Defense Notices

Online Social Networks (OSNs) are integrated into business, entertainment, politics, and education; they are integrated into nearly every facet of our everyday lives. They have played essential roles in milestones for humanity, such as the social revolutions in certain countries, to more day-to-day activities, such as streaming entertaining or educational materials. Not surprisingly, social networks are the subject of study, not only for computer scientists, but also for economists, sociologists, political scientists, and psychologists, among others. In this dissertation, we build a model that is used to classify content on the OSNs of Reddit, 4chan, Flickr, and YouTube according the types of lifespan their content have and the popularity tiers that the content reaches. The proposed model is evaluated using 10-fold cross-validation, using data mining techniques of Sequential Minimal Optimization (SMO), which is a support vector machine algorithm, Decision Table, Naïve Bayes, and Random Forest. The run times and accuracies are compared across OSNs, models, and data mining algorithms. 
Our experiments compared the runtimes and accuracy of SMO, Naïve Bayes, Decision Table, and Random Forest to classify the lifespan of content on Reddit, 4chan, and Flickr as well as classify the popularity tier of content on Reddit, 4chan, Flickr, and YouTube. The experimental results indicate that SMO is capable of outperforming the other algorithms in runtime across all OSNs. Decision Table has the longest observed runtimes, failing to complete analysis before system crashes in some cases. The statistical analysis indicates, with 95% confidence, there is no statistically significant difference in accuracy between the algorithms across all OSNs. Reddit content was shown, with 95% confidence, to be the OSN least likely to be misclassified. All other OSNs, were shown to have no statistically significant difference in terms of their content being more or less likely to be misclassified when compared pairwise with each other.

To enhance the capabilities of autonomous flight systems for Unmanned Aircraft Systems (UASs), a multichannel Frequency-Modulated Continuous Wave (FMCW) collision-avoidance radar with a center frequency of 1.445 GHz is designed. The radar is intended to provide situational awareness for a 40% Yak-54 model aircraft by providing in real time range, radial velocity and angle-of-arrival (AoA) information on surrounding targets with an update rate of 10 Hz. A target’s range and Doppler is determined by employing a two-dimensional (2-D) Fast Fourier Transform (FFT) on the received signal which maps the target to a specific range-Doppler bin. Tests have shown that the proto-type radar is capable of providing range detection up to 430 m with an accuracy of 0.6 m for a target with a 1-m2 radar cross section (RCS). The radar is designed to provide a Doppler resolution of 10 Hz. An array of receiving antennas is used to determine a target’s elevation and azimuth angles by exploiting the received signal’s phase difference at each individual antenna. The AoA measurement error due to thermal noise was found to be less than 3° for a signal-to-noise ratio (SNR) of 18 dB.

Nowadays superlattices (SLs) are widely used as optical materials due to optical absorption properties. Short-period superlattices with certain optical properties such as InAs/GaSb type-II superlattices and ZnTe/CdTe superlattices can serve for mid-infrared (MIR) detection and solar cells. In this study, a standard Kronig-Penney model is applied to calculate the mini band structure of such SLs. On the basis of the energy-balance equation derived from the Boltzmann equation, a simple approach is used to calculate the optical absorption coefficient for the corresponding SL systems. Comparison of simulation results and experimental findings will be made in this study. And reasonable causes of error and future work will be discussed.

Ever since the advent of electronic fuel injection, auto manufacturers have been able to increase fuel efficiency and power production, and to meet stricter emission standards. Most of these systems use engine sensors (RPM, Throttle Position, etc.) in concert with look-up tables to determine the correct amount of fuel to inject. While these systems work well, it is time and labor intensive to fine tune the parameters for these look-up tables. Automobile manufacturers are able to absorb the cost of this calibration since the variation between engines in a new model line is often small enough as to be inconsequential for a specific calibration. 

However, a growing number of drivers are interested in modifying their vehicles with the intent of improving performance. While some aftermarket performance upgrades can be accounted for by the original manufacturer equipped (OEM) electronic control unit (ECU), other more significant changes, such as adding a turbocharger or installing larger fuel injectors, require more drastic accommodations. These modifications often require an entirely new ECU calibration or an aftermarket ECU to properly control the upgraded engine. The problem then, is that the driver becomes responsible for the calibration of the ECU of this “new” engine. However, most drivers are unable to devote the resources required to achieve a calibration of the same quality as the original manufacturers. At best, this results in reduced fuel economy and performance, and at worst, unsafe and possibly destructive operation of the engine. 

The purpose of this thesis is to design and develop—using machine learning techniques—an approximate predictive model from current engine data logs, which can be used to rapidly and incrementally improve the calibration of the engine. While there has been research into novel control methods for engine air-fuel ratio control, these methods are inaccessible to the majority of end users, either due to cost or the required expertise with engine calibration. This study shows that there is a great deal of promise in applying machine learning techniques to engine calibration and that the process of engine calibration can be expedited by the application of these techniques.

Linear amplification using nonlinear components (LINC) is a design approach that can suppress the effects of the nonlinear distortion introduced by the transmitter. A typical transmitter design requirement is for the high power amplifier to be operated in saturation. The LINC approach described here employs a polyphase-coded FM (PCFM) waveform that is able to overcome this saturated amplifier distortion to greatly improve the spectral containment of the transmitted waveform. A two stage optimization process involving simulation and hardware-in-the-loop routines is used to create the final PCFM waveform code.

Mobile Ad Hoc networks~(MANETs) present great challenges to new protocol design, especially in scenarios where nodes are high mobile. Routing protocols performance is essential to the performance of wireless networks especially in mobile ad-hoc scenarios. The development of new routing protocols requires comparing them against well-known protocols in various simulation environments. Furthermore, application traffic like transactional application traffic has not been investigated for domain-specific MANETs scenarios. Overall, there are not enough performance comparison work in the past literatures. This thesis presents extensive performance comparison work with MANETs and uses inclusive parameter sets including both highly-dynamic environment as well as low-mobility cases.

Property-directed verification of functional programs tends to take one of two paths. 
First, is the traditional testing approach, where properties are expressed in the original programming language and checked with a collection of test data. 
Tools following this technique have the advantage of a direct integration with the host system, but their resultant statement about a program's correctness is anything but a guarantee. 
Alternatively, for those desiring a more rigorous approach, properties can be written and checked with a formal tool; typically, an external proof system. 
This process delivers a well reasoned argument for a program's correctness, however, it comes at the cost of a more complex system integration requiring additional expertise. 

We propose a hybrid approach that captures the best of both worlds: the formality of a proof system paired with the native integration of an embedded, domain specific language for testing. 
Presented in this document is a description of the hybridization, a theorem prover as a library, as well as a classification of our target properties for case study. 
As we attempt to verify these properties, our goal is to document and formalize the logical connection between language and tool. 
The resultant process will be evaluated both for the strength of its reasoning power and its viability for real world application.

A multichannel sense-and-avoid radar system targeted for small unmanned aerial vehicles (UAVs), such as the 40% Yak-54 RC aircraft, is being developed to assist the integration of UAVs into the national air space. This frequency-modulated continuous-wave (FMCW) radar system utilizes a two-dimensional fast-Fourier transform process to detect targets in range and Doppler. Interferometry using a 5-element receiver array allows the radar to calculate the azimuth/elevation angles of the target relative to itself. These tasks are being performed in real time with a targeted update rate of 10 Hz utilizing highly-integrated radar-ready components and an FPGA based processor. The focus of the research is on analysis and enhancement of the radar performance by implementing various detection and predictive algorithms such as extended Kalman filtering and constant false alarm rate detection. By tracking targets and predicting their future location, false alarms caused by anomalies can be minimized. Furthermore, targets located at the same range and Doppler will corrupt each other’s signals during interferometic processing thus giving the autopilot corrupted angle information. Using a predictive algorithm these occurrences can be avoided with some level of confidence.

The path geo-diversity routing protocol described in this report takes advantage of geographical diversity of physical network topology, and quickens the routing selection. By importing this mechanism, a more accurate path could be provided instead of multiple useless attempts when area-based challenges occur in the network. A k-shortest path algorithm is introduced, followed by a modified algorithm. These two algorithms are implemented in ns-3, and tested in both grid network and real network. Simulation results show that they provide better performance compared to OSPF, as multiple geo-diverse paths are calculated to provide reliable performance.

In a multi-agent system, an idle agent may be available to assist other agents in the system. An agent architecture called intent recognition is proposed to accomplish this with minimal communication. 
In order to assist other agents in the system, an agent performing recognition observes the tasks other agents are performing. Unlike the much studied field of plan recognition, the overall intent of an agent is recognized instead of a specific plan. The observing agent may use capabilities that it has not observed. This study focuses on the key research questions of: (1) What are intent recognition systems? (2) How can these be used in order to have agents autonomously assist each other effectively and efficiently? A conceptual framework is proposed for intent recognition systems. An implementation of the conceptual framework is tested and evaluated. We hypothesize that using intent recognition in a multi-agent system increases utility (where utility is domain specific) and decreases the amount of communication. We test our hypotheses using two experimental series in the domains of Box Pushing, where agents attempt to push boxes to specified locations; and Cow Herding, where agents attempt to herd cow agents into team corrals. A set of metrics, including task time and number of communications, is used to compare the performance of plan recognition and intent recognition. In both sets of experimental series, intent recognition agents communicate fewer times than plan recognition agents. In addition, unlike plan recognition, when agents use the novel approach of intent recognition, they select unobserved actions to perform, which was seen in both experimental series. Intent recognition agents were also able to outperform plan recognition agents by sometimes reducing task completion time in the Box Pushing domain and consistently scoring more points in the Cow Herding domain.