Steve Goddard
Department of Computer Science and Engineering
University of Nebraska-Lincoln
September 2003
This section briefly describes my funded research projects since joining the faculty at UNL. The grants and contracts for which I am the PI have or will provide nearly $2.7M in external funding for research projects at UNL. I am also a Co-Investigator of a $2.5M NSF EPSCoR grant and a Co-PI on a NSF grant that provides $270,000 in scholarships to qualified computer science, engineering, or mathematics students—though that grant is described in Folder 4.
The research grants and contracts described here have provided funds for traditional expenditures: support for dozens of graduate students, full summer support for me, and tens of thousands of dollars in new equipment for my research lab and the department. However, the level of funding has also helped to provide a new research infrastructure that will help me and my colleagues in CSE reach new levels of research quality and productivity. My grants and contracts are currently (or will be) providing funds for six personnel funded on soft research money:
1. Marilyn Augustyn, Staff Assistant (direct report to me)
2. Ian Cottingham , Research Programmer (direct report to me),
3. Bill Waltman, Research Associate Professor (originally hired by a Nebraska Research Initiative funded project in which I am not a PI or Co-PI)
4. Xueming Wu, Visiting Research Scholar (direct report to me)
5. You Jinsheng, Post-Doctorate Researcher (Climatology)
6. Shashank Mehta, Visiting Research Associate Professor
In addition to the UNL employees identified above, research and support contracts have been established with Kacy Steiner (Technical Writing Tutor); National Soil Survey Center, USDA-Natural Resources Conservation Service; USGS EROS Data Center; and GIS Workshop, Inc. to provide services to my research projects.
The remainder of this section provides a brief summary of my UNL research projects, a full list of funding sources since joining the faculty at UNL, and an abstract of each funded proposal or contract; abstracts are provided only for grants or contracts over $10,000. (I received over $1.2M in contracts, as a consultant, before joining the faculty at UNL, but that work is not described here.)
· Grants [1, 4] have provided funding for the National Agricultural Decision Support System (NADSS), http://nadds.unl.edu. This is a broad inter-disciplinary research project at the University of Nebraska in the area of distributed geospatial decision support systems for agriculture. It includes CSE faculty members and scientists from the School of Natural Resources. This effort is expanding and at least three more proposals will be submitted this year to provide additional support for this project.
· Grants [2, 3, 9, 10] have provided funding for my research in real-time systems. As described in the summary of my research activities, these projects have focused on new flexible, dynamic task models and scheduling theory that guarantees temporal requirements will be met under specified conditions.
· Grants [5, 8] funded my research in high assurance networked-clustered servers and an e-commerce broker architecture, which uses some of the same technology developed for networked clustered servers. The work funded by an industrial partner [8] began as a technology transfer project, but resulted in new research on fault-tolerant firewall sandwiches.
· Grants [6, 7, 11] supported undergraduate research activities in distributed geospatial support systems and the application of computer science to the field of climatology.
Grants and Contracts:
[1] “Risk Assessment and Exposure Analysis on the Agricultural Landscape,” U.S. Department of Agriculture Federal Crop Insurance Corporation, Risk Management Agency, 10/1/02–9/30/04, $1,326,623. PI: S. Goddard (Co-PIs: W. Waltman, J. Deogun, M. Hayes, K. Hubbard, D. Jose, S. Reichenbach, M. Svoboda, D. Wilhite).
[2] “Collaborative Research: Rate-Based Resource Allocation Methods for Real-Time Embedded Systems,” National Science Foundation, 9/1/02–8/31/05, $120,000. PI: S. Goddard.
[3] “Improving Aviation Safety through Real-Time Spatio-Temporal Resource Allocation,” NASA Nebraska Space Grant & EPSCoR, 11/1/01-10/31/02, $13,193. PI: S. Goddard.
[4] “DIGITAL GOVERNMENT: A Geospatial Decision Support System for Drought Risk Management,” National Science Foundation, 7/1/01–6/30/04, $1,007,914. PI: S. Goddard (Co-PIs: J. Deogun, M. Hayes, K. Hubbard, S. Reichenbach, P. Revesz, W. Waltman, D. Wilhite).
[5] “Next Generation Enterprise Resource Planning Systems,” NSF EPSCoR, 5/1/01-4/30/04, $2,575,680 ($1,214,937 from NSF, $1,360,743 from UNL matching funds). Co-Investigator: S. Goddard (PI: S. Henninger; Co-PI: F. Choobineh; Many other Co-Investigators).
[6] “Undergraduate Creative Activity and Research Experience (UCARE),” University of Nebraska-Lincoln, 6/1/02–5/31/03, $2,400. Faculty Advisor: S. Goddard.
[7] “Undergraduate Creative Activity and Research Experience (UCARE),” University of Nebraska-Lincoln, 8/1/01–5/31/02, $2,400. Faculty Advisor: S. Goddard.
[8] “Intelligent Server Farm Management,” Flextel S.p.A., 10/1/00–8/31/01. $144,316. PI: S. Goddard.
[9] “Assured Quality of Service Resource Allocation Models,” University of Nebraska-Lincoln Center for Communication and Information Science, 5/1/01–6/31/01, $7,380. PI: S. Goddard.
[10] “Schedulability Analysis with UML-RT Models,” University of Nebraska-Lincoln Center for Communication and Information Science, 5/1/00–6/31/00, $3,000. PI: S. Goddard.
[11] “Undergraduate Creative Activity and Research Experience (UCARE),” University of Nebraska-Lincoln, 8/1/00– 5/31/02, $2,000. Faculty Advisor: S. Goddard.
RISK ASSESSMENT AND EXPOSURE ANALYSIS ON THE AGRICULTURAL LANDSCAPE
A Holistic Approach to Spatio-Temporal Models and
Tools for Agricultural Risk Assessment and Exposure Analysis
Principal
Investigator:
Steve Goddard
Co-Principal
Investigators:
W.J. Waltman, Jitender Deogun, Michael J. Hayes, Kenneth G. Hubbard, H. Douglas
Jose, Stephen Reichenbach, and Mark D. Svoboda, Donald A. Wilhite, University
of Nebraska-Lincoln (UNL)
Co-Investigators: Norman Bliss, EROS Data
Center; Sioux Falls, SD: Sheri K. Harms, University of Nebraska-Kearney; and
J.S. Peake, University of Nebraska-Omaha; Ray Sinclair and Sharon Waltman, USDA
Natural Resources Conservation Service, National Soil Survey Center, Lincoln,
NE; and Marcus Tooze, GIS Workshop, Lincoln, NE.
Amount: $1,326,623
USDA agencies face the difficult task of integrating multiple geospatial databases to assess natural hazards on the agricultural landscape. Timely, accurate climate information and forecasts coupled with soil and land use characteristics can potentially reduce production risks among agricultural producers and provide new insights into the spatial and temporal characteristics of natural hazards. Within USDA there are many natural resource inventories that describe aspects of environmental risks and the characteristics of agricultural lands; however there is a lack of analytical tools designed to operate within and between these databases to evaluate patterns, trends and the spatial context of climate hazards in comparison with soils and the agricultural infrastructure.
Our proposal
describes a holistic approach to the spatial analysis of climate and soil
landscape hazards. It incorporates
traditional geostatistical methods, geographic information systems (GIS) and
data mining techniques in a framework that builds associations between oceanic
parameters and major crop production regions (agroecozones) along with
agricultural practices, climate and soils characteristics.
Specific outcomes are to develop a suite of web-based, (industry standard) GIS tools and applications that: 1) Calculate and map frequencies (probabilities), durations, and intensities of drought and flood events and their effect on crop and animal production; 2) Develop a new generation of FCI-33 maps (across multiple scales; farm to regional levels) through automated mapping routines (GIS framework) that delineate regions of high/low risk landscapes, reflecting multiple hazards to crops; 3) Develop a framework of agroecozones that define landscapes of similar climate and soil characteristics, leading to delineation of zones or regions of similar agronomic or horticultural behavior (crop suitability or adaptation). The use of agroecozones will improve the delineation of high/low risk landscapes in the new generation of FCI-33 maps; 4) Generate a near-real time exposure analysis using climate indicators or weather events superimposed on multiple themes from the Census of Agriculture (2002) and National Agricultural Statistics Service’s historical data to derive maps of potential impacts on the agricultural infrastructure; 5) Calculate the probabilities and extremes of frost dates and a phenology calendar based upon heat units (growing degree-days) in a planting date guide tool; 6) Help enhance the quality of climate databases and lead to an automated procedure to generate missing values using surrounding weather stations; 7) Translate drought indices into potential fire danger indicators (Keetch-Byram Drought Index, SPI, and PDSI, and Newhall) applicable to the rangeland ecosystems and grazing lands in the western Great Plains; and 8) Contribute to a series of risk education workshops to assist RMA staff in training and tailoring the tools to their program responsibilities and extensions to other USDA agencies.
Collaborative Research:
Rate-Based Resource Allocation Methods for Real-Time Embedded Systems
Principal Investigator: Steve Goddard
Funded by the National Science Foundation
Amount: $120,000
Date: 9/1/02–8/31/05
We propose an investigation into the use of rate-based resource allocation methods for constructing embedded systems with real-time execution constraints. The investigation will have both an algorithm design and analysis component as well as an implementation and use component. In the design/analysis component we will develop a framework for analyzing and comparing different rate-based schedulers from the literature. Specifically, we consider a taxonomy of rate-based resource allocation consisting of proportional share scheduling, polling server-based scheduling, and rate-based extensions to classical Liu and Layland scheduling. The goal here will be to relate the different scheduling models and abstractions to one another to understand the fundamental principles of rate-based resource allocation such as the form and nature of timing guarantees and the algorithmic overhead. In addition we will extend the existing theory of rate-based resource allocation to deal with practical considerations such as preemption constraints.
The implementation and use component of this research will consider the reduction to practice of rate-based resource allocation in operating system kernels and applications. The goal here will be to assess the fit between the formal task model used to develop a particular allocation algorithm and implementation constraints that arise in practice. To fully explore the utility of rate-based resource allocation, three scheduling problems will be considered: application-level scheduling (i.e., scheduling of user programs or application threads), scheduling the execution of system calls made by applications (“top-half” operating system-level scheduling), and scheduling asynchronous events generated by devices (“bottom-half” operating system-level scheduling). This treatment is motivated by the logical structure of traditional, monolithic real-time (and general purpose) operating systems and kernels with hardware enforced protection boundaries. Moreover, this simple taxonomy of scheduling problems emphasizes that in real systems one must consider issues of intra-kernel resource allocation and scheduling (e.g., buffer management) as well as application task scheduling. A sub-goal will be to prove the thesis that that “one size does not fit all” — one rate-based resource allocation scheme does not suffice for all scheduling problems that arise within the layers of a real system. For realistic environments that are likely to be encountered in practice, the best results will be achieved by employing different rate-based allocation schemes at different levels in the operating system. The thesis will be shown by constructing small-memory-footprint versions of FreeBSD and Linux that employ different forms of rate-based scheduling and resource allocation at different levels in the system. We will work with external researchers from industry to use two commercial embedded systems to evaluate our rate-based allocation schemes and their implementation.
Collaborative Research:
Rate-Based Resource Allocation Methods for Real-Time Embedded Systems
Principal Investigator: Steve Goddard
Funded by the NASA Nebraska Space Grant & EPSCoR Seed Grant
Amount: $13,193
Date: 11/1/01-10/31/02
The primary
objective of this project is to improve aviation safety by developing
new real-time resource allocation algorithms for avionics systems such as radar
tracking and signal processing, which execute in dynamic spatial environments
with hard temporal requirements. Other prominent examples of such systems
include autonomous vehicles and semi-autonomous mobile robots. We refer to this special class of adaptive
real-time systems as real-time spatio-temporal systems.
The challenge in engineering such systems is allocating resources so that critical real-time components are guaranteed to complete all processing within given time constraints. These guarantees must be met even when the duration and execution frequency of the components changes dynamically. We are developing new real-time resource allocation models that explicitly support both spatial and temporal properties of the application environment to efficiently use available computing resources.
DIGITAL GOVERNMENT: A Geospatial Decision Support
Principal Investigator: Steve Goddard
Co-Principal Investigators: Stephen E. Reichenbach, Jitender Deogun,
Michael J. Hayes, Kenneth G. Hubbard, Peter Z. Revesz, Mark D. Svoboda, William
J. Waltman, and Donald A. Wilhite
Funded by the National Science Foundation
Amount: $1,007,914
Date: 7/1/01–6/30/04
This research
develops and integrates new information technologies for improved government
services in the U.S. Department of Agriculture (USDA) Risk Management Agency
(RMA). The mission of the RMA is to strengthen the safety net for
agricultural producers (farmers) through sound risk management programs and
education. Risk management in agriculture is critically important to
producers, their communities, and the nation’s economy, but it involves many
complex problems. New information technologies hold great promise for dealing
with the complexities of risk management and offer an opportunity for improved
government services. This project develops innovative approaches in constraint
databases, data mining, information retrieval, and geospatial information
analyses that are integrated using advances in cluster computing to support
faster and better risk management.
Our objective is to improve, through research
and advanced development, the RMA’s risk assessment services in three important
ways:
Accomplishing
these objectives requires basic computer and information science research in
constraint databases, data mining, information retrieval, geospatial
information analyses, and network clustered servers. The project will integrate
ongoing research in these areas in advanced development of a Geospatial
Decision Support System (GDSS) for drought risk management.
Next Generation Enterprise
Resource Planning Systems
Principal Investigators: S. Henninger and F. Choobineh
Investigators: Steve Goddard, Sebastian Elbaum, David Olson, and many others.
Funded by the National Science Foundation EPSCoR Program
Amount: $2,575,680 ($1,225,312 from NSF EPSCoR, $1,350,368 from UNL matching funds).
Date: 5/1/01-4/30/04
Enterprise resource planning (ERP) systems are revolutionizing the way organizations streamline business processes, share information within and across businesses, and conduct electronic business. Though promoted as a “silver bullet” in solving organizational IT integration problems, current ERP packages do have major shortcomings. First, significant amount of resources and time needs to be invested in customizing these ERP systems. Despite the investment, less than satisfactory outcome or outright implementation failure is not uncommon. Existing ERP systems have proven difficult to change and extend. They do not integrate well with legacy systems or support the integration of ERP systems across different organizations.
There is an urgent need to study the problems of current ERP systems and propose a next generation ERP architecture (coined here as ERP-II) that is viable in the e-commerce and Internet era. Trends indicate that next-generation ERP will be valued less on initial implementations and more on creating an infrastructure or “backbone” that consolidates databases, transaction services and other business processes [Radding 1999], enabling development of business applications that would otherwise be cost prohibitive or technically infeasible.
The goal of this project is to develop an infrastructure for researching the foundations of next-generation ERP systems. Distributed, component-based technology will be developed to address critical ERP/ERP-II issues and create an infrastructure that is more flexible, robust, and responsive to user and business needs. Features of the new ERP-II architecture will include: (i) scalability, (ii) availability, (iii) quality of service, (iv) maintainability, (v) security, and (vi) data sharing. To achieve this goal, the following infrastructure-building activities will be carried out:
(i) Creation of a Design Environment that supports component-based application development.
(ii) Creation of a Scalable Distributed Systems Layer using Enterprise Servers.
Development of this infrastructure will create a potential for the University of Nebraska to become a national center for ERP-related research. The multidisciplinary nature of our team combined with our close association to the ERP industry (through JD Edwards & Company and the ASP project group at UNL that is implementing SAP in the University system) will be instrumental in the successful definition and refinement of ERP-II strategies and technical architectures that can meet future needs in the industry.
Intelligent Server Farm
Management
Principal Investigator: Steve Goddard
Funded by Flextel S.p.A.
Amount: $144,316
Date: 10/1/00–8/31/01
The overall objective for this project is to develop a Flextel Intelligent Server Farm Manager (FISFM). FISFM will manage multiple network servers from a single chassis to support e-business applications with Quality of Service (QoS). The collection of network servers appears to be a single, high performance server to clients. It is actually a cluster of servers managed by an intelligent dispatching entity, called the Dispatcher, which assigns client requests to servers for processing using specific load sharing algorithms while providing a single server interface to clients over the Internet.
The project is organized as a collection of software orders that evolves the research prototype dispatchers LSNAT and LSMAC, which were developed in the University of Nebraska-Lincoln Advanced Networking and Distributed Experimental Systems (ANDES) Laboratory, into a fully functional, scalable, and hierarchical commercial dispatcher for the FISFM.
The first software (SW) order is composed by a set of modules running at user or kernel level under Linux OS (2.2.14). The SW is intended to process and dispatch client requests towards a generic server pool, with the goal of virtualizing the pool so that clients are unaware of how many servers compose it, and balancing the load of requests on servers according to a set of policies (e.g., Round Robin). The SW must accept a client request arriving through a standard Ethernet interface driver, parse it, select an appropriate set of eligible servers where the request can be sent and dispatch the request through a standard Ethernet interface diver. After that, the selection of one among the set of eligible server is performed with an algorithm that balances the load among them, according to some defined policies. The parsing of the request is limited to Layer 4. Supported services should be HTTP and FTP. The SW is also in charge of handling server responses, performing IP address translation as appropriate.
The final product is expected to provide the following features:
· Load sharing of client requests towards a server pool
· Support of client affinity for FTP
· Support of URL analysis
· Support of request scheduling disciplines different from FIFO
· Software configurable by a set of user level commands or a configuration file.
· Support of client affinity for SSL
· Support of IPB interface driver
· Support of client affinity for HTTP through cookie analysis
· Support of logging of profiling information
· Support of SSL3 and telnet
· Support of multiple dispatching modules
· Support of L4/2 accompanying load sharing modules
· Support of fault tolerance and state reconstruction
· Export of profiling information about bandwidth requirements associated with server responses