Defense Notices

The dynamic thinning of fast-flowing glaciers is so poorly understood that its potential impact on sea level rise remains unpredictable. Therefore, there is a dire need to predict the behavior of these ice bodies by understanding their bed topography and basal conditions, particularly near their grounding lines (the limit between grounded ice and floating ice). The ability to detect previous VHF radar returns in some key glacier regions is limited by strong clutter caused by severe ice surface roughness, volume scatter, and increased attenuation induced by water inclusions and debris. 
The work completed in the context of this thesis encompasses the design, integration, and field testing of a new compact light-weight radar and antenna system suitable for low-frequency operation onboard Uninhabited Aerial Systems (UASs). Specifically, this thesis presents the development of two tapered dipole antennas compatible with a 4-meter wingspan UAS. The bow-tie shaped antenna resonates at 35 MHz, and the meandering and resistively loaded element radiates at 14 MHz. Also discussed are the methods and tools used to achieve the necessary bandwidth while mitigating the electromagnetic coupling between the antennas and on-board avionics in a fully populated UAS. The influence of EM coupling on the 14 MHz antenna was nominal due to relatively longer wavelength. However, its input impedance had to be modified by resistive loading in order to avoid high power reflections back to the transmitter. The antenna bandwidths were further enhanced by employing impedance matching networks that resulted in 17.3% and 7.1% bandwidths at 35 MHz and 14 MHz, respectively. 
Finally, a compact 4 lbs. system was validated during the 2013-2014 Antarctic deployment, which led to echo sounding of more challenging temperate ice in the Arctic Circle. The thesis provides results obtained from data collected during a field test campaign over the Russell glacier in Greenland compared with previous data obtained with a VHF depth sounder system operated onboard a manned aircraft. 

Technological advancements in analog and digital systems have enabled new approaches to study networks of physical and artificial neurons. In biological systems, a standard method to record neuronal activity is through cortically implanted micro electrode arrays (MEAs). As advances in hardware continue to push channel counts of commercial MEAs upwards, it becomes imperative to develop automatic methods for data acquisition and analysis with high accuracy and throughput. Reliable, low latency methods are critical in closed-loop neuroprosthetic paradigms such as spike-timing dependent applications where the activity of a single neuron triggers specific stimuli with millisecond precision. This work presents an adapted version of an online spike detection algorithm, previously employed successfully on in vitro recordings, that has been improved to work under more stringent in vivo environments subject to additional sources of variability and noise. The algorithm’s performance was compared with other commonly employed detection techniques for neural data on a newly developed and highly tunable extracellular recording model that features variable firing rates, adjustable SNRs, and multiple waveform characteristics. The testing framework was created from in vivo recordings collected during quiescence and electrical stimulation periods. The algorithm presents superior performance and efficiency in all evaluated conditions. Furthermore, we propose a methodology for online signal integrity analysis from MEA recordings and quantification of neuronal variability across different experimental settings. This work constitutes a stepping stone toward the creation of large scale neural data processing pipelines and aims to facilitate reproducibility in activity dependent experiments by offering a method for unifying various metrics calculated from single unit activity. Precise spike detection becomes crucial for experiments studying temporal in addition to rate coding mechanisms. To further study and exploit the potential of temporal coding, a delay-feedback-based reservoir (DFB) has been implemented in software. This artificial network is found to be capable of processing spikes encoded from a benchmark task with performance comparable to that of more complex networks. This work allows us to corroborate the capabilities of temporal coding in a minimally-complex system suitable for implementation in physical hardware and inclusion in low-power circuit applications where computational power is also necessary.

The worldwide energy demand is being increased day by day, anticipated to increase for 48% from 2012 to 2040. The distributed generation (DG) including renewable energy resources such as wind and solar are part of the solution in terms of lowering electricity cost, power reliability, and environmental concerns and therefore must function efficiently. Designing a robust maximum power point tracking (MPPT) technique can ensure maximized energy harvesting from PV solar systems and increases conversion efficiency which is the significant hindrance for their growth. The maximum power point (MPP) varies with intrinsic and climate changes nonlinearly. Thus, MPPT methods are expected to seek the MPP regardless of the solar module and ambient changes. The proposed method is based on the concept of Model Predictive Control (MPC) with unique properties. MPC is a powerful class of controllers that uses a system modeling to predict future behavior and optimize performance objectives. Unlike the traditional techniques that are prone to lose a tracking direction and their consequences on the stability, the proposed technique treats the photovoltaic (PV) module as a plant and uses a digital observer for predicting the behavior of the PV module and tracking the MPP. Further, it unifies the simplicity of implementation, enhances the overall dynamics performance and is robust against atmosphere changes.

Historical subsurface sensing methods applied to planar ice and snow sheets rely on underlying assumptions about the physical situation governing volumetric backscatter. Namely, the stratification of the natural medium under investigation consists of layered material with distinctly different dielectric properties. While appropriate for recovering sharp spatial discontinuities in the relative permittivity, the layer stripping approach [1] is not applicable to smooth permittivity variations about a common mean. In this project we developed techniques to model both the forward scattering from one-dimensional permittivity variation and the inverse problem - estimating the permittivity profile from the reflected energy. The underlying assumption is that smoothly varying inhomogeneities may be decomposed into wavelet basis functions which efficiently represent natural perturbations about an effective mean. Potential applications for this method are in ground penetrating radar, ionospheric sounding, nondestructive evaluation, and medical imaging.

Simulating the mechanics of a beating heart requires the numerical solution of partial differential equations. An application like this is a good candidate for high order computational methods that deliver higher solution accuracy at a lower cost than their low order counterparts. 
To fully leverage these high order computational methods, they must be paired with an accurate discretization of the domain. For a geometry like the heart, this requires a high order mesh. Thus robust high order mesh generation is a critical component to the widespread adoption of high order computational methods for numerically solving partial differential equations. Toward this end, we are developing high order mesh generation and untangling methods. As our first step, we have developed an optimization-based second order mesh generation method that employs triangles and tetrahedra. We will also develop generation methods for quadrilateral and hexahedral elements. Finally, we will develop untangling methods that can be used to untangle our generated meshes, as well as untangle any tangled elements that occur during motion (e.g. the beating of the heart). 

Unmanned Aerial Vehicles (UAVs) are increasingly demanded in civil, military and research purposes. However, they also possess serious threats to the society because faults in UAVs can lead to physical damage or even loss of life. While increasing their intelligence, for example, adding vision-based sense-and-avoid capability, has a potential to reduce the safety threats, increased software complexity and the need for higher 
computing performance create additional challenges—software bugs and transient hardware faults—that must be addressed to realize intelligent and safe UAV systems. 
This work present a fault tolerant system design for UAVs. Our proposal is to use two heterogeneous hardware and software platforms with distinct reliability and performance characteristics: High-Assurance (HA) and High-Performance (HP) platforms. The HA platform focuses on simplicity and 
verfiability in software and uses a simple and transient fault tolerant processor, while the HP platform focuses on intelligence and functionality in software and uses a complex and high performance processor. During the normal operation, the HP platform is responsible for controlling the UAV. However, if it fails due to transient hardware faults or software bugs, the HA platform will take over until the HP platform recovers. 
We have implemented the proposed design on an actual UAV using a low-cost Arduino and a high-performance Tegra TK1 multicore platform. Our case-studies show that our design can improve safety without compromising performance and intelligence of the UAV. 

Scene understanding in both static images and dynamic videos is the ultimate goal in computer vision. As two important sub-tasks of this endeavor, object detection and tracking have been extensively studied in the past decades, however, the problem is still not well addressed. The main challenge is that the appearance of objects is affected by a number of factors, such as scale, occlusion, illumination, and so on. Recently, deep learning has attracted lots of interests in the computer vision community. However, how to tackle these challenges in object detection and tracking is still an open problem. In this work, we propose a method for detecting objects in images using a single deep neural network, which can be optimized end-to-end and predict the object bounding boxes and class probabilities in one evaluation. To handle the challenges in object tracking, we propose a framework, which consists of a novel deep Convolutional Neural Networks (CNNs) to effectively generate robust spatial appearance, and a Long Short-term Memory (LSTM) network that incorporates temporal information to achieve long-term object tracking accuracy in real-time.

According to Moore’s law the number of transistors per square inch double every two years. Scaling down technology reduces size and cost however, also increases the number of problems. Our current computers using Von-Neumann architectures are seeing progressive difficulties not only due to scaling down the technology but also due to grid-lock situation in its architecture. As a solution to this, scientists came up architectures whose function resembles that of the brain. They called these brains inspired architectures, neuromorphic computers. The building block of the brain is the neuron which encodes, decodes and processes the data. The neuron is known to accept sensory information and converts this information into a spike train. This spike train is encoded by the neuron using different ways depending on the situation. Rate encoding, temporal encoding, population encoding, sparse encoding and rate-order encoding are a few encoding schemes said to be used by the neuron. These different neural encoding schemes are discussed as the primary focus of the thesis. A comparison between these different schemes is also provided for better understanding, thus helping in the design of an efficient neuromorphic computer. This thesis also focusses on hardware implementation of a neuron. Leaky Fire and Integrate neuron model has been used in this work which uses spike-time dependent encoding. Different neuron models are discussed with a comparison as to which model is effective under which circumstances. The electronic neuron model was implemented using 180nm CMOS Technology using Global Foundries PDK libraries. Simulation results for the neuron are presented for different inputs and different excitation currents. These results show the successful encoding of sensory information into a spike train.

Due to the increasing need for greater Radio Frequency (RF) spectrum by mobile apps like Facebook and Instagram, high data-rate communication protocols like 5G and the Internet of Things, it has led to the issue of spectrum congestion as radar systems have traditionally maintain the largest share of the RF spectrum. To resolve the spectrum congestion problem, it has become even necessary for users from both types of systems to coexist within a finite spectrum allocation. However, this then leads to other problems such as the increased likelihood of mutual interference experienced by all users that are coexisting within the finite spectrum. 

In this dissertation, we propose to address the problem of spectrum congestion via a two-step approach. The first step of this approach involves designing an optimal sparse spectrum allocation scheme to radar systems such that the radar range resolution performance can be maintained with a smaller resulting bandwidth at a cost of degraded sidelobe performance. The second step of this approach involves designing radar waveforms that possesses good spectral containment property by expanding the framework of Polyphase-coded Frequency Modulated (PCFM) waveforms to higher-order representations such that these waveforms will mitigate issues of interference experienced by other systems when both systems are coexisting within the same band. 

Von Neumann Bottleneck, which refers to the limited throughput between the CPU and memory, has already become the major factor hindering the technical advances of computing systems. In recent years, neuromorphic systems started to gain the increasing attentions as compact and energy-efficient computing platforms. As one of the most crucial components in the neuromorphic computing systems, neural encoder transforms the stimulus (input signals) into spike trains. In this report, I will present my research work on spike-time-dependent encoding schemes and its relevant energy efficient encoders’ design. The performance comparison among rate encoding, latency encoding, and temporal encoding would be discussed in this report. The proposed neural temporal encoder allows efficient mapping of signal amplitude information into a spike time sequence that represents the input data and offers perfect recovery for band-limited stimuli. The simulation and measurement results show that the proposed temporal encoder is proven to be robust and error-tolerant.