Electrical & Computer Engineering

Permanent URI for this community

http://hdl.handle.net/1903/2234

Browse

Now showing 1 - 20 of 1298

1D-CROSSPOINT ARRAY AND ITS CONSTRUCTION, APPLICATION TO BIG DATA PROBLEMS, AND HIGHER DIMENSION VARIANTS
(2022) An, Taeyoung; Oruc, Yavuz A; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Increased chip densities offer massive computation power to deal with fundamental bigdata operations such as sorting. At the same time the proliferation of processing elements (PEs) in settings such as High Performance Computers(HPCs) or servers together with the employment of more aggressive parallel algorithms cause the interprocessor communications to dominate the overall computation time, potentially resulting in reduced computational efficiency. To overcome this issue, this dissertation introduces a new architecture that uses simple crosspoint switches to pair PEs instead of a complex interconnection network. This new architecture may be viewed as a “quadratic” array of processors as it uses O(n^2) PEs rather than O(n) as in linear array processor models. In addition, three different models for sorting big data in a distributed com- puting environment such as Cloud computing are presented. With the most realistic model of the three, we demonstrate that the high parallelism made possible by the simple communication channels overcomes the seemingly excessive hardware complexity and performs comparable to or better than existing algorithms. Furthermore, two additional algorithms of matrix multiplica- tion and triangle counting for the 1D-Crosspoint Array are introduced and analyzed. Lastly, two higher dimensional variants, 2D- and 3D-Crosspoint Array are also proposed with a construction method, which succeeds in reducing the number of PEs required by utilizing the communication channels in the added dimensions.
3D Fast Geometric Collision Avoidance Algorithm (FGA) and Decision-Making Approach based on the Balance of Safety and Cost for UAS
(2021) lin, zijie; Xu, Mumu; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Unmanned Aircraft System (UAS) is a fast-growing industry with extensive economic implications and would be integrated into the national airspace system (NAS), which requires UAS to have the efficient sense and avoidance capability. This thesis develops a fast geometry-based algorithm FGA which shortens the collision avoidance computation time, path length and balances the probability of safety and energy cost by calculating and giving UAS proper selective avoidance starting time tc, meaning the last possible point for the UAS to avoiding the potential threaten and itis based on the UAS kinematic, conflicts likelihood map, and navigation constraints. This operation enables the update path to be as close as possible to the UAVs resume designed path, decreasing the length of path variation and the corresponding time cost. In comparison to a current geometry method, the sampling-based method and the search algorithm, the FGA algorithm shows 40% to 90% of reduction in computational time and length of path for the same obstacle avoidance scenarios. Quantitative analysis of the efficiency by different avoiding trigger times is also provided. FGA with critical avoidance time tc not only could improve the geometry methods, but also could be used for (1) research on the bounds of general geometry based collision avoidance, and (2) solving the multiple obstacles avoidance problem.For a scenarios with crowded obstacles which cannot be avoided at the same time, an applicable algorithm for obstacles classification is provided. It divides the obstacles into small groups with different urgent levels by their critical avoidance trigger time tc, and then avoids them in sequence. Simulation validates the efficiency of this application. Extremely difficult obstacle avoidance such as the UAV working under maneuver limitation and the obstacles are time-variant are discussed and solved in the following chapters. Monte Carlo simulation, statistical method and Machine learning algorithms especially the supervised logistic learning methods are implemented later to analyze the weight of the factors such as sensor detection distance, ratio of the speed, number of obstacles, which have impacts on the geometric based obstacle avoidance methods. Finally, flight missions in an aircraft simulator and the hardware fixed-wing aircraft experiments validate the algorithm effectiveness with successful results.
3D Integration, Temperature Effects, and Modeling
(2005-05-02) Parker, Latise; Goldsman, Neil; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Practical limits to device scaling are threatening the growth of integrated circuit (IC) technology. A breakthrough architecture is needed in order to realize the increased device density and circuit functionality that future high performance ICs demand. 3D integration is being considered as this breakthrough architecture. In this thesis, the limits to scaling are noted and the feasibility of overcoming these limits using 3D integration is presented. The challenges and considerations, most notably dangerously high chip temperatures, are provided. To address the temperature concern, a mixed-mode simulator that calculates temperature as a function of position on chip is detailed. The simulator captures the important link between individual device and full chip heating. Lastly, circuit simulations and lab experiments are performed to experimentally validate the claims that differences in device activity on chip leads to dangerously high local and overall chip temperatures.
3D Multimodal Image Registration: Application to equine PET and CT images
(2017) Regani, Sai Deepika; Chellappa, Rama; Beylin, David; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Positron Emission Tomography (PET) is being widely used in veterinary medicine in recent years. Although it was limited to small animals because of its classical design and the large amount of radionuclide doses required, PET imaging in horses became possible with the introduction of a portable PET scanner developed by Brain Biosciences Inc. It was observed that this new modality could capture abnormalities like lesions that Computed Tomography (CT), Magnetic Resonance Imaging (MRI) and other modalities could not. Since 2016, PET imaging in horses is being studied and analysed. While PET provides functional information characterizing the activity of lesions, it is useful to combine information from other modalities like CT and match the structural information to develop an accurate spatial representation of the data. Since biochemical changes occur much earlier than structural changes, this helps detect lesions and tumours during the early stages. Multimodal image registration is used to achieve this goal. A series of steps are proposed to automate the process of registration of equine PET and CT images. Multimodal image registration using landmark-based and intensity-based techniques are studied. It is observed that a few tissues are not imaged in the PET, which makes image segmentation, an important preprocessing step in the registration process. A study of the segmentation algorithms relevant to the field of medical imaging is presented. The performance of segmentation algorithms improved with the extent of manual interaction and intensity-based registration gave the smallest time complexity with reasonable accuracy.
Accelerated Imaging Using Partial Fourier Compressed Sensing Reconstruction
(2016) Chou, Chia-Chu; Babadi, Behtash; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Accelerated imaging is an active research area in medical imaging. The most intuitive way of image acceleration is to reconstruct images from only a subset of the whole raw data space, so that the acquisition time can be shortened. This concept has been formalized in recent years, and is known as Compressed Sensing (CS). In this dissertation, we developed a new image reconstruction method, Partial Fourier Compressed Sensing (PFCS), which combines the advantages of partial Fourier transform and compressed sensing techniques. Then, we explore its application on two imaging modalities. First, we apply PFCS to Electron Paramagnetic Resonance Imaging (EPRI) reconstruction for the purpose of imaging the cycling hypoxia phenomenon. We begin with validating PFCS with the prevailing medical acceleration techniques using CS. Then, we further explore its capability of imaging the oxygen distribution in the tumor tissue. Our results show that PFCS is able to accelerate the imaging process by at least 4 times with-out losing too much image resolution in comparison to conventional CS. Further, the ox-ygen map given by PFCS precisely captures the oxygen change inside the tumor tissue. In the second part, we apply PFCS to 3D diffusion tensor image (DTI) acquisition. We develop a new sampling strategy specified to diffusion weighted images and optimize the reconstruction cost function for PFCS. The results show that PFCS can reconstruct the accurate color FA map using only 30% of the k-space data. Moreover, PFCS can be further combined with Echo-Planar Imaging (EPI) to achieve an even faster acquisition speed. In summary, PFCS is shown to be a promising image acceleration method in medical imaging which can potentially benefit many clinical applications.
Achievable Rates, Optimal Signalling Schemes and Resource Allocation for Fading Wireless Channels
(2005-08-04) Kaya, Onur; Ulukus, Sennur; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
The proliferation of services involving the transmission of high rate data traffic over wireless channels makes it essential to overcome the detrimental effects of the wireless medium, such as fading and multiuser interference. This thesis is devoted to obtaining optimal resource allocation policies which exploit the transmitters' and receiver's knowledge about the fading to the network's advantage, to attain information theoretic capacity limits of fading wireless channels. The major focus of the thesis is on capacity results for fading code division multiple access (CDMA) channels, which have proved to be a robust way of combatting the multiuser interference in practical wireless networks. For these channels, we obtain the capacity region achievable with power control, as well as the power control policies that achieve the desired rate points on the capacity region. We provide practical one-user-at-a-time iterative algorithms to compute the optimal power distributions as functions of the fading. For the special case of sum capacity, some properties of the optimal policy, such as the number of simultaneously transmitting users, are obtained. We also investigate the effects of limited feedback on the capacity, and demonstrate that very coarse channel state information (CSI) is sufficient to benefit from power control as a means of increasing the capacity. The selection of the signature sequences also plays an important role in determining the capacity of CDMA systems. This thesis addresses the problem of jointly optimizing the signature sequences and power levels to maximize the sum capacity. The resulting policies are shown to be simple, consisting of orthogonal transmissions in time or signal space, and requiring only local CSI. We also provide an iterative way of updating the joint resource allocation policy, and extend our results to asynchronous, and multi-antenna CDMA systems. Rather than treating the received signal at the transmitters as interference, it is possible to treat it as free side information and use it for cooperation. The final part of the thesis provides power allocation policies for a fading Gaussian multiple access channel with user cooperation, which maximize the rates achievable by block Markov superposition coding, and also simplify the coding strategy.
Achieving Utility Arbitrarily Close to the Optimal with Limited Energy
(IEEE, 2000-07) Qu, Gang; Potkonjak, Miodrag
Energy is one of the limited resources for modern systems, especially the battery-operated devices and personal digi- tal assistants. The backlog in new technologies for more powerful battery is changing the traditional system design philosophies. For example, due to the limitation on battery life, it is more realistic to design for the optimal benefit from limited resource rather than design to meet all the applica- tions' requirement. We consider the following problem: a system achieves a certain amount of utility from a set of applications by providing them certain levels of quality of service (QoS). We want to allocate the limited system re- sources to get the maximal system utility. We formulate this utility maximization problem, which is NP-hard in gen- eral, and propose heuristic algorithms that are capable of finding solutions provably arbitrarily close to the optimal. We have also derived explicit formulae to guide the alloca- tion of resources to actually achieve such solutions. Simu- lation shows that our approach can use 99.9% of the given resource to achieve 25.6% and 32.17% more system utilities over two other heuristics, while providing QoS guarantees to the application program.
Active Attention for Target Detection and Recognition in Robot Vision
(2017) Luan, Wentao; Baras, John S; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
In this thesis, we address problems in building an efficient and reliable target detection and recognition system for robot applications, where the vision module is only one component of the overall system executing the task. The different modules interact with each other to achieve the goal. In this interaction, the role of vision is not only to recognize but also to select what and where to process. In other words, attention is an essential process for efficient task execution. We introduce attention mechanisms into the recognition system that serve the overall system at different levels of the integration and formulate four problems as below. At the most basic level of integration, attention interacts with vision only. We consider the problem of detecting a target in an input image using a trained binary classifier of the target and formulate the target detection problem as a sampling process. The goal is to localize the windows containing targets in the image, and attention controls which part of the image to process next. We observe that detectors’ response scores of sampling windows fade gradually from the peak response window in the detection area and approximate this scoring pattern with an exponential de- cay function. Exploiting this property, we propose an active sampling procedure to efficiently detect the target while avoiding an exhaustive and expensive search of all the possible window locations. With more knowledge about the target, we describe the target as template graphs over segmented surfaces. Constraint functions are also defined to find the node and edge’s matching between an input scene graph and target’s template graph. We propose to introduce the recognition early into the traditional candidate proposal process to achieve fast and reliable detection performance. The target detection thence becomes finding subgraphs from the segmented input scene graph that match the template graphs. In this problem, attention provides the order of constraints in checking the graph matching, and a reasonable sequence can help filter out negatives early, thus reducing computational time. We put forward a sub-optimal checking order, and prove that it has bounded time cost compared to the optimal checking sequence, which is not obtainable in polynomial time. Experiments on rigid and non-rigid object detection validate our pipeline. With more freedom in control, we allow the robot to actively choose another viewpoint if the current view cannot deliver a reliable detection and recognition result. We develop a practical viewpoint control system and apply it to two human-robot interaction applications, where the detection task becomes more challenging with the additional randomness from the human. Attention represents an active process of deciding the location of the camera. Our viewpoint selection module not only considers the viewing condition constraints for vision algorithms but also incorporates the low-level robot kinematics to guarantee the reachability of the desired viewpoint. By selecting viewpoints fast using a linear time cost score function, the system can deliver smooth user interaction experience. Additionally, we provide a learning from human demonstration method to obtain the score function parameters that better serves the task’s preference. Finally, when recognition results from multiple sources under different environmental factor are available, attention means how to fuse the observations to get reliable output. We consider the problem of object recognition in 3D using an ensemble of attribute-based classifiers. We propose two new concepts to improve classification in practical situations, and show their implementation in an approach implemented for recognition from point-cloud data. First, we study the impact of the distance between the camera and the object and propose an approach to classifier’s accuracy performance, which incorporates distance into the decision making. Second, to avoid the difficulties arising from lack of representative training examples in learning the optimal threshold, we set in our attribute classifier two threshold values to distinguish a positive, a negative and an uncertainty class, instead of just one threshold value. We prove the theoretical correctness of this approach for an active agent who can observe the object multiple times.
Active Instrument Cables with Buffer Amplification
(2011) Babich, Timothy A.; O'Shea, Patrick; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
ABSTRACT Title of Thesis: ACTIVE INSTRUMENT CABLES WITH BUFFER AMPLIFICATION Timothy A. Babich, Master of Science, 2011 Thesis Directed by: Professor Patrick O'Shea Department of Electrical and Computer Engineering "When connecting electronic instruments to power amplifiers with traditional passive instrument cables signal attenuations can occur due to several factors such as signal reflections and cable capacitance. This thesis examines the possibility of better preserving electronic instrument signals by using buffer amplification built into instrument cables. Active instrument cables were developed that are powered by an internal AA battery in a form factor similar to traditional passive cables. The goal was to achieve a design that was small and had substantial battery life such that the product would be marketable to musicians."
Active microring and microdisk optical resonators on indium phosphide
(2006-04-27) Amarnath, Kuldeep; Ho, Ping-Tong; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Photonic or optical logic holds the promise of ultra-fast logic circuits with capability for speeds beyond what is possible using conventional silicon electronics. However, the jump from theory to practice has a high barrier set by critical issues such as integration, scalability and power requirements. Optical micro-resonator based schemes have the potential of addressing some of these issues. This thesis focuses on the development of active InGaAsP/InP microdisk and microring optical resonators to lower that barrier a little. Microrings and disks provide a compact and cascadable device platform to achieve resonance enhancement of optical non-linearity. By incorporating gain in such devices, the optical power needed for carrying out switching can be greatly reduced. Electrically pumped microring and microdisk resonators are fabricated on indium phosphide in both vertically and laterally coupled bus-waveguide configurations. The gain saturation non-linearity is used to demonstrate all-optical switching and bistable operation at optical powers more than two orders of magnitude lower compared to passive devices. The shift in the ring/disk resonances caused by the refractive index change due to a pump beam is used to switch a weaker probe beam tuned to one of the resonances. The non-linear response and switching mechanism is modeled numerically. A novel pseudodisk configuration that combines the best of microdisks and microrings is used to minimize device heating and surface recombination as well as provide near single-mode operation. Additionally, optical amplifiers based on microrings are also developed for cascading passive optical gates. Optical amplification up to 10 db in pulsed mode has been observed for 20 µm radius microrings. The control of surface recombination on the microring sidewalls is critical to avoid carrier loss and device heating. A sulfur passivation scheme is used to reduce the surface recombination velocity. The lateral carrier transport and surface recombination in microrings is analyzed by an ambipolar diffusion model.
Active Power Decoupling (APD) Converter for PV Microinverter Applications
(2024) Shen, Yidi; Khaligh, Alireza; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Under global challenges in climate change, the demand for renewable energy is continuously growing. Photovoltaic (PV) power and its integration into the utility grid are gaining increasing traction. To lower the levelized cost of energy (LCOE) of PV systems, enhance the adoption of PV applications, and ensure the delivery of high-quality power to the utility grid, there is a growing need for reliable, cost-effective, efficient, and compact PV inverters. One key challenge in single-phase PV systems is the short lifetime and poor reliability of electrolytic capacitors used for decoupling the double line frequency (DLF) power. To eliminate the less reliable electrolytic capacitor, the active power decoupling (APD) technique is widely adopted. Various topologies can be used for APD, but the selection of proper topology, modulation scheme, and circuit components, along with the control strategy, will enhance the efficiency, power density, reliability, and cost of the overall PV microinverter. This Ph.D. dissertation proposes an APD converter circuit suitable for PV microinverters, designed for optimized efficiency, power density, and cost. The proposed APD converter is controlled to achieve good power decoupling performance and to optimize the system's maximum power point tracking (MPPT) efficiency. The proposed APD converter circuit is analyzed in the low-frequency domain for power flow and in the high-frequency domain for modulation strategy, where different topologies are considered, taking into account the voltage and current ratings of active devices and decoupling capacitors. Two modulation approaches, continuous conduction mode (CCM) and critical conduction mode (CRM), are compared, considering detailed zero voltage switching (ZVS) operation and different loss mechanisms. Parametric design and multi-objective optimization are performed for CCM and CRM to select circuit components and switching frequency for each modulation strategy to minimize power loss, volume, and costs. With the results of multi-objective optimization, Pareto-optimal designs for CCM and CRM are analyzed in terms of the impact of various circuit elements, namely: switching device output capacitance and on-state resistance, inductor winding turns and core geometries, as well as capacitor dimensions and capacitance. With the optimal CCM- and CRM-operated APD realizations, closed-loop control algorithms are designed, and the corresponding system characteristics are compared. A simple pulse width modulation (PWM) based control strategy that does not rely on zero-crossing detection (ZCD) is proposed to implement closed-loop CRM modulation. In addition, advanced control technologies, including double sampling-based average current control, current observer-based reduced sensor control, and sensorless predictive control, are proposed to improve APD converter performance, reduce system complexity, and lower circuit cost. The proposed APD converter operation is extended to different application scenarios, including burst-mode operation and non-sinusoidal power delivery, including systems with non-linear circuit components, non-linear local loads, or non-ideal grids. A feed-forward control solution is proposed to enable power decoupling for non-sinusoidal power with improved control accuracy and reduced closed-loop design burden. The circuit design, associated analyses, and control approaches are validated by the design, development, and testing of 400 VA APD hardware prototypes.
Activity Detection in Untrimmed Videos
(2023) Gleason, Joshua D; Chellappa, Rama; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
In this dissertation, we present solutions to the problem of activity detection in untrimmed videos, where we are interested in identifying both when and where various activity instances occur within an unconstrained video. Advances in machine learning, particularly the widespread adoption of deep learning-based methods have yielded robust solutions to a number of historically difficult computer vision application domains. For example, recent systems for object recognition and detection, facial identification, and a number of language processing applications have found widespread commercial success. In some cases, such systems have been able to outperform humans. The same cannot be said for the problem of activity detection in untrimmed videos. This dissertation describes our investigation and innovative solutions for the challenging problem of real-time activity detection in untrimmed videos. The main contributions of our work are the introduction of multiple novel activity detection systems that make strides toward the goal of commercially viable activity detection. The first work introduces a proposal mechanism based on divisive hierarchical clustering of objects to produce cuboid activity proposals, followed by a classification and temporal refinement step. The second work proposes a chunk-based processing mechanism and explores the tradeoff between tube and cuboid proposals. The third work explores the topic of real-time activity detection and introduces strategies for achieving this performance. The final work provides a detailed look into multiple novel extensions that improve upon the state-of-the-art in the field.
Activity Representation from Video Using Statistical Models on Shape Manifolds
(2010) Abdelkader, Mohamed F.; Chellappa, Rama; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Activity recognition from video data is a key computer vision problem with applications in surveillance, elderly care, etc. This problem is associated with modeling a representative shape which contains significant information about the underlying activity. In this dissertation, we represent several approaches for view-invariant activity recognition via modeling shapes on various shape spaces and Riemannian manifolds. The first two parts of this dissertation deal with activity modeling and recognition using tracks of landmark feature points. The motion trajectories of points extracted from objects involved in the activity are used to build deformation shape models for each activity, and these models are used for classification and detection of unusual activities. In the first part of the dissertation, these models are represented by the recovered 3D deformation basis shapes corresponding to the activity using a non-rigid structure from motion formulation. We use a theory for estimating the amount of deformation for these models from the visual data. We study the special case of ground plane activities in detail because of its importance in video surveillance applications. In the second part of the dissertation, we propose to model the activity by learning an affine invariant deformation subspace representation that captures the space of possible body poses associated with the activity. These subspaces can be viewed as points on a Grassmann manifold. We propose several statistical classification models on Grassmann manifold that capture the statistical variations of the shape data while following the intrinsic Riemannian geometry of these manifolds. The last part of this dissertation addresses the problem of recognizing human gestures from silhouette images. We represent a human gesture as a temporal sequence of human poses, each characterized by a contour of the associated human silhouette. The shape of a contour is viewed as a point on the shape space of closed curves and, hence, each gesture is characterized and modeled as a trajectory on this shape space. We utilize the Riemannian geometry of this space to propose a template-based and a graphical-based approaches for modeling these trajectories. The two models are designed in such a way to account for the different invariance requirements in gesture recognition, and also capture the statistical variations associated with the contour data.
Adaptation in Standard CMOS Processes with Floating Gate Structures and Techniques
(2007-04-25) Wong, Yanyi Liu; Abshire, Pamela A; Abshire, Pamela A; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
We apply adaptation into ordinary circuits and systems to achieve high performance, high quality results. Mismatch in manufactured VLSI devices has been the main limiting factor in quality for many analog and mixed-signal designs. Traditional compensation methods are generally costly. A few examples include enlarging the device size, averaging signals, and trimming with laser. By applying floating gate adaptation to standard CMOS circuits, we demonstrate here that we are able to trim CMOS comparator offset to a precision of 0.7mV, reduce CMOS image sensor fixed-pattern noise power by a factor of 100, and achieve 5.8 effective number of bits (ENOB) in a 6-bit flash analog-to-digital converter (ADC) operating at 750MHz. The adaptive circuits generally exhibit special features in addition to an improved performance. These special features are generally beyond the capabilities of traditional CMOS design approaches and they open exciting opportunities in novel circuit designs. Specifically, the adaptive comparator has the ability to store an accurate arbitrary offset, the image sensor can be set up to memorize previously captured scenes like a human retina, and the ADC can be configured to adapt to the incoming analog signal distribution and perform an efficient signal conversion that minimizes distortion and maximizes output entropy.
Adaptive Analysis and Processing of Structured Multilingual Documents
(2006-01-19) Ma, Huanfeng; Chellappa, Rama; Doermann, David S; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Digital document processing is becoming popular for application to office and library automation, bank and postal services, publishing houses and communication management. In recent years, the demand for tools capable of searching written and spoken sources of multilingual information has increased tremendously, where the bilingual dictionary is one of the important resource to provide the required information. Processing and analysis of bilingual dictionaries brought up the challenges of dealing with many different scripts, some of which are unknown to the designer. A framework is presented to adaptively analyze and process structured multilingual documents, where adaptability is applied to every step. The proposed framework involves: (1) General word-level script identification using Gabor filter. (2) Font classification using the grating cell operator. (3) General word-level style identification using Gaussian mixture model. (4) An adaptable Hindi OCR based on generalized Hausdorff image comparison. (5) Retargetable OCR with automatic training sample creation and its applications to different scripts. (6) Bootstrapping entry segmentation, which segments each page into functional entries for parsing. Experimental results working on different scripts, such as Chinese, Korean, Arabic, Devanagari, and Khmer, demonstrate that the proposed framework can save human efforts significantly by making each phase adaptive.
Adaptive Log Domain Filters Using Floating Gate Transistors
(2004-12-06) Zhai, Yiming; Abshire, Pamela; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
In this thesis, an adaptive first order lowpass log domain filter and an adaptive second order log domain filter are presented with integrated learning rules for model reference estimation. Both systems are implemented using multiple input floating gate transistors to realize on-line learning of system parameters. Adaptive dynamical system theory is used to derive robust control laws in a system identification task for the parameters of both a first order lowpass filter and a second order tunable filter. The log domain filters adapt to estimate the parameters of the reference filters accurately and efficiently as the parameters are changed. Simulation results for both the first order and the second order adaptive filters are presented which demonstrate that adaptation occurs within milliseconds. Experimental results and mismatch analysis are described for the first order lowpass filter which demonstrates the success of our adaptive system design using this model-based learning method.
An Adaptive Mac Protocol for Wireless Multi-Hop Networks with Multiple Antennas
(2005-08-10) Li, Haipeng; Gligor, Virgil D.; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
Radio links that use multiple antennas at both transmitter and receiver sides are referred to as Multiple-Input Multiple-Output (MIMO) links. MIMO links are known to provide multiplicative increase in capacity and spectral efficiency by simultaneously transmitting multiple independent data streams in the same channel. However, current medium access control (MAC) protocols can't fully exploit the bandwidth and capacity of the MIMO links. In this thesis, we present a new MAC protocol, Achieving Maximum Transmit Antenna MAC (AMTA-MAC), which can fully utilize the feature of MIMO links to achieve better performance in terms of fairness and throughput. We implement the AMTA-MAC protocol in the network simulator ns-2 for a system with two antennas. Simulation results show that the AMTA-MAC outperforms the throughput of IEEE 802.11 and MIMA-MAC and mitigates the unfairness problem of IEEE 802.11.
Adaptive Sensing and Processing for Some Computer Vision Problems
(2014) Warnell, Garrett; Chellappa, Rama; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
This dissertation is concerned with adaptive sensing and processing in computer vision, specifically through the application of computer vision techniques to non-standard sensors. In the first part, we adapt techniques designed to solve the classical computer vision problem of gradient-based surface reconstruction to the problem of phase unwrapping that presents itself in applications such as interferometric synthetic aperture radar. Specifically, we propose a new formulation of and solution to the classical two-dimensional phase unwrapping problem. As is usually done, we use the wrapped principal phase gradient field as a measurement of the absolute phase gradient field. Since this model rarely holds in practice, we explicitly enforce integrability of the gradient measurements through a sparse error-correction model. Using a novel energy-minimization functional, we formulate the phase unwrapping task as a generalized lasso problem. We then jointly estimate the absolute phase and the sparse measurement errors using the alternating direction method of multipliers (ADMM) algorithm. Using an interferometric synthetic aperture radar noise model, we evaluate our technique for several synthetic surfaces and compare the results to recently-proposed phase unwrapping techniques. Our method applies new ideas from convex optimization and sparse regularization to this well-studied problem. In the second part, we consider the problem of controlling and processing measurements from a non-traditional, compressive sensing (CS) camera in real time. We focus on how to control the number of measurements it acquires such that this number remains proportional to the amount of foreground information currently present in the scene under observations. To this end, we provide two novel adaptive-rate CS strategies for sparse, time-varying signals using side information. The first method utilizes extra cross-validation measurements, and the second exploits extra low-resolution measurements. Unlike the majority of current CS techniques, we do not assume that we know an upper bound on the number of significant coefficients pertaining to the images that comprise the video sequence. Instead, we use the side information to predict this quantity for each upcoming image. Our techniques specify a fixed number of spatially-multiplexed CS measurements to acquire, and they adjust this quantity from image to image. Our strategies are developed in the specific context of background subtraction for surveillance video, and we experimentally validate the proposed methods on real video sequences. Finally, we consider a problem motivated by the application of active pan-tilt-zoom (PTZ) camera control in response to visual saliency. We extend the classical notion of this concept to multi-image data collected using a stationary PTZ camera by requiring consistency: the property that each saliency map in the set of those that are generated should assign the same saliency value to distinct regions of the environment that appear in more than one image. We show that processing each image independently will often fail to provide a consistent measure of saliency, and that using an image mosaic to quantify saliency suffers from several drawbacks. We then propose ray saliency: a mosaic-free method for calculating a consistent measure of bottom-up saliency. Experimental results demonstrating the effectiveness of the proposed approach are presented.
AN ADAPTIVELY SAMPLED PATH PLANNER USING WAYPOINTS: AN ANY-ANGLE VARIANT
(2014) Gefen, Yonatan; Martins, Nuno C; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
This thesis develops a low-cost grid-based path planner that intrinsically supports smooth, curved vehicle dynamics. There are many advantages to grid-based planners, including working natively in the digital space of most sensors, and efficiency in low dimensional space. However, discrete planners create jaggedness in most paths. Further, the dimensionality must be limited for efficiency, usually by limiting vehicle steering angles to a small finite set. The algorithm presented here, Waypoint-A*, extends A* to produce low-cost curved trajectories, taking the dynamics of the vehicle into account explicitly post-planning. Considering the path generated by A* as composed of a set of waypoints, Waypoint-A* calculates the minimum-cost heading on a continuum, to direct the vehicle to the waypoint at the location resulting in the lowest total cost. Smoothness of these curves is invariant to terrain resolution and computation.
The ADI-FDTD Method for High Accuracy Electrophysics Applications
(2006-11-24) Haeri Kermani, Mohammad; Ramahi, Omar M; Electrical Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)
The Finite-Difference Time-Domain (FDTD) is a dependable method to simulate a wide range of problems from acoustics, to electromagnetics, and to photonics, amongst others. The execution time of an FDTD simulation is inversely proportional to the time-step size. Since the FDTD method is explicit, its time-step size is limited by the well-known Courant-Friedrich-Levy (CFL) stability limit. The CFL stability limit can render the simulation inefficient for very fine structures. The Alternating Direction Implicit FDTD (ADI-FDTD) method has been introduced as an unconditionally stable implicit method. Numerous works have shown that the ADI-FDTD method is stable even when the CFL stability limit is exceeded. Therefore, the ADI-FDTD method can be considered an efficient method for special classes of problems with very fine structures or high gradient fields. Whenever the ADI-FDTD method is used to simulate open-region radiation or scattering problems, the implementation of a mesh-truncation scheme or absorbing boundary condition becomes an integral part of the simulation. These truncation techniques represent, in essence, differential operators that are discretized using a distinct differencing scheme which can potentially affect the stability of the scheme used for the interior region. In this work, we show that the ADI-FDTD method can be rendered unstable when higher-order mesh truncation techniques such as Higdon's Absorbing Boundary Condition (ABC) or Complementary Derivatives Method (COM) are used. When having large field gradients within a limited volume, a non-uniform grid can reduce the computational domain and, therefore, it decreases the computational cost of the FDTD method. However, for high-accuracy problems, different grid sizes increase the truncation error at the boundary of domains having different grid sizes. To address this problem, we introduce the Complementary Derivatives Method (CDM), a second-order accurate interpolation scheme. The CDM theory is discussed and applied to numerical examples employing the FDTD and ADI-FDTD methods.

Browse

Browsing Electrical & Computer Engineering by Title

Results Per Page

Sort Options