Summer 2012 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Prior Year TRUST
-- 2011 REU
|
The TRUST Research Experiences for Undergraduates (TRUST REU) promotes access to graduate education
for undergraduate students who have been educationally or economically disadvantaged or who may not have
exposure to the academic environment of a research university.
The program offers a research experience in cyber security and trustworthy systems and other TRUST Center related topics. The TRUST REU is an eight-week program that will commence on Monday, June 4, 2012 and end on Friday, July 27, 2012. Dorm move-in will be Sunday, June 3, 2012 and dorm move-out will be Saturday, July 28, 2012. The goal of the TRUST REU program is to increase the level of diversity among students entering graduate programs by providing research opportunities for undergraduates. Program objectives are (1) to provide students with preparation to become research scholars, (2) to stimulate serious consideration of graduate study, and (3) to increase the number of underservd students successfully enroll in graduate school. The TRUST REU provides students an opportunity to conduct research under the direction and guidance of a TRUST Center faculty advisor and/or graduate student mentor. Special consideration will be given to applicants who have shown potential for success, but may have had limited access to graduate research or other academic opportunities. The program strongly encourages applications from undergraduates who have been educationally or economically disadvantaged and who would benefit from exposure to the environment of a research university. All applicants will be considered without regard to race or gender. For more information on the TRUST REU, including information on applying, please click HERE |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Summer 2012 TRUST REU Research Projects
|
| Research Topic | Managing Exhaustion: An Analysis of Regional Variations in IPv4 Allocation Strategies |
|---|---|
| Sponsoring TRUST Faculty |
Prof. John Chuang, School of Information, University of California, Berkeley |
| Description | The amount of free IPv4 address space is currently running out, resulting in a coordinated push to transition to
IPv6, which will offer substantially more address space. However, the shortage of IPv4 addresses varies quite radically by
region, due to both historical and market conditions. For instance, North America holds a substantial chunk of so-called legacy
address space, allocated in the early days of the Internet, while emerging economies in Asia are rapidly exhausting their
available address space. Accordingly, policies for handling the exhaustion of IPv4 address space have also varied amongst the
five Regional Internet Registries responsible for handling the allocation of critical Internet resources in their regions of the
world. In the face of scarcity, recurrent debates about IP address allocations have gained in prominence: whether IP addresses
are to be viewed as resources to be held in common, or property to be exchanged in markets. This has crucial implications for
the nature of trust in ICANN and the Regional Internet Registries, the institutions of Internet governance - are they to be
viewed as guarantors of property rights in IP addresses, or shepherds of common pool resources? The relationship between
institutional structures, policy design and the trustworthiness of large scale computing systems is a core concern of this
research project.
In this project, we will examine variations amongst regional IPv4 allocation policies, with particular focus on how these policies have changed in the last few years, as IPv4 exhaustion became a more pressing concern. This will involve qualitative research into the language and implications of policies, analysis of conversations on policy mailing lists that led to particular policy decisions, and a quantitative examination of actual IPv4 allocations as a consequence of these policy variations. This research project will provide important insights into the functioning of the global Internet governance regime, over a crucial period of reaction to IPv4 address space exhaustion. |
| Campus Location | University of California, Berkeley |
| Research Topic | On the Usage of Brainwave Data in Safety and Security Applications |
|---|---|
| Sponsoring TRUST Faculty |
Prof. Shankar Sastry, Electrical Engineering and Computer Sciences, University of California, Berkeley |
| Description | This project focuses on neuro-based safety bounds for human-machine interactions. We want to use brainwave signals to help us
determine whether subjects are able and qualified to perform complex tasks, such as safety-critical operational decisions about
network control systems or electric dispatch. This research relates to a broader agenda of investigating the possibilities to
utilize brainwave data in safety and security applications.
Experiments: Conduct studies with human subjects, including designing experiment protocols and collecting data from human subjects using brain wave equipment and controls to measure the subject’s vital signs. Most likely, we will use the NeuroSky MindSet headset. Statistical Data Analysis: Learn to analyze the experimental data collected via the experiments and perform statistical analysis of the data. Working with Literature: Review the current state of art of brain wave data applications in various domains by researching the literature and summarize the existing studies about the possible uses of brainwave data including limitations and constraining problems of brain wave usability. An emphasis will be on applications targeting non-traditional, non-diagnostic (non-medical) purposes (e.g., games and neuromarketing) and special attention will be paid to privacy issues and the privacy repercussions of using brain wave signals. Equipment Testing: Design and perform a series of tests to collect comparative performance statistics (signal quality and precision) of existing, commercially-available brainwave hardware. |
| Campus Location | University of California, Berkeley |
| Research Topic | A Study of Captcha Security |
|---|---|
| Sponsoring TRUST Faculty |
Prof. Dawn Song, Electrical Engineering and Computer Sciences, University of California, Berkeley |
| Description | Human interaction tests (better known as Captchas) are used to distinguish human activity from automated activity on websites. The main goal is to prevent various forms of online abuse like spam messages, fake registrations, automatic login attempts, poll rigging. Captchas are traditionally in the form of a visual or audio test in which the users are asked to recognize a sequence of distorted alphanumeric characters. Such recognition is assumed to be hard for machines while being relatively easy for humans. Studying Captcha security, e.g., breaking captchas, requires a large effort from the attackers in terms of labeling thousands of captcha samples by hand, in order to train the classifiers. In this project, we want to explore the possibility of using semi-supervised learning approaches to break captchas. With only a small number of solved captchas given, semi-supervised learning can exploit a huge corpus of unsolved captchas. First, we transfer the unlabelled observations to a low-dimensional representation then distribute the few available labels to close-by observations. |
| Campus Location | University of California, Berkeley |
| Research Topic | RACS: Research in Cloud Storage Diversity |
|---|---|
| Sponsoring TRUST Faculty |
Prof. Hakim Weatherspoon, Computer Science, Cornell University |
| Description | The increasing popularity of cloud storage is leading organizations to consider moving data out of their own data centers and into the cloud. However, success for cloud storage providers can present a significant risk to customers; namely, it becomes very expensive to switch storage providers. In this summer project, we make a case for applying RAID-like techniques used by disks and file systems, but at the cloud storage level. We argue that striping user data across multiple providers can allow customers to avoid vendor lock-in, reduce the cost of switching providers, and better tolerate provider outages or failures. We will work on a project called RACS (redundant array of cloud storage), a proxy that transparently spreads the storage load over many providers. We will build and evaluate a prototype of RACS system and estimate the costs incurred and benefits reaped. |
| Campus Location | Cornell University |
| Research Topic | SoNIC Boom |
|---|---|
| Sponsoring TRUST Faculty |
Prof. Hakim Weatherspoon, Computer Science, Cornell University |
| Description | State-of-the-art networks require state-of-the-art methodologies to understand and secure them, and use them efficiently. Our SoNIC-enabled (software defined network interface) networks are a crucial enabling step. Informed by the improved understanding, control, and flexibility given by being able to control the entire network stack in software, we expect to develop better protocols for moving large quantities of data securely and reliably in modern networks. In this summer project, we investigate end-to-end system dependability, focusing on the flow dynamics introduced by a state-of-the-art 10 Gbps wide-area network carrying a variety of extremely steady data streams. We intend to show that the burstiness introduced by this network causes endpoint buffer overflows and resultant packet loss, and that the degree of loss can be far more severe than would be expected purely on the basis of the packet chain lengths. Further, we would like to investigate ways in which data transfer protocols, like TCP, could be modified to remedy the problem. The issue is important: enterprises ranging from geographically dispersed scientific projects that move large data sets, to cloud computing applications shipping data between data centers (or directly to end-users), are building networks of the sort we used in our studies. As a result, a substantial community faces the performance issues we investigate, and would benefit from the remedial steps we would research and suggest. |
| Campus Location | Cornell University |
| Research Topic | Securing Web Applications from Logic Flaws |
|---|---|
| Sponsoring TRUST Faculty |
Prof. Yuan Xue, Electrical Engineering & Computer Science, Vanderbilt University |
| Description | The World Wide Web has evolved from a system that delivers static pages to a platform that supports distributed applications, known as web applications, and has become one of the most prevalent technologies for information and service delivery over Internet. Web applications usually interact with back-end database systems that may store sensitive information (e.g., financial, health) and are increasingly used to deliver security critical services. Web applications also become a primary and valuable target for cyber attacks, which brings about serious security concerns for users and corporations that rely on web applications. A breach report from Verizon shows that web applications now reign supreme in both the number of breaches and the amount of data compromised. For instance, in June 2010 it was reported that a vulnerability of the AT&T website allowed an attacker to harvest Apple iPad subscriber e-mails by enumerating ICC-ID numbers, which affected over 100,000 Apple customers. As web applications become deeply embedded in business activities and are required to support sophisticated business functions, the design and implementation of web applications become more and more complicated. The increasing complexity is compounded by insufficient security assurance from both current widely-used web application development and testing frameworks plus developers with insufficient security skills or awareness. As a result, a high percentage of web applications deployed on the Internet are exposed to security vulnerabilities. The goal of this research is to harden web applications and secure them from logic flaws and state violation attacks. In particular, we aim to develop a software testing technique that is able to identify potential logic vulnerabilities within web applications so as to prevent successful state violation attacks. |
| Campus Location | Vanderbilt University |
| Research Topic | Evaluating the Impact of Security Attacks on Cyber-Physical Systems |
|---|---|
| Sponsoring TRUST Faculty |
Dr. Yuan Xue, Electrical Engineering & Computer Science, Vanderbilt University |
| Description | Cyber-physical systems (CPS) are characterized by the tight coupling and coordination among sensing, communications, computational, and physical resources. As CPS become more complex through distributed architectures and expanded mission capability, it becomes more challenging to assure the performance, stability, safety, and security properties of their behavior. There is a pressing need to evaluate both cyber and physical systems together and holistically in realistic network environments, especially when under security attacks. The goal of this project is to perform experimental studies for CPS sytems in an integrated environment of simulation tools (e.g., MatLab) and emulation environments (e.g., DETERlab). Specific tasks will include (1) implementing/running a simple networked control system in DETERLab, (2) using network attack generation tools to generate network attacks during the execution of the network control system, and (3) collecting traces, measuring NCS system performance, and studying the results. |
| Campus Location | Vanderbilt University |
| Research Topic | A Study of the Human Action Database |
|---|---|
| Sponsoring TRUST Faculty |
Prof. Ruzena Bajcsy, Electrical Engineering and Computer Sciences, University of California, Berkeley |
| Description | Forthcoming |
| Campus Location | University of California, Berkeley |