Pandya, Abhijit S.

Artificial Neural Networks have been widely used for obtaining solutions for combinational optimization problems. Traveling Salesman problem is a well known nonlinear combinational optimization problem. In Traveling Salesman problem, a fixed number of cities is given. An optimal tour of all these cities is required such that each city is visited only once and the total tour distance to be covered has to be minimized. Hopfield Networks have been applied for generating an optimal solution. However there are certain factors which result in instability and local optimization of Hopfield Networks. In such cases the solutions obtained may not be optimal and feasible. In this thesis, the application of the K-Means algorithm is combined with the Hopfield Networks to generate more stable and optimum solutions to traveling salesperson problem.

This thesis discusses the implementation of a feed forward NN using time series model to predict the sudden rise or sudden crash of a company's stock prices. The theory behind this prediction system is Pattern recognition. Pattern recognition techniques for time-series prediction are based on structural matching of the current state of the time-series with previously occurring states in historical data for making predictions. This study reports the result of attempts to predict the Motorola stock price index using artificial neural networks (ANN). Daily data from January 1999 to December 2001 were taken from the NYSE. These data are classified based on criteria of an n% fall or rise of price corresponding to the previous day close price. A novel method using Hurst exponent is used in selecting the data set. These data are fed into a Back Propagated Neural Network. The number of hidden layers and number of neurons are systematically selected to implement a better predicting machine. The implemented model is tested using both interpolated and extrapolated data. Fundamental limitations and inherent difficulties when using neural networks for processing of high noise, small sample size signals are also discussed. Results of the prediction are presented and an elaborate discussion is made comparing the results.

In this thesis, application of GMDH Algorithm to real life problems is studied. A particular type of GMDH Algorithm namely TMNN is chosen for this purpose. An effort is made to forecast S&P Index Closing Value with the help of the forecaster. The performance of the TMNN Algorithm is simulated by implementing a tool in C++ for developing forecast models. The validation of this simulation tool is carried out with Sine Wave Values and performance analysis is done in a noisy environment. The noisy environment tests the TMNN forecaster for its robustness. The primary goal of this research is to develop a simulation software based on TMNN Algorithm for forecasting stock market index values. The main inputs are previous day's closing values and the output is predicted closing index.

The objective of this work is to apply and investigate the performance of a neural network-based receiver for interference cancellation in multiuser direct sequence code division multiple access (DSCDMA) wireless networks. This research investigates a Receiver model which uses Neural Network receiver in combination with a conventional receiver system to provide an efficient mechanism for the Interference Suppression in DS/CDMA systems. The Conventional receiver is used for the time during which the neural network receiver is being trained. Once the NN receiver is trained the conventional receiver system is deactivated. It is demonstrated that this receiver when used along with an efficient Neural network model can outperform MMSE receiver or DFFLE receiver with significant advantages, such as improved bit-error ratio (BER) performance, adaptive operation, single-user detection in DS/CDMA environment and a near far resistant system.

Backpropagation is a standard algorithm that is widely employed in many neural networks. Due to its wide acceptance and implementation, a standard benchmark for evaluating the performance of the algorithm is a handy tool for software design and development. The object of this thesis is to propose the use of the classic XOR problem for the performance evaluation of the backpropagation algorithm, with some variations on the input data sets. This thesis covers background work in this area and discusses the results obtained by other researchers. A series of test cases are then developed and run to perform the performance analysis of the backpropagation algorithm. As the performance of the networks depends strongly on the inputs, the effect of variation of the design parameters for the networks are evaluated and discussed.

This thesis is an effort to present the performance analysis of the RLP in cdma2000, which uses the NAK based ARQ scheme for Random Error Channels. The performance analyses is done in terms of throughput and mean extra delay, which are calculated analytically and are compared with the results generated from the simulations. As the demand for higher data rates increases over the wireless channels, this thesis studies the effect of the random errors over the different types of RLP frame formats and also the performance of the NAK based ARQ mechanism used in these conditions. The simulation provides with the overall characteristics of the throughput and the mean extra delay in terms of realistic environment parameters like Eb/No and probability of packet error (Pe), based on the channel conditions.

Time series is a phenomena which appears in the financial world in various forms. One of the objectives of time series is to forecast the future based on the past. The goal of this thesis is to use foreign exchange time series, and predict its future values and trends using neural networks. The thesis covers background work in this area and discusses the results obtained by other researchers. A neural network is then developed to predict the future values of the USD/GBP and USD/DEM exchange rates. Both single-step and iterated multi-step predictions are considered. The performance of neural networks strongly depends on the inputs supplied. The effect of the changes in the number of inputs is also considered, and a method suggested for deciding on the optimum number. The forecasting of foreign exchange rates is a challenge because of the dynamic nature of the FOREX market and its dependencies on world events. The tool used for building the neural network and validating the approach is "Brainmaker".

Information access systems have traditionally focused on retrieval of documents consisting of titles and abstracts. The underlying assumptions of such systems are not necessarily appropriate for full text, structured documents. Context and structure should play an important role in information access from full text document collections. When a system retrieves a document in response to a query, it is important to indicate not only how strong the match is (e.g., how many terms from the query are present in the document), but also how frequent each term is, how each term is distributed in the text and where the terms overlap within the document. This information is especially important in long texts, since it is less clear how the terms in the query contribute to the ranking of a long text than a short abstract. This thesis does research in the application of information visualization techniques to the problem of navigating and finding information in XML files which are becoming available in increasing quantities on the World Wide Web (WWW). It provides a methodology for presenting detailed information about a specific topic while also presenting a complete overview of all the information available. A prototype has been developed for visualization of search query results. Limitations of the prototype developed and future direction of work have also been discussed.

Wireless personal area networks (WPAN) are becoming more and more popular for use by mobile professionals in areas like airports, hotels, or convention centers. The demand for wireless networks is expected to undergo an explosive growth as Bluetooth(TM) capable devices become more and more popular. In such a scenario, it is imperative that designer are aware of the performance characteristics of several Bluetooth(TM) networks operating within the same area. There are several issues that need consideration like security, self-interference and adjacent network interference. The objective of this research is to evaluate the performance of a Bluetooth(TM) network in the presence of self-interference which included adjacent and co-channel interference from neighboring Bluetooth(TM) networks. Specific to the above topics of interest, the following research tasks are performed: (1) The magnitude of self-interference problem in Bluetooth(TM) networks. (2) The system throughput is evaluated by varying duty cycles of the various networks. (3) The pathloss difference is measured between the desired and the interfering device.

This dissertation proposes amodular Artificial Neural Network (ANN) based buffer allocation and routing control model for ATM switching networks. The proposed model considers limited buffer capacity which can adversely impact the switching performance of ATM switching networks. The proposed ANN based approach takes advantage of the favorable control characteristics of neural networks such as high adaptability and high speed collective computing power for effective buffer utilization. The proposed model uses complete sharing buffer allocation strategy and enhances its performance for high traffic loads by regulating the buffer allocation process dynamically via a neural network based controller. In this study, we considered the buffer allocation problem in the context of routing optimization in ATM networks. The modular structure of the proposed model separates the buffer allocation from the actual routing of ATM cells through the switching fabric and allows adaptation of the neural control for routing to different switching structures. The influence of limited buffer capacity, routing conflicts, statistical correlation between arriving ATM cells and cell burst length on ATM switching performance are analyzed and illustrated through computer simulation.