Computer Science Graduate Projects and Theses

An Exploratory System for Collaborative Decision-Making in Community Planning

Aaron Dale Wells — Mon, 20 May 2013 14:57:51 PDT

Community planning problems differ from those of science, technology, and mathematics as they are not solvable with logical-empiricism. Their solutions are influenced by technology, politics, style, economics, as well as the personalities and experience of those individuals collaborating on the solution. Obtaining cooperation of the stakeholders to implement community planning solutions can be cumbersome or simply cause failure in the implementation of plans. Yet, if the stakeholders had a real handle on the cost and benefits the literature suggests that cooperation can evolve.

In this project, we explore building a reliable cost and benefit model for a set of input parameters that may allow a collaborative solution to emerge more easily. Furthermore we hypothesize that in the decision process there is a tipping point between costs related to a decision and its benefits.

In order to test the hypotheses, we have designed and tested a software framework with focus groups that included locally elected officials, economic development specialist, planners, and citizens. The software framework allowed the stakeholders to explore an interactive cost-benefit model, and researchers to collect those interactions and visualize them in real-time. The software framework developed for the study, its set up, and findings based on a focus group study are discussed.

The software framework developed for this study and the included analysis tool provided were shown to be effective in identifying the \tipping-point" moments in the group dynamics.

Parallel Copying Tools for Distributed File Systems

Kevin Matthew Nuss — Fri, 22 Mar 2013 09:02:08 PDT

Parallel distributed files systems are increasingly being used on clusters to allow greater throughput of data to the many compute nodes. They are also an effective way to store massive amounts of data. However, using the standard core utility cp does not make good use of the potential parallelism of the file systems. Using multiple cp commands has inherent problems too.

Two utilities were created to help recursively copy directories containing large amounts of data on parallel distributed file systems. One of the test data sets contains very many files, and the other contains large files. One utility is a C program that submits a single job on a user specified number of nodes. The work of copying the files is dynamically distributed among those nodes using MPI communications. Multiple threads are used to traverse the directories. Speedups of 9.57 and 7.36 were attained for the many files set and the large files set, respectively. A second utility is written in Java. It also uses multiple threads to traverse the directories, but it performs the copying by creating Bash scripts and submitting them to the job scheduler. The work is balanced among those scripts and the number of jobs is specified by the user. It reached speedups of 3.67 and 7.32 for the same two data sets. Both utilities can also be used to track the progress of the jobs they have submitted.

Object Oriented Implementation of the Parallel Toolkit Library

Sandhya Vinnakota — Mon, 07 Jan 2013 10:18:05 PST

With manufacturing efficiencies and technological innovation the computing power of commodity machines has been increasing accompanied by decreasing costs. With the very favorable price/performance ratio the computing community has shifted from monolithic machines to networked machines.

This has created the need for software to manage the parallelism of the network. One such work has been the Parallel Toolkit Library. The Parallel Toolkit Library provides support for common design functionalities used throughout parallel programs.

This work extends the PTK C library for C++ parallel programs. The motivation for the current project stems from the need to let parallel programs reap the benefits of a library with an object oriented programming approach. This also fits well with the introduction of C++ bindings in MPI. The library has been structured on object-oriented concepts. The functionality of the PTK-C has been encapsulated into various classes. Individual functionalities have also been split into multiple classes leading to modularity and reusability of code.

Template programming has been used to ensure type safety. The testing results are consistent with expectations in that the PTK-C++ is very much comparable to the PTK-C in terms of performance. In most cases, it would be more efficient to use the toolkit than to rewrite the code to recreate the efficiencies already present in the library.

Hadoop and Hive as Scalable Alternatives to RDBMS: A Case Study

Marissa Rae Hollingsworth — Thu, 23 Aug 2012 11:06:55 PDT

While high-performance, cost-effective data management solutions, such as Hadoop, exist for Big Data analysis, small and medium businesses with moderate-sized data sets would also like to implement low budget data management systems that will perform well on existing data and scale as the amount of accumulated data increases. Parallel database management systems may provide a high-performance solution, but are expensive and complex to implement. The purpose of this project was to compare the scalability of open-source relational database management systems and distributed data management systems for small and medium data sets. To make this comparison, a business intelligence case study was investigated using three data management solutions: MySQL, Hadoop MapReduce, and Hive. This experiment involved a payment history analysis which considers customer, account, and transaction data for predictive analytics. Experiments were executed on data sets ranging from 200MB to 10GB. The results show that the single server MySQL solution performs best for trial sizes ranging from 200MB to 1GB, but does not scale well beyond that. MapReduce outperforms MySQL on data sets larger than 1GB and Hive outperforms MySQL on sets larger than 2GB. This demonstrates MapReduce and Hive as viable techniques for small and medium businesses who want to implement scalable data management techniques.

An Efficient Time-Bound Hierarchical Key Assignment Scheme with a New Merge Function: A Performance Study

Rajasree Shyam — Thu, 05 Nov 2009 17:10:26 PST

The advent of digital age has resulted in more television consumers switching to Digital TV with considerable improvement in image quality and ease-of-use. Consumers are able to select and view television programs and channels of choice by using a pay-per-use model or streaming video from their computer terminals. In all these use cases, the media provider requires a means by which they can restrict the consumers from watching selected programs for a pre-approved temporal interval. The consumer needs to be prevented access to certain pay-per-use channels and programs upon expiry of this pre-approved access. This necessitates the media provider to have a way to generate and assign time-bound secure access keys which could be granted and removed easily. In conventional key assignment schemes, one has to renew the keys periodically and redistribute the keys to the users accordingly. To allow a user to access all the authorized data over some temporal window, this straightforward implementation requires him/her to keep a lot of keys which is very inefficient. In contrast to conventional schemes, a time-bound hierarchical key assignment scheme updates the keys periodically according to the class hierarchy and an entity only keeps a small amount of information for deriving all his entitled keys. Wang and Laih (WL) proposed a scheme with a concept of merging, which provides a systematic way to solve the problem. Yeh-Shyam (YS) scheme has improved on the WL scheme and has theoretically shown polynomial improvement in both memory and performance requirement. In this project, we will compare and contrast these secure key generation techniques and provide comparative analytical results.