Theses and Dissertations
Permanent URI for this collection
Browse
Search Collection
1 - 5 of 31740
-
PublicationTerritoriality: Theoretical building blocks for planning office workstations( 2000)Territoriality is a known as one of the most fundamental factors in the office environment which remains eternally as a basic instinct of humankind, regardless of trend of office design. This study examined factors that affect office territoriality and their relationship to the physical components of the workstation. The following questions were addressed: 1. What are the fundamental human factors that affect territoriality in the office environment? 2. What is the relationship between physical office workstation components and territoriality, especially the reinforcement of territoriality? 3. What is the most effective method of developing a set of data that can be used for practitioners? The end product of this study is a collection of principles, developed and summarized by translating complicated information on territoriality into a clear, comprehensible data set incorporating symbols and illustrations. It will help designers understand the nature of territoriality in terms of workstations. Moreover, it will stimulate and guide interior designers to be prepared for designing various types of work places, today and in the future. The method employed encourages further validation of the data set and its refinement as a tool for the designers of office environments.
-
PublicationAgent toolkit for distributed knowledge networks( 2000)This thesis focuses on the design and development of an agent toolkit which can provide developers with agent models that are useful for rapid design and prototyping of applications in a given domain. Most of the existing agent systems belong to either of the two categories agent systems specific to a problem or generic agent architectures that describe functions and properties of an agent at a very abstract level. There is a need to devise a toolkit that can provide agent models that can be reused for applications belonging to a specific domain. These agent models are generic for a domain of applications and are more specific than the models provided by the generic agent architecture. The application which motivated us to build such a toolkit is distributed knowledge networks. Advances in sensor, high throughput data acquisition, and digital storage techniques have made it possible to acquire and store large volumes of data. Distributed Knowledge Networks are proposed as a solution for selective, reactive, proactive information retrieval, knowledge discovery, distributed problem solving to translate these advances into fundamental scientific advances and technological advances. We have designed a toolkit which gives agent prototypes for such networks. We propose formal specifications of agent models which are useful in domain of distributed knowledge networks. Agent toolkit implements some of these prototypes. This would help developers to concentrate more on agent related aspects of the system rather than generic features. It also helps to develop the generic domain specific features that can be used across applications. To demonstrate utility of the agent toolkit, design and implementation of applications built using the toolkit is described. Recent advances in data storage and acquisition technologies made possible to produce increasingly large data repositories. Moreover most of these data sources are physically distributed and assembling them together at a central site is an expensive process in terms of network bandwidth. Many organizations are not willing to provide the actual data for security reasons but they are willing to give some summary of data. This generates a need for Learning Algorithms that re able to learn from distributed data without actually collecting data. In this thesis, we explore a general technique involving information extraction to enable the current learning algorithms to learn from the distributed data. We propose distributed learning techniques for horizontally and vertically distributed data. These techniques are illustrated with the help of decision tree learning algorithm. The results obtained in case of distributed learning is same as in the centralized case where the complete data is at a central location. We give time, space and communication cost analysis for the distributed algorithms.
-
PublicationShort-pulsed laser deposition of thin films of YBaCuO( 2000)This thesis investigates deposition of superconducting thin films of YBaCuO using short pulsed laser deposition technique. The effects of temporal pulse width and target density on the deposition of thin films of YBaCuO were studied. Experiments were conducted with pulse widths of 27 ns, 16 ns, and 150 fs, and target densities of 80% and 90%. Scanning electron microscope photomicrographs and profilometer traces show a striking difference between nanosecond and femtosecond laser irradiation. Shortening the pulse width reduced particulate formation, provided stoichiometry, and improved the film properties. Decreasing the target density produced thicker but nonuniform films, and reduced particulate formation. A thermal model was applied to predict the temperature distributions inside the target and thereby identify the effects of short pulses on the collateral thermal damage. The surface ablated by the ns-laser were highly disrupted, underwent melting and resolidification, contained micron-sized pores, and exhibited large heat affected zones. In contrast, the surface ablated by the fs-laser were smooth and free from recast layer, underwent direct evaporation, and consisted of cone-like structures.
-
PublicationA state machine approach for network intrusion detection( 2000)The past two decades have seen the emergence of computer networks, which, while bringing increased computing power, greater flexibility and better price/performance ratio, have also brought many risks and challenges. To account for this, there has been a simultaneous increase in the research effort focussed on developing techniques to ensure the security and reliability of computer networks. This is particularly true in today's day and age, where network-based attacks are increasing in frequency, severity and sophistication. Thus, development of network intrusion detection techniques that are dedicated to identifying these attacks has gained prime importance. In this thesis, we present a new approach to network intrusion detection based on modeling the normal network traffic behavior by reconstructing sessions, in terms of finite state machines, at different protocol layers in the protocol suite and learning relevant information about this normal traffic with reference to these models. We use a rule-based specification language to specify a unique finite state machine for each protocol, at different layers in the protocol suite, such that each instance of a state machine represents a distinct session for the protocol which it models. Relevant information is learnt from normal network traffic in association with these state machine models to establish an accurate model for normal behavior. The actual behavior of the system is monitored with reference to this model and intrusions are detected as deviations from normal behavior. We present this technique of intrusion detection using state machine models, describe its implementation and present experimental results.
-
PublicationThe use of travel demand models with geographic information systems in transportation planning( 2000)Geographic information systems (GIS) have become a recognized tool in the travel demand modeling field. Various techniques have been devised to use the power of both GIS and travel demand models for transportation planning projects. By adding pull-down menu options to ArcView's graphical user interface the ArcView-Tranplan interface forms a seamless bond between the travel demand modeling package Tranplan and ArcView GIS. The interface was created for the Iowa Department of Transportation at Iowa State University's Center for Transportation Research and Education (CTRE). Three articles discussing the tools incorporated into the interface have been compiled into this thesis. The first article discusses the standard travel demand model analysis tools incorporated into the interface such as screenline, turning movement display, calibration plot and shortest path algorithms. The second article covers the quality assurance functions incorporated into the interface. These functions include model documentation, verification of centroid loading, external trip table editing, one-way link versus turn prohibitor conflict and traffic assignment noise. Article three details the incorporation of the Federal Highway Administration's Benefit Efficiency Analysis Module (BEAM) economic analysis program. BEAM uses travel demand model outputs to produce a benefit-cost analysis for a proposed scenario against the base scenario condition. The ArcView-Tranplan interface allows users to quickly gather all required data for a BEAM analysis, then display the BEAM output with aerial photography, land use or infrastructure coverages or other data sources in ArcView. The ArcView-Tranplan interface is provided free of charge on the CTRE website and is targeted for use by both the Iowa DOT and metropolitan planning organizations in the Iowa area.