Engineering and Computer Science.

TRANSPORT IN GRASS SWALES. Yukio Nara and Robert E. Pitt, Dept. of Civil and Env. Eng., Univ. of Alabama, Tuscaloosa, AL 35487.

I am conducting research in investigating the effectiveness of grass swales for sediment transport. Grass swales are vegetated open channels where collected stormwater runoff flows through. They are often used as an alternative to a concrete gutter for stormwater management because of their advantages of infiltration and filtration of stormwater. The aim of research is to understand the effects of different variables. The variables are slope, grass types, time interval, depth of flow, and length of swales. To analyze the variables, solutions with known concentrations of sands and fine particles of Aluminum Oxides are pumped onto inclined grass swales in different conditions for each variable. 108 samples were collected and analyzed by turbidity, total solids, and particle size distribution analysis to investigate the effects of the variables.

VALIDATING DISTRIBUTED REAL-TIME AND EMBEDDED SYSTEMS IN UNIFRAME. Shih-Hsi Liu, Dept. of Computer and Information Sciences, Univ. of Alabama at Birmingham, Birmingham AL 35294.

Distributed Real-Time and Embedded (DRE) Systems are widely applied to large scale and mission critical application domains such as avionics systems and medical imaging. Due to the rapid lifecycle and the complexity of design of the systems, there is an urgent demand to manage the development of robust, COTS, and effective software for vendors. The UniFrame project is a unified framework for seamless integration of distributed heterogeneous software components. This paper describes how the UniFrame approach is applicable to DRE software components. Our project starts from both functional and nonfunctional natural language specification. Then, leveraging the Two-Level Grammar (TLG) specification language and the Vienna Development Method (VDM), a formal methodology for developing DRE components and system code generation at the microarchitecture level is developed for UniFrame.

           TRADITIONAL SIMULATION FOR SOLVING FUZZY WEB QUEUING SYSTEM
         PROBLEMS. Xidong Zheng and Kevin Reilly, Dept. of Computer &
         Information Science, Univ. of Ala., Birmingham, AL 35294.
           In this project, we want to solve web queuing system problems
         for web servers. We first introduce the system models, and then
         apply fuzzy theory to our models. We have two methods to obtain
         required system performance variables, such as Utilization,
         Number (of requests) in the System, Throughput, and Response
         Time. Firstly, we use fuzzy optimization to do the computation.
         Then, we used crisp queuing systems simulation. By appropriate
         analysis of the results we can obtain fuzzy performance
         results. Applying crisp simulation to solve fuzzy optimization
         problems constitutes an area of research which has not been
         explored. At the detailed level, we use crisp Poisson
         distributed arrival rates and exponentially distributed service
         times as input, simulate the system for a period of time,
         collect results, and through analysis obtain the fuzzy system
         performance. We deal with both single and multiple server
         cases. Comparison of the results of these two methods
         (optimization and simulation) reveals that they match very
         well. So, in this first time demonstration of conventional
         simulation being used to solve fuzzy optimization problems,
         we offer a methodology that reduces computation time relative
         to the typically very time-consuming (direct fuzzy)
         optimizations. There is an added 'bonus' in the simulations
         in that it provides distributions of variables for "free" i.e.,
         as a natural outcome of the effort. The simulation methods,
         it would appear, can be applied to a whole host of other fuzzy
         optimization problems as well.

ADAPTIVE WEB BASED RESOURCE BROKER FOR THE GRID. Enis Afgan, Dept. of Computer and Information Sciences, Univ. of Ala. at Birmingham, Birmingham, AL 35294.

Today, as Grid Computing is becoming a reality, there is a need for managing and monitoring the available resources worldwide, as well as the need for conveying these resources to the everyday user. This paper describes a resource broker with its main function being to match the available resources to the user's requests. One way of doing this is for a user to manually find available resources, manually connect to the remote computer and, manually enter the commands to be executed. This method is superseded with the use of the resource broker that provides a uniform interface to access any of the available and appropriate resources in the name of the user. This paper will discuss the process of creating the resource broker as well as provide insight in how it connects and relates to underlying software and protocols. The resource broker runs on top of the Globus Toolkit. Therefore, it provides security and current information about the available resources and serves as a link to the diverse systems available in the Grid.

STATISTICAL DISTRIBUTION ANALYSIS FOR STORMWATER CONSTITUENTS. Alexander A. Maestre and Robert Pitt. Department of Civil and Environmental Engineering, University of Alabama, Tuscaloosa, AL 35487.

The University of Alabama and the Center for Watershed Protection were awarded an EPA Office of Water 104(b)3 grant in 2001 to collect and evaluate stormwater data from a representative number of NPDES (National Pollutant Discharge Elimination System) MS4 (municipal separate storm sewer system) stormwater permit holders. The initial version of this database, the National Stormwater Quality Database (NSQD, version 1.1) is currently being analyzed. It has been found that stormwater constituents can follow lognormal, normal or gamma distributions. When the lognormal distribution has been assumed only the mean and the standard deviation has been considered during the analysis. This paper evaluate the effect of include the third parameter in the fit if the lognormal distribution to stormwater constituents. Maximum likelihood and L-moments were used during the evaluation.

SCALABLE LINEAR ALGEBRA LIBRARY AND BENCHMARK. Yin Liu and Anthony Skjellum, Dept. of Computer and Information Sciences, Univ. of Ala. at Bhm., Birmingham, AL 35294.

According to Amdahl's law, the speedup in solving a problem due to more computing power detains at a certain limit no matter how many processors are further added. We're making a linear algebra library scalable in the sense that it can keep efficiently solving larger and larger sizes of problems along with the increasing number of processors/computers. Exploiting hierarchical memory reuse in basic matrix operations at algorithmic level is our way to essentially raise the computing performance in a single processor. Divide and conquer algorithms, outstanding in its cache-friendly feature, are our recipe for distributed memory computing environment, as well as the load balance issue in parallel computing. As a consequence of the algorithm exploration, we're generating benchmark programs to capture the performance profiles of given single/parallel computing systems. The various benchmark numbers can not only reveal information like the best/half performance and the maximum capable problem size for a given system, but also give insights on how to enhance hardware resources to improve the system computing capability.

SOURCE QUANTIFICATION OF INAPPROPRIATE DISCHARGES TO STORM DRAINAGE SYSTEMS. Veerabhadra Rao Karri and Dr. Robert Pitt, Department of Civil & Environmental Engineering, University of Alabama, Tuscaloosa, Alabama-35487.

Inappropriate discharges are non-stormwater discharges into a municipal separate storm sewer system (MS4) that are not covered by an existing National Pollutant Discharge Elimination System (NPDES) permit. Identifying and eliminating these inappropriate discharges to storm sewers is an important and cost-effective Best Management Practice (BMP) in a non-point source water pollution problem. Field screening procedures of dry weather flows which monitor for certain chemical and visual tracers that indicate potential sources are necessary to identify the sources. In this research a chemical mass balance model with Monte Carlo statistical simulation using Microsoft Visual Basic 6.0 was developed, which involves a statistical analysis of the tracers using their mass balance at the outfall. The numerical technique used in this model is to estimate mass contributions of different identified sources for a mixture water quality data set. Each mixture-water quality data set consisted of statistical parameters like mean, coefficient of variation and the type of distribution (Uniform, Normal, or Log-Normal) for an individual tracer from every source that is to be evaluated. Simulations were performed on this data set, to estimate these mass contributions. The input for this model is observed outfall concentrations for the sources selected for evaluation. The output of this model gives the most likely fraction of flow for each source type and shows the spread of the solutions. Good agreement was obtained between the prediction using the current model and the experimental results. Thus, using this model one can predict the main contributor to the inappropriate discharges in storm drainage systems at the outfall considered.

SEQUENCE ALIGNMENT ALGORITHM: MULTI-DIMENSION, MULTI-PROCESS. Dr. Robert Hyatt & Garret Cox, Dept. of Comp & Info Sciences, UAB, Birmingham, AI 35294

A sequence alignment is performed on a pair (or more) of protein, DNA, or RNA sequences by inserting gaps between the bases (or amino acids) of the sequences in such a way that the sequences match to some degree. An optimal alignment provides the best such matching. Once an optimal alignment is complete, it is trivial to calculate the similarity of the set of aligned sequences, or to determine what regions are present, missing, and changed in each member of the set. Such information can be used to determine the evolution of a sequence or to investigate a new sequence's coding behavior. Whereas it is trivial to calculate the optimal alignment of a pair of sequences using a dynamic programming algorithm, it is much more difficult to derive the optimal alignment of 3 or more sequences. Existing methods generally return only a nearly optimal alignment, or take unfeasibly long to compute. Distributing the work over a large cluster, however, can drastically reduce this run time.

PROCESS ENGINEERING AGENT FRAMEWORK USING ENTROPY MANAGEMENT. Rajani S. Sadasivam, Department of Electrical and Computer Engineering, University of Alabama at Birmingham, Birmingham, AL, [e-mail].

We must increase automation of the software development process because of the wicked nature of its problems, which are both difficult to define and difficult to solve. We must overcome classic software problems such as semantic gap to achieve this automation. Mature engineering disciplines have achieved great deal of success by developing products by integrating standardized components of a supply chain. The Web Services initiative is focused on developing standards and tools for developing such a supply chain of software components. Therefore, we propose that software development should just become a matter of correlating problems to system of Services. We argue that once we achieve this correlation we can develop systems by dynamically forming agencies, as Minsky noted. In our research, we are developing a process model for the Intelligent Correlation of Services Agent (ICSA) framework for providing systematic approach for automated selection of Services. We are using the ICSA framework for developing the virtual university system. We are also developing an Entropy Model (EM) for modeling the clusters of Services. As test cases, we have successful applied EM for modeling cross-disciplinary partnerships of organizations and selecting packet trunk groups in telephony systems.

Elders' Impressions of Ethicolegal Issues in Healthcare: A Qualitative Study. Gabriella H. Cooper and Ellen B. Buckner, University of Alabama School of Nursing, University of Alabama at Birmingham, Birmingham, AL 35294-1210

Ethicolegal issues in healthcare play a leading role in the satisfaction and comfort patients relate to hospital care. Ensuring that patients perceive their care in a positive light not only helps the hospital, but also aids geriatric patients by decreasing anxiety that may inhibit their healing. Ethicolegal issues are those topics that involve both ethical and legal principles, such as autonomy and informed consent. The issues addressed herein include advance directives, basic human needs, healthcare decision-making, and general nursing care. The purpose of this study was to investigate the perspectives of elders regarding geriatric care. Seniors ages 65 and older were interviewed with the Seniors' Perspectives Regarding Elder Care Issues (SPRECI) questionnaire to obtain qualitative responses. Eight open-ended questions developed by the investigator comprised the SPRECI. This questionnaire was reviewed for content validity by a CNS with cardiovascular and administrative experience, a faculty member with experience in geriatric nursing, a faculty member with experience in ethicolegal issues, and another with experience in research methods. Nine elders have been interviewed to date. Findings show that most elders commend hospital care as effectively provided to geriatric patients however these responses may reflect socially acceptable answers. To counter this tendency, future interviews are planned for community settings. This study was supported in part by a grant from the UAB Center for Aging.

SEDIMENT TRANSPORT IN GRASS SWALES. Yukio Nara and Robert E. Pitt, Dept. of Civil and Env. Eng., Univ. of Alabama, Tuscaloosa, AL 35487.

Grass swales are vegetated open channels through which collected stormwater runoff flows. They are often used as an alternative to concrete gutters for stormwater management because of their advantages of infiltration and filtration of stormwater. The objective of this research is to understand the effect different variables have on stormwater filtration and infiltration in grass swales. The variables to be studied are slope, grass type, time, depth of flow, and length of swales. To analyze these variables, solutions with known concentrations of sands and fine particles of aluminum oxide are pumped onto inclined grass swales and a variety of configurations of the study variables are measured. To date, 108 samples have been collected and analyzed for turbidity, total solids, and particle size distribution. Outdoor experiments will be conducted to extend the length of grass swale and to confirm the results obtained from indoor experiments.

     SURFACE RECONSTRUCTION USING GAMMA SHAPES, Ying Sun, Kenneth R.
     Sloan Dept. of Computer Information & Sciences, Univ. of Ala at
     Birmingham, Marietta E. Cameron Dept. of Computer Science
     Birmingham-Southern College.

        In this paper, we examine a modification to our
     three-dimensional point cloud reconstruction method, Gamma Shapes.
     Gamma Shapes is an extension to Alpha Shapes with the advantage
     that the Gamma Shapes method allows the automatic selection of
     local scaling factors. This presentation examines scaling methods
     based on simple approximation of the point sets medial axis. And we
     compared the results of different methods usng the information of
     normals of points.

TRACKING STORMWATER HEAVY METAL SPECIATION CHANGES FROM SOURCE AREAS TO RECEVING WATERS. Renee E. Morquecho and Robert E. Pitt, Dept. of Civil and Env. Eng., Univ. of AL, Tuscaloosa, AL 35487.

Many studies have identified heavy metals in urban runoff as a major contributor to the degradation of urban streams and rivers. Metals can occur as dissolved, colloidal and particulate-bound species, and their environmental health effects and control strategies can be quite different, based on their specific forms. Therefore, it is important to measure all forms of heavy metals, especially the particulate and filterable fractions, when determining their fate and effects. The objectives of the proposed research presented are to measure the changes in metal association/speciation from source areas to small urban receiving waters. The associations of heavy metals (along with some major constituents and nutrients) with different-sized particulates will be determined using cascade sieves and filters. Sequential extraction experiments will also be conducted to examine the treatability and other characteristics of the filterable (<0.45 [micro]m) portion of the heavy metals using Chelex-100 resin, UV-light exposure, and Anodic Stripping Voltammetry (ASV). It is hypothesized that metal associations will change as stormwater moves from source area(s) to receiving water(s). Those metals more associated with large particulates will be removed from the flow water due to natural sedimentation processes first, while the metal fractions associated with the smaller and/or dissolved particles will be the prevalent forms found in receiving water(s). Also, metals in the filterable fraction will be more likely to occur in free ionic form as opposed to being associated with organic/inorganic or colloidal complexes.

SOURCE VERIFICATION OF INAPPROPRIATE DISCHARGES TO STORM DRAINAGE SYSTEMS. Soumya Chaturvedula and Robert E. Pitt, Dept. of Civil and Env. Eng., Univ. of Alabama, Tuscaloosa, AL 35487.

Investigation of non-stormwater discharges into storm drainage proceeds along a hierarchy of procedures. Exploratory techniques involve an extensive mapping effort to identify the locations of all outfalls for sampling. This is followed by the screening analyses at the outfalls which include sampling at repeated intervals at the outfalls, to measure chemical tracers which would identify the general categories of non-stormwater flows. Using a Flow chart method and a modeling package, the most probable source of inappropriate discharge into the storm drain system can be identified. The final verification process entails the identification of problem outfalls and field investigation of these problem outfalls by surveying the contributing watershed. The above methodology has been employed at Tuscaloosa, Alabama, to study the sources of illicit discharges into the Cribbs Mill Creek. The field verification of the results is being completed.

THE NATURE OF THE IMPERVIOUSNESS IN URBAN WATERSHEDS. Celina Micu and Robert E. Pitt, Dept. of Civil and Env. Eng., Univ. of Alabama, Tuscaloosa, Al 35487-0205.

Local development characteristics are the most important elements that affect stormwater quality and quantity. In this study, 136 neighborhoods were surveyed to determine the critical development characteristics representing 18 major land use areas located in Little Shades Creek Watershed, Birmingham, AL.

Field data were collected during an earlier research project to determine the sources of urban runoff and the associated pollutants and to examine the alternative controls in this rapidly developing area. This research uses the same field data to calibrate WinSLAMM (Source Loading and Management Model), to determine the runoff quantity generated by the 136 sites and by the 18 major land uses. Statistical analyses will be then conducted at several levels to determine how land development variability affects the calculated yields. WinSLAMM runoff data will be validated against the National Stormwater Quality Database (NSQD, version 1.1) for Jefferson County, AL, which is part of the University of Alabama, Department of Civil and Environmental Engineering Research and is used as a benchmark for comparison with locally collected data.

STORMWATER TREATMENT AT CRITICAL SOURCE AREAS USING UPFLOW FILTERS. Uday Khambhammettu and Robert E. Pitt, Dept. of Civil and Env. Eng., Univ. of Ala., Tuscaloosa, AL 35487.

A number of studies have linked specific pollutants in stormwater runoff with critical sources such as paved parking lots and gas stations. Runoff from these areas has been observed to be contaminated with concentrations of many critical pollutants especially heavy metals and PAHs. These paved areas are usually found to contribute more of the pollutant loadings than their areas would normally indicate. Numerous manufacturers have developed proprietary devices to treat stormwater runoff at these critical source areas. These devices have been designed to treat a variety of stormwater pollutants such as solids, metals, oil and grease, nutrients and bacteria. Very few have been designed to treat a broad list of these pollutants in a single device. This research is focusing on the demonstration of the effectiveness of an upflow filtration setup for the treatment of stormwater runoff at critical source areas and will be tested for the removal of many stormwater pollutants. This project will also look into the head loss and associated treatment rate change during filtration.

WEBRUN: A UNIFIED PLATFORM SUPPORTING GRID COMPUTING ENVIRONMENTS. Zhijie Guan, Yin Liu, Vijay Velusamy, and Purushotham V. Bangalore, Dept. of Computer and Information Sciences, Univ. of Ala. at Bhm., Birmingham, AL 35294.

The evolution of the Grid has influenced the ever-growing need to have a uniform platform to access various kinds of programs, hiding as much heterogeneity among programs as possible. Current research adopts a hierarchical architecture that provides users a programmatic interface to access the web services on remote machines hiding all the underlying functionality within the actual running environment. However, programmatic interfaces are insufficient to support the requirements of the user. A friendly web interface may be more beneficial to users who do not wish to concentrate their efforts on programming. This paper presents the development of a unified platform, namely WebRun, which provides users with both a web interface and a programmatic interface to command-line programs located on remote computing resources. The web interface, a browser/JSP implementation of the user/system ports, supports the finding, starting, controlling, and utilization of command-line application programs. The programmatic interface enabled by WASP client/service model provides users a Java interface to access the programs stored on the remote resources that are otherwise not accessible.

GENPARSE: AN EVOLUTIONARY APPROACH TO CONTEXT-FREE GRAMMAR INDUCTION. Faizan Javed, Department of Computer & Information Sciences, University of Alabama at Birmingham, Birmingham, AL, 35294-1170.

While grammar inference is used in areas like natural language acquisition and pattern recognition, its application to the programming language domain has been limited. We propose facilitating domain-specific language development, and building renovation tools for legacy systems, as new application domains for grammar inference. Many well established techniques exist for inferring regular languages however inferring context-free grammars (CFG's), which are more expressive and powerful than regular languages, is still an open research problem. Our current research focus is to investigate the genetic programming (GP) paradigm for inferring CFG's. Preliminary work using grammar-specific heuristic operators in tandem with non-random construction of the initial grammar population resulted in successful induction of small grammars. We extend that work and propose the use of syntax graphs and derivation trees to be able to infer a more comprehensive set of CFG's. Our project aim is to research on methodologies of CFG induction under various constraints (use of positive or negative samples, or both) limited not only to the GP model of computation, but also open to investigating other existing models of grammar inference like the MDL approach, in the hope of being able to infer real world grammars.

QUALITY OF SERVICE PREDICTION IN COMPONENT-BASED DISTRIBUTED SYSTEMS. Changlin Sun, Dept. of Computer and Information Sciences, University of Alabama at Birmingham, Birmingham, AL 35294

Developing distributed software systems by assembling commercial-off-the-shelf (COTS) software components has well-known advantages of reducing the development cost and time to market. However, the component-based software development poses some new problems and challenges due to its use of a different underlying process from traditional software development approaches. One of the major challenges is the compositional reasoning about the system Quality of Service (QoS). Compositional reasoning about system QoS is necessary in (COTS) component-based software development. It allows the static validation of the QoS of a software system based on the QoS assured by individual software components without a need to examine or otherwise analyze the implementations of those individual components. Compositional reasoning about QoS makes it possible that the component developers can develop and certify components independently and system designers can build a system and predict the system QoS with the given QoS of its components. This shifts the burden of QoS assurance from the system developer to the component developer. Software component developers provide rich information for predictable assembly (design for reuse) and the system developers reuse the QoS information provided by components during system integration (design with reuse). We examine the categories of software artifacts that are needed to facilitate the QoS prediction in developing component-based distributed systems. This research was supported by US Office of Naval Research under award number N00014-01-1-0746.

OVERLAPPING OF COMPUTATION, COMMUNICATION, AND I/O FOR HIGH PERFORMANCE PARALLEL COMPUTING. Vijay Velusamy, Dept. of Computer and Information Sciences, University of Alabama, Birmingham, AL 35294.

There has been growing interest in large-scale cluster/grid computing systems having thousands of processors, for computational simulations, including but not limited to bioinformatics, and earth simulations. Although there has been significant research on novel algorithms, computing environments, high performance interconnects for such systems, and overlapping of computation and communication, there has been little interest in studying the issues and benefits of overlapping I/O. This study analyzes some of the emerging technologies for storage networks such as InfiniBand and Remote Dynamic Memory Architecture (RDMA). These technologies support CPU-offload for their communication protocols, allowing the CPU's to be used more effectively for computations, rather than for communication, or Input/Output (I/O). This is beneficial for overlapping computation with communication and I/O, and for effective resource utilization. The study shows that overlapping of computation, communication, and I/O will allow better application performance for large-scale systems. It is even more interesting to analyze similar issues relating to storage access in grid computing environments. Typical applications of this would include efficient check-pointing for rollback recovery to create fault-tolerant environments, and predictive reading of data for real-time visualization of biomedical simulations on a visualization-wall.