing protocol for MANET

In general, performances of routing protocols are analyzed through simulation. Tools like statistical design of experiments also provide a variety of significant outcomes. To improve the functioning of routing protocols it is required to evaluate the factors that affect its performance. Such evaluation is carried out with the help of response metrics. Two kinds of evaluation are possible: single-response and multi-response. By varying single factor at a time, the effect on the protocol performance is measured through response metrics individually. It is referred as single-response metric analysis. On contrary, the effects of such variation are quantified through more than one response metrics simultaneously and it is referred as multi-response analysis. The paper presents the multi-response analysis of Adaptive Fault Tolerant Replication (AFTR) routing protocol using five response metrics like packet delivery ratio, delay, throughput, routing overhead and packet drop under the variation of five factors named network size, transmission rates, mobility speed, optimal number of copies and pause time. The findings show network size is the leading factor that optimizes all the metrics of the AFTR protocol simultaneously. The second most important factor is the optimal number of copies which have been evaluated using adaptive fault tolerant replication strategy.

Face recognition using user specific features

Holistic approaches of face recognition are not robust to illumination, scale, occlusion and age variations. Various studies indicate that the performance of holistic approaches degrades as the face database size increases. In this paper, we propose a user specific landmark geometry based approach that assigns weights to different geometrical distances according to their role in face recognition process. The term user specific implies that, the feature vector of each face in the database comprises a unique set of features. The idea is inspired by the fact that, humans recognize individuals by identifying ones unique characteristics (e.g. unusually long nose or big eyes). We propose a method to identify these characteristics and use them as features for particular face. Experimental results show that our specific feature approach has better accuracy and faster than available technique.

Avoiding state explosion problem of generated AJAX web application state machine using BDD

There is growing tendency of users to use web application in place of desktop application because of technological advancement such as AJAX. AJAX is used to build single page web application because content and structure can be changed using AJAX features like Asynchronous communication and run time DOM manipulation. To understand and analyze extreme dynamism of web application, we implemented a tool which is used to generate state machine model of dynamic behavior of user session. In this research paper, we validated and evaluated efficiency of the generated model to detect faults embedded area. However, the state machine can be huge and unbounded and may hit by state explosion problem for large number of user session traces and for extreme dynamism. In this paper to avoid this problem, we used binary decision diagram, a model checking technique to reduce state space at the time of state machine generation. Finally, we are able to control the size of generated state machine without affecting faults embedded area.