Defense Notices

Websockets is a protocol that provides a full-duplex communication channel over a single TCP connection between a web server and web client. Kansas-comet is long polling solution that allows web servers written in the functional programming language Haskell to push data to browser clients. Implementing kansas-comet with websockets enables pushing data from web servers to clients with reduced data loads and network latency, which helps in scaling web applications. Other applications, like the graphics library blank canvas, use kansas-comet, so improving kansas-comet also improves these applications as well. 

In this project, we add websockets to kansas-comet for the sake of improving client-server communications by providing a modern full duplex communication channel. Modern web browsers support the websocket protocol but it is important for kansas-comet to also provide backward compatibility. So, the new kansas-comet now implements a mechanism that falls back to long polling strategy when browser does not support websocket or when applications using kansas comet does not implement websockets. We use JavaScript and the kansas-comet JavaScript library on client browsers, and we use websocket, wai-websockets and warp libraries on the server side to implement websockets in kansas comet.

Increasing focus on global warming has challenged the scientific community to develop ways to mitigate its adverse effects. This is more so important as different technologies become an integral part of daily human life. Mobile wireless networks and mobile devices form a significant part of these technologies. It is estimated that there are over four billion mobile phone subscribers worldwide and this number is still growing as more people get connected in developing countries. In addition to the growing number of subscribers, there is an explosive growth in high data applications among mobile terminal users. This has put increased demand on the mobile network in terms of energy needed to support both the growth in subscribers and higher data rates. The mobile wireless industry therefore has a significant part to play in the mitigation of global warming effects. To achieve this goal, there is a need to develop and design energy efficient communication schemes for deployment in future networks and upgrades to existing networks. This is not only done in the wireless communication infrastructure but also in mobile terminals. In this project a practical power consumption model which includes circuit power consumption from the different components in a transceiver chain is analyzed. This is of great significance to practical system design when doing energy consumption and energy efficiency analysis. The proposed power consumption model is then used to evaluate the energy efficiency in the context of cooperative Multiple Input Multiple Output(MIMO)systems.

The reliance of Unmanned Aerial Vehicles (UAVs) on Global Navigation Satellite System (GNSS) for autonomous operation represents a significant vulnerability to their reliable and secure operation due to signal interference, both incidental (e.g. terrain shadowing, ionospheric scintillation) and malicious (e.g. jamming, spoofing). An accurate and reliable alternative UAV navigation system is proposed that exploits Signals of Opportunity (SOP) thus offering superior signal strength and spatial diversity compared to satellite signals. Given prior knowledge of the transmitter's position and signal characteristics, the proposed technique utilizes triangulation to estimate the receiver's position. Dual antenna interferometry provides the received signals' Angle of Arrival (AoA) required for triangulation. Reliance on precise knowledge of the antenna system's orientation is removed by combining AoAs from different transmitters to obtain a differential Angles of Arrival (dAoAs). Analysis, simulation, and experimental techniques are used to characterize system performance; a path to miniaturized system integration is also presented.

Batteries are used everywhere and with the rise of the portable devices it is crucial to lower the power dissipation and to improve the battery runtime. An efficient way to describe the electrical behavior of a battery helps the designer to better predict and optimize the battery runtime and circuit performance. In this project a suggested electrical circuit model is used to evaluate the battery characteristics of an alkaline cell and a rechargeable NiMH cell and the same is implemented in Cadence environment. The measured data is compared with the simulated data and the results are discussed further. This circuit model is efficient in modeling the behavior of the batteries used in this project and can be extended to various other types of batteries.

This paper proposes a novel extension to the Attribute Blocks visualization technique that can be applied to visualizations containing many attributes. The Attribute Blocks visualization scheme is a technique that divides the visualization space into a regular pattern of small cells where each cell displays only one attribute. This paper recommends that the goal of a pattern design should be to have each attribute share equal length edges with each other attribute. This goal imposes new constraints on the number of attributes that can be simultaneously displayed, hence one significant challenge was to develop a new strategy that would allow more flexible pattern geometry and evaluating the effectiveness of this strategy with real data sets.

A face detection and recognition system is a biometric identification mechanism which compared to other methods such as finger print identification, speech, signature, hand written and iris recognition is shown to be more important both theoretically and practically. In principle, the biometric identification methods use a wide range of techniques such as machine learning, machine vision, image processing, pattern recognition and neural networks. The methods have various applications such as in photo and film processing, control access networks, etc. In recent years, the automatic recognition of a human face has become an important problem in pattern recognition. The main reasons are that structural similarity of human faces and great impact of illumination conditions, facial expression and face orientation. Face recognition is considered one of the most challenging problems in pattern recognition. A face recognition system consists of two main components, face detection and recognition. In this dissertation we will design and implement a detection and recognition face system using color images with multiple faces. In color images, the information of skin color is used in order to distinguish between the skin pixels and non-skin pixels, dividing the image into some components. The next step is to decide which of these components belong to human face. After face detection, the faces which were detected in the previous step are to be recognized. Appearance based methods used in this work are one of the most important methods in face recognition due to the robustness of the algorithms to head rotation in the images, noise, low quality images, and other challenges.

This work formally introduces a generalized fuzzy logic and interval clustering (FLIC) technique which,when integrated with existing supervised learning algorithms, improves their performance. FLIC is a method that was first integrated with neural network in order to improve neural network’s performance in drug discovery using high throughput screening (HTS). This research strictly focuses on binary classification problems and generalizes the FLIC in order to incorporate it with other machine learning algorithms. In most binary classification problems, the class boundary is not linear. This pose a major problem when the number of outliers are significantly high, degrading the performance of the supervised learning function. FLIC identifies these misclassifications before the training set is introduced to the learning algorithm. This allows the supervised learning algorithm to learn more efficiently since it is now aware of those misclassifications. Although the proposed method performs well with most binary classification problems, it does significantly well for data set with high class asymmetry. The proposed method has been tested on four well known data sets of which three are from UCI Machine Learning repository and one from BigML. Tests have been conducted with three well known supervised learning techniques: Decision Tree, Logistic Regression and Naive Bayes. The results from the experiments show significant improvement in performance. The paper begins with a formal introduction to the core idea this research is based upon. It then discusses a list of other methods that have either inspired this research or have been referred to, in order to formalize the techniques. Subsequent sections discuss the methodology and the algorithm which is followed by results and conclusion. 

Keyword: supervised learning, binary classification, fuzzy logic, clustering 

IT Service Management (ITSM) is an IT function associated with resolving user issues through the support of a service desk. Some of the widely used ticket management tools that service desk utilizes include Remedy, Falcon and ServiceNow. These tools typically use a web portal as a front end for users to submit issues. Alternately, these tools may have a dedicated application that can be installed on a device. However, an application may not be compatible with various devices and is also very costly to maintain compatibility with current technology. Access to web portals requires a high bandwidth internet connection and connectivity could be a challenge in restricted areas. In these cases, a user’s only option is to report an issue via email. Email is supported on most connected devices and has very low internet bandwidth requirement. It also tends to be an ideal solution for traveling professionals. However, none of these ITSM tools provide a convenient mechanism to log tickets via email. Emails have to be manually converted to a ticket by the service desk. This process has a potential for human errors. 

With this objective, we have implemented an auto ticketing tool, 'TicketWise' that will automatically convert email requests into service tickets. This tool provides the necessary technological bridge for interfacing an email service with a ticketing system. This is a new feature that can be integrated with existing ITSM tools. New tickets get created for users who are registered with the system. Non-registered emails are automatically filtered out. Upon receiving a confirmation email the user can also send a follow up email. This information also gets updated in the ticket work log. 

TicketWise has been integrated with an application, 'TicketMe' that simulates a ticketing system. Validation has been successfully conducted by sending emails from a registered and a non-registered email address. In the former case, a new ticket was successfully created. In the latter, the email was filtered out. Contents from a follow up email for the ticket confirmation were also successfully added to the ticket work log. The results of the validation were satisfactory. 

Dementia is a major public health problem affecting 35 million people in USA. The caregivers of dementia patients experience many types of physical and psychological stress while dealing with disruptive behaviors of dementia patients. This will also result in frequent hospitalizations and re-admissions. In this project we design, implement, and measure the performance of an advanced video based monitoring system to aide the caregivers in managing the behavioral symptoms of dementia patients. The caregivers will be able to easily capture and share the antecedents, consequences, and the function of behavior, through a video clip, and get the real-time feedback from clinical experts. Overall the system will help in reducing the hospital admission/readmission, improve the quality of life for caregivers, and in general result in reduced cost of health care systems. System is developed using python scripts, open source web frameworks, FFmpeg tool chain, and commercial off-the-shelf IP camera and mini-PC. WebRTC is used for video based coaching of caregivers. A framework has been developed to evaluate the storage and retrieval latency of video clips to public and On-premise clouds, video streaming performance in LAN and WLAN environments, and WebRTC performance in different types of access networks. InstaGENIrack, a GENI rack in KU is used as on-premise cloud infrastructure for the evaluation. OpenSSL utilities are employed for secured transport and storage of captured video clips. We conducted the trials in Google fiber ISP in Kansas city, and compared the performance with other traditional ISPs..

Program execution can be classified into phases. These phases can be repeated during a single execution of the application. Ability to identify and classify the phases statically will help prepare the system early for the next phase, which can benefit overall program performance at run time. While static program phase detection algorithms have been explored for binary executable with promising results, to the best of our knowledge, such algorithms have not been targeted and evaluated for managed language, specifically Java, programs. Accurate detection of future program phases can allow the Java virtual machine to perform phase-specific optimizations to improve performance. 

In this project, we build a framework to detect program phases and insert software phase markers in Java byte code. We employ an existing algorithm to detect program phases and adapt it to detect phases for Java binaries. We modify the control flow graph generated by the byte code analysis tool WALA (Watson library for Analysis) to integrate program loops. We analyze each method in the control flow graph produced by WALA to detect loops, and convert the call graph into a "call loop graph". We then rely on program profiling to provide data on the number of times each basic block and edge is reached at run-time. We use this profiling information to determine the average number of instructions executed along each graph edge, the average number of times an edge is executed and the standard deviation for instructions on these edges in our algorithm to identify the software phase markers.