CSE Defenses This Week

CSE graduate defenses
CSE graduate defenses

Several CSE graduate students will defend their projects, theses, and dissertations this upcoming week. Here is the full list of students and their defenses:

Master's Thesis Defense: "Optimization of Irrigation Recommendation in CornSoyWater"

Dharmic Payyala

Committee Members: Dr. Jitender Deogun (Advisor), Dr. Vinod Variyam and Dr. Haishun Yang

Thursday, November 17, 2016, 3:30 p.m.
347 Avery Hall

Abstract:
A crop simulation model is used to estimate crop production as a function of weather conditions, soil parameters, and plant related inputs. These crop simulation models are extensively used by farmers, corporations and policy makers for agronomical decision making. CornSoyWater is one such application which provides irrigation recommendation for soy and corn farmers using hybrid maize and soy sim models. As this is a simulation technology, the accuracy of results depends on the quality of data provided to it. One such important input parameter is weather data. CornSoyWater simulates field and crop conditions by retrieving the updated weather data from nearest weather station for that field. However, the closest weather station could be far enough that the simulation model cannot rely on that weather station's data. Currently, in CornSoyWater twenty miles is considered as the threshold distance beyond which the nearest weather station's data might not represent the field conditions. A significant amount of fields which produces corn and soybean doesn’t have access to weather station within the threshold distance, which arises the necessity to optimize the existing models to work for real-world scenarios. In this thesis, we solved this problem using a new approach which uses quantification of the shape, distance, and position of weather stations to choose the optimal ones and performs inverse distance weighing on them. The results demonstrate that this approach can be used for those fields which don't have access to weather stations within the threshold distance.





Master's Project Defense: "MOHIN: A Modular, Mobile and Multi-Platform Solution for Healthcare Interventions"

Vishnu Sivadasan

Committee Members: Dr. Byrav Ramamurthy (Advisor), Dr. Witawas Srisa-an and Dr. Sheng Wei

Monday, November 21, 2016, 2:00pm
112 Schorr Center

Abstract:
Smartphones have become a commonplace object in everyone's lives and its popularity is increasing at a very fast rate. Many companies and organizations in different sectors have leveraged this by making apps. This includes the Healthcare sector too. There are several apps that can monitor a patients vital signs such as heart rate, sleeping pattern, temperature, blood pressure etc. There are apps that encourage users to engage in physical activities by providing motivational messages. Many of these apps provide a platform for healthy competition with your friends on health related activities such as steps walked, calories burned etc.

In our project we focus on healthcare intervention, where a healthcare practitioner can use the features of a smartphone to perform interventions. We developed a system, MOHIN, which includes many of the popular services used by a healthcare practitioner to perform an intervention such as scheduled messaging, real-time messaging and discussion forums. MOHIN is customizable in that the healthcare practitioner can choose the services they require for the study based on the nature of the intervention. The patients and the healthcare practitioner can interact with MOHIN using an Android app, iOS app or a web browser.

MOHIN enables the healthcare practitioner to perform interventions with very little supervision. This helps the healthcare practitioner to communicate with a large number of patients at the same time. MOHIN automates the task of sending out intervention messages using a scheduler. The patients can easily communicate with the healthcare practitioner using real-time messaging. The patients can also discuss issues with other patients in a moderated discussion forum.

This project is the result of a collaboration between UNL and UNMC College of Nursing. A feasibility study was conducted using volunteers as participants and MOHIN was found to be effective in delivering customized healthcare interventions to the study participants.




Master's Thesis Defense: "Power Management In Heterogeneous Mapreduce Cluster"

Rojee Sunuwar

Committee Members: Dr. Ying Lu (Advisor), Dr. David Swanson and Lisong Xu

Monday, November 21, 2016, 9:00 a.m.
347 Avery Hall

Abstract: The growing expenses of power in data centers as compared to the operation costs has been a concern for the past several decades. It has been predicted that without an intervention, the energy cost will soon outgrow the infrastructure and operation cost. Therefore, it is of great importance to make data center clusters more energy efficient which is critical for avoiding system overheating and failures. In addition, energy inefficiency causes not only the loss of capital but also environmental pollution. Various Power Management(PM) strategies have been developed over the years to make system more energy efficient and to counteract the sharply rising cost of electricity. However, it is still a challenge to make the system both power efficient and computation efficient due to many underlying system constraints.

In this thesis, we investigate the Power Management technique in heterogeneous MapReduce clusters while also maintaining the required system QoS (Quality of Service). For a cluster that supports MapReduce jobs, it is necessary to develop a PM technique that also considers the data availability. We develop our PM strategy by exploiting the fact that the servers in the system are underutilized most of the time. Hence, we first develop a model of our testbed and study how the server utilization levels affect the power consumption and the system throughput. With the established models, we form and solve the power optimization problem for heterogeneous MadReduce clusters where we control the server utilization levels intelligently to minimize the total power consumption.

We have conducted simulations and shown the power savings achieved using our PM technique. Then we validate some of our simulation results by running experiments in a real testbed. Our simulation and experimental data have shown that our PM strategy works well for heterogeneous MapReduce clusters and it scales with the number of servers in the system.





Master's Thesis Defense: "Testing the Mutation Independence Hypothesis for Pairs of Adjacent Amino Acids in Protein Sequences"

Jyotsna Ramanan

Committee Members: Dr. Peter Revesz (Advisor), Dr. Juan Cui and Dr. Stephen Scott

Monday, November 21, 2016, 3:30 p.m.
347 Avery Hall

Abstract:
Evolutionary studies usually assume that the genetic mutations are independent of each other. However, that does not imply that the observed mutations are independent of each other because it is possible that when a nucleotide is mutated, then it may be biologically beneficial if an adjacent nucleotide mutates too.

With a number of decoded genes currently available in various genome libraries and online databases, it is now possible to have a large-scale computer-based study to test whether the independence assumption holds for pairs of adjacent amino acids. Since the coding regions of DNAs encode proteins, and the mutations in those regions are reflected in the mutation of the encoded amino acids. Hence the independence question also arises for pairs of adjacent amino acids within proteins. The independence question can be tested by considering the evolution of proteins within a closely related sets of proteins, which are called protein families.


In this thesis, we test the independence hypothesis for three protein families from the PFAM library, which is a publicly available online database that records a growing number of protein families. For each protein family, we construct a hypothetical common ancestor, or consensus sequence. We compare the hypothetical common ancestor of a protein family with each of the descendant protein sequences in the family to test where the mutations occurred during evolution. The comparison yields actual probabilities for each pairs of amino acids changing into another pair of amino acids. By comparing the actual probabilities with the theoretical probabilities under the independence assumption, we identify anomalies that indicates that the independence assumption does not hold for many pairs of amino acids.





Master's Thesis Defense: "Using Software Testing Techniques to Infer Biological Models"

Mikaela Cashman

Committee Members: Dr. Myra Cohen (Advisor), Dr. Matthew Dwyer, Dr. Massimiliano Pierobon, and Dr. Nicole Buan

Tuesday, November 22, 2016, 9:00 a.m.
347 Avery Hall

Abstract:
Years of research in software testing has given us novel ways to reason about and test the behavior of complex software systems that contain hundreds of thousands of lines of code. Many of these techniques have been inspired by nature such as genetic algorithms, swarm intelligence, and ant colony optimization. However, they use a unidirectional analogy—taking from nature without giving back.

In this thesis we invert this view and ask if we can utilize techniques from testing and modeling of highly-configurable software systems to aid in the emerging field of systems biology which aims to model and predict the behavior of biological organisms. Like configurable systems, the underlying source code (metabolic model) contains both common and variable code elements (reactions) that are executed only under particular configurations (environmental conditions) and these directly impact an organism's observable behavior. We propose the use of sampling, classification, and modeling techniques commonly used software testing and combine them into a process called BioSIMP which can lead to simplified models and biological predictions.

We perform two case studies, the first of which explores and evaluates different classification techniques to infer influential factors in microbial organisms. We then compare several sampling methods to limit the number of experiments required in the laboratory. We show that we can reduce testing by more than two thirds without negatively impacting the quality of our models. Finally, we perform an end-to-end case study on BioSIMP using both laboratory and simulation data and show that we can find influencing environmental factors in two microbial organisms, some of which were previously unknown to biologists.

Our findings suggest that the configurable-software analogy holds, and we can identify the variable and common regions of reactions that change with respect to the environment.





Dissertation Defense: "Autonomous UAVs for Near Earth Environmental Sensing"

David Anthony

Committee: Dr. Carrick Detweiler (Advisor), Dr. Sebastian Elbaum, Dr. Witawas Srisa-an and Dr. Richard Ferguson

Tuesday, November 22, 2016, 10:00 a.m.
211 Schorr Center

Abstract:
Unmanned aerial vehicles (UAVs) are playing an increasing role in large scale environmental monitoring. Small UAVs are increasingly used to monitor agricultural field, infrastructure projects, and disaster areas. The combination of their sensors, ease of use, and portability make them an ideal tool for collecting information on demand about geographic regions. These small UAVs do have several significant limitations. The UAVs have very limited autonomy and fly pre-determined flight paths far above the underlying terrain, limiting the spatial resolution of the collected data. Current battery technology several limits their flight times, which in turn limits the temporal resolution of the data collection processes. What all users, from farmers to first responders, require are improved systems with higher levels of autonomy for long term environmental monitoring.

Our research uses two complementary approaches to address these limitations in autonomy and endurance. Our first contribution introduces novel techniques for localizing small UAVs in outdoor environments. These improved localization techniques allow the vehicle to operate within one meter of mature agricultural fields throughout the growing season. This near earth operation allows the UAV to fly missions that are not possible for human pilots or current UAVs with standard localization systems to fulfill. Improved localization not only improves the vehicles' autonomy, but it also increases the spatial resolution of data collected from the airborne vehicles. Furthermore, the UAVs can utilize shorter range or lighter weight sensors that are not effective on higher flying aircraft.

We complement the short-term near earth sensing capabilities by deploying long lasting wireless sensor network (WSN) nodes from the same small UAVs. We improve upon traditional WSN deployment mechanisms by performing tactile surface classification from the UAV prior to deploying a node. This pre-deployment procedure changes haphazard sensor deployments to controlled installations, which improves deployment outcomes. The UAVs also performs a post-deployment assessment of the installation outcome, enabling preemptive replacement or maintenance of ineffective nodes.

These complementary approaches exploit the high mobility and powerful sensors of small UAVs with energy efficient and long lasting WSNs to maximize the information collected from the environment. Our localization techniques improve the autonomy and capabilities of small UAVs in agricultural and other complex outdoor environments. We validate our approach in outdoor field trials that demonstrate our approach is capable of localizing and navigating UAVs closer to crops than traditional GPS based guidance allows. Using these low flying vehicles we show that we can replicate manually collected observations that are currently used in phenotyping trials and gain deep insight into field structure and development. Our work on WSN deployments bridges the gap between the WSN and robotics communities to create comprehensive environmental monitoring capabilities that use the strengths of both groups.





Master's Thesis Defense: "Characterization Of Molecular Communication Based On Cell Metabolism Through Mutual Information and Flux Balance Analysis"

Zahmeeth Sayed Sakkaff

Committee Members: Dr. Massimiliano Pierobon (Advisor), Dr. Myra Cohen and Dr. Juan Cui

Tuesday, November 22, 2016, 3:30pm
347 Avery Hall

Abstract:
Synthetic biology is providing novel tools to engineer cells and access the basis of their molecular information processing, including their communication channels based on chemical reactions and molecule exchange. Molecular communication is a discipline in communication engineering that studies these types of communications and ways to exploit them for novel purposes, such as the development of ubiquitous and heterogeneous communication networks to interconnect biological cells with nano and biotechnology-enabled devices, i.e., the Internet of Bio-Nano Things. One major problem in realizing these goals stands in the development of reliable techniques to control the engineered cells and their behavior from the external environment. A possible solution may stem from exploiting the natural mechanisms that allow cells to regulate their metabolism, the complex network of chemical reactions that underlie their growth and reproduction, in function of chemical compounds in the environment.

In this thesis, molecular communication concepts are applied to study the potential of cell metabolism, and its regulation, to channel information from the outside environment into the cell as function of chemical compounds in the environment, and quantify how much information of the internal state of the metabolic network can be perceived from the outside environment. For this, cell metabolism is characterized in this work through two abstractions, namely, as a molecular communication encoder and a modulator, respectively. The former models the cell metabolism as a binary encoder of the mechanisms underlying the regulation of the cell metabolic network state in function of the chemical composition of the external environment. The latter models the metabolic network inside the cell as a digital modulator of metabolite exchange/growth according to the information contained in its state. Based on these abstractions, the aforementioned potential of cell metabolism is quantified with the information theoretic mutual information parameter obtained through the use of a well-known and computationally efficient metabolic simulation technique. Numerical results are obtained through simulation of cell metabolism based on the standard processes of Genome Scale Modeling (GEM) and Flux Balance Analysis (FBA). These preliminary proof-of-concept results are based on the following three main cellular species: Escherichia coli (E. coli), the “standard” organism in microbiology, and two important human gut microbes studied in our collaborators’ lab, namely, the Bacteroides thetaiotaomicron (B. theta) and Methanobrevibacter smithii (M. smithii), which provide a direct connection of this work to future practical applications.