Computer Engineering

We propose a novel method based on artificial neural network (ANN) for beam position calculation and fault diagnosis of beam position monitors (BPM) in electron storage rings. BPM are diagnostic devices used to determine position of stored electron beam in storage rings. BPM commonly uses four button type electrodes as sensors to sense the electric field of electron beam circulating in storage ring. A mathematical polynomial of suitable degree is generally used to compute beam position from the button electrode signals of the BPM when beam excursions are large from the design orbit. The coefficients of the polynomial are derived from the bench calibration data of BPM. In the proposed new method, a neural network predicts the position of electron beam using four electrode signals of a BPM. A feed-forward network with three hidden layers using backpropagation training algorithm has been designed and trained with the bench calibration data of each BPM. The beam position predicted by the network was compared with conventional polynomial method. Neural network based method was tested on the BPM of 2.5 GeV electron synchrotron radiation source named Indus-2 at Raja Ramanna Centre for Advanced Technology, Indore, India. The root mean square (rms) error in neural network predicted beam position in horizontal plane and vertical plane was 24 and 26 microns respectively as compared with 100 and 101 microns with first polynomial and 61 and 66 microns with second polynomial in the central region of ± 5 mm on bench calibration data. The reliability of beam position measurement was assured by another neural network by doing self consistency check. In this paper we present architecture, training of neural network and improvement in beam position measurement in comparison with polynomial method.

Cloud Computing involves aggregation of computing, storage and network resources into a single entity called cloud in which task is performed. Cloud computing is an evolution of the virtualization, Service-Oriented Architecture (SOA) and Utility Computing. In efficient resource management, the virtualized data center is always a practical concern and it has attracted significant attention. This allocation mechanism is desired to maximize the spaces for technical cloud providers. In this paper uses the overbooking and automatic from physical machine management to avoid resource over-provision problem according to its runtime demand. It proposes an automatic model to control the overbooking policy while it provides usages probability based on the performance and risk estimation. To cooperate with overbooking policy, it optimizes the VM placement with resource-aware strategy to satisfy application's QoS requirement. In this paper, Automated VM provisioning approach in which multiple VMs are consolidated and provisioned based on an estimate of their aggregate capacity needs. To implement cloud analytics in cloud computing in which it have reached the goal of achieves the overload avoidance and green computing concept successfully.

The growing importance of quality software systems has brought about increasing demand for efficient software testing. Traditionally, there is a typical approach for testing the software manually, which consumes a lot of time and effort. Hence there is a need to decrease the amount of resources needed and to increase the efficiency and throughput. One attractive solution to this problem is test automation, i.e. allocating certain testing tasks to computers. This thesis opens up the discussion on test automation framework made by us. Here we have used Selenium Web driver, Java, HTML and other necessary languages and packages for implementing the framework. The framework uses the Driver Script and allows to automatically fetch data from the spreadsheets and uses them in testing the web application automatically and also generating report automatically for the same.

The watershed algorithm is a commonly used method of solving the image segmentation problem. Watershed based image segmentation is selected for the implementation, as it exhibits least computational complexity, good segmentation quality and can be implemented in the FPGA. This paper proposes a new parallel watershed based image segmentation technique and its architecture is implemented on Virtex-4 FPGA board. The results show that the proposed architecture requires minimum hardware resources, low execution time and is also suitable for use in real time applications.

The major aim of bringing forth this project is to design a system for Text Dependent Speech Recognition. The speech recognition system contains two main modules, namely “feature extraction” and “feature matching”. In this project, the MFCC (Mel-Frequency Ceptral Co-efficient) is used to simulate the feature extraction module. In MFCC algorithms the Ceptral co-efficients are calculated on the Mel frequency scale where the frequency bands are equally spaced on the Mel scale. For the reduction of amount of data in order to reduce the computation time the VQ (Vector Quantization) is used. VQ is the data compression method based on the principle of block coding. Because of the accuracy of the used algorithms the accuracy of this speech recognition system is high. Using these algorithms we achieve the text dependent speaker identification and speech recognition and thus providing improved efficiency

It is been a long “Time” ,it is usually the most story’s of fairy tales started and also the second most important word is “ago” all these words which give as implication of past are related to Time, in our teachings, we learn History a lot, we learn about Mohan-ja-dare, but never seen its site, don’t we known that it is the most important part of Indus Valley civilization. In the coming years, we may not be able to see it’ ,learn about it & understand the first architecture of sanitation and hygiene gifted by our own dynastic people, therefore we must protect these site’s & try to teach more about them as these site are aspiration for young Indians to make economic cost technology for future. In the coming of ages the problem remained how to remember things and also how to preserve ,make available to our coming generations at last. Hence forth we require more wide ,simple approach to understand this. We propose a need for augmentation, we call it as augmented room .In this paper we have studied conceptual of Augmentation and observed that how Augmentation “a mixed real-virtual environment technology” will help us reach our goal. Our major aim is to prevent these monuments for our future generations so that next generation will learn and live with it. Making it possible through augmented room. This paper gives introduction to new kind of perseverance of heritage site and new transit to technology.

This paper presents a new technique, used to register two images of the same modality and that are assumed to be differ from each other, by a rotation and a scale. Recently, researchers have introduced image registration techniques using log-polar transform for its scale and rotation invariant properties. However, it suffers from non-uniform sampling and it leads to inaccurate registration. Inspired by log polar transform, a new technique, fast uniform polar transform (FUPT) is used to overcome these disadvantages. The proposed method yields more accurate registration than log polar transform.

Given a collection of cities and the cost of travel between each pair of them, the traveling salesman problem, or TSP for short, is to find the cheapest way of visiting all of the cities and returning to your starting point. In the standard version we study, the travel costs are symmetric in the sense that traveling from city X to city Y costs just as much as traveling from Y to X. The Traveling Salesman Problem is typical of a large class of "hard" optimization problems that have intrigued mathematicians and computer scientists for years. Most important, it has applications in science and engineering I attempt to apply simulated annealing to find (Sub-Optimal) solutions to TSP with 100-200 cities randomly. The performance by changing the parameter values and try to understand how fast and effective simulated annealing algorithm can generate a solution to TSP problem.

Existing and in recent times proposed clustering algorithms are studied and it is known that the k-means clustering method is mostly used for clustering of data due to its reduction of time complexity. But the foremost drawback of k-means algorithm is that it suffers from sensitivity of outliers which may deform the distribution of data owing to the significant values. The drawback of the k-means algorithm is resolved by k-medoids method where the novel approach uses user defined value for k. As a result, if the number of clusters is not chosen suitably, the accuracy will be minimized. Even, K-medoids algorithm does not scale well for huge data set. In order to overcome the above stated limitations, a new grid based clustering method is proposed, where time complexity of proposed algorithm is depending on the number of cells. Simulation results show that, the proposed approach has less time complexity and provides natural clustering method which scales well for large dataset.

Speaker identification followed by speech recognition system is developed. The system makes use of MFCC (mel frequency cepstrum coefficients) to process the input signal and extract the features. VQ (Vector quantization) is used to identify the speaker. LPC (Linear Predictive Coding) and BNN (Back Propagation Neural Network) technique of hyperbolic tangent function under ANN (Artificial Neural Network) is used for speech recognition system. The implementation is done using MATLAB. The results of the developed system proved to be efficient and faster.