Scholarly Works, Electrical and Computer Engineering

Permanent URI for this collection

Research articles, presentations, and other scholarship

Browse

Recent Submissions

Now showing 1 - 20 of 849
  • Hybrid Modular Multilevel Converters for High-AC/Low-DC Medium-Voltage Applications
    Motwani, Jayesh Kumar; Liu, Jian; Boroyevich, Dushan; Burgos, Rolando; Zhou, Zhi; Dong, Dong (IEEE, 2024-02-12)
    With ever-increasing power-density requirements, technologies such as energy storage systems and electric-vehicles can benefit greatly from interfacing medium-voltage (MV)-AC grid like 13.8kV or 30kV using high-AC/low-DC voltage converter. Using modular high-AC/low-DC voltage converter can help increase power-density and efficiency, while reducing total conversion steps and providing flexibility. Full-bridge modular multilevel converters (FB-MMC) and solid-state transformers are existing solutions for such operations, but suffer from limitations of high semiconductor requirements, large submodule capacitors and/or many high-frequency transformers. Three new hybrid-MMC (HMMC) topologies are proposed in this paper as alternative solutions for such high-AC/low-DC voltage operations. Each of the three developed HMMCs utilizes a unique combination of low-frequency high-voltage switches and fast-switching lowvoltage switch based submodules to generate multilevel-AC voltage. HMMCs are compared extensively to state-of-the-art FB-MMC and are shown to have semiconductor savings of over 27%, 38% lower submodule capacitor size, and 53% lower losses for 13.8-kV-AC/6-kV-DC operation. Due to these benefits like higher efficiency, significantly smaller submodule capacitance requirements, and fewer semiconductors, HMMCs can be an excellent option for high-AC/low-DC applications. Practical considerations like snubber and DC split-capacitor requirement are also elaborated for developing and commercializing HMMCs. Comparison results are verified using a 17.5 kW three-phase MV laboratory prototype.
  • Edge-Connected Jaccard Similarity for Graph Link Prediction on FPGA
    Sathre, Paul; Gondhalekar, Atharva; Feng, Wu-chun (IEEE, 2022-01-01)
    Graph analysis is a critical task in many fields, such as social networking, epidemiology, bioinformatics, and fraud de-tection. In particular, understanding and inferring relationships between graph elements lies at the core of many graph-based workloads. Real-world graph workloads and their associated data structures create irregular computational patterns that compli-cate the realization of high-performance kernels. Given these complications, there does not exist a de facto 'best' architecture, language, or algorithmic approach that simultaneously balances performance, energy efficiency, portability, and productivity. In this paper, we realize different algorithms of edge-connected Jaccard similarity for graph link prediction and characterize their performance across a broad spectrum of graphs on an Intel Stratix 10 FPGA. By utilizing a high-level synthesis (HLS)-driven, high-productivity approach (via the C++-based SYCL language) we rapidly prototype two implementations - a from-scratch edge-centric version and a faithfully-ported commodity GPU implementation - which would have been intractable via a hardware description language. With these implementations, we further consider the benefit and necessity of four HLS-enabled optimizations, both in isolation and in concert - totaling seven distinct synthesized hardware pipelines. Leveraging real-world graphs of up to 516 million edges, we show empirically-measured speedups of up to 9.5 x over the initial HLS implementations when all optimizations work in concert.
  • Scaling out a combinatorial algorithm for discovering carcinogenic gene combinations to thousands of GPUs
    Dash, Sajal; Al-Hajri, Qais; Feng, Wu-chun; Garner, Harold R.; Anandakrishnan, Ramu (IEEE, 2021-05-01)
    Cancer is a leading cause of death in the US, second only to heart disease. It is primarily a result of a combination of an estimated two-nine genetic mutations (multi-hit combinations). Although a body of research has identified hundreds of cancer-causing genetic mutations, we don't know the specific combination of mutations responsible for specific instances of cancer for most cancer types. An approximate algorithm for solving the weighted set cover problem was previously adapted to identify combinations of genes with mutations that may be responsible for individual instances of cancer. However, the algorithm's computational requirement scales exponentially with the number of genes, making it impractical for identifying more than three-hit combinations, even after the algorithm was parallelized and scaled up to a V100 GPU. Since most cancers have been estimated to require more than three hits, we scaled out the algorithm to identify combinations of four or more hits using 1000 nodes (6000 V100 GPUs with ≈ 48× 106 processing cores) on the Summit supercomputer at Oak Ridge National Laboratory. Efficiently scaling out the algorithm required a series of algorithmic innovations and optimizations for balancing an exponentially divergent workload across processors and for minimizing memory latency and inter-node communication. We achieved an average strong scaling efficiency of 90.14% (80.96%-97.96% for 200 to 1000 nodes), compared to a 100 node run, with 84.18% scaling efficiency for 1000 nodes. With experimental validation, the multi-hit combinations identified here could provide further insight into the etiology of different cancer subtypes and provide a rational basis for targeted combination therapy.
  • Electrically Small Antennas' Design Criteria and Measurement Challenges
    Manteghi, Majid (2023-05-24)
    The contrast between the design criteria for electrically small transmit and receive antennas is studied in this work. On the transmit side, radiation efficiency (ohmic loss plus return loss) and data throughput are critical. However, a higher impedance mismatch on the receiving front end may reduce ohmic loss and expand the frequency bandwidth. So, a calculated mismatch can be added to improve the performance of the receiving ESA by lowering the overall noise figure and widening the frequency bandwidth. These contradictory design criteria suggest utilizing separate transmit and receive antennas to improve the transmitter and receiver performances.
  • On the Role of Uncertainty in Poisson Target Models Used for Placement of Spatial Sensors
    Kim, Mingyu; Yetkin, Harun; Stilwell, Daniel J.; Jimenez, Jorge (SPIE, 2023-01-01)
    This paper addresses the role of uncertainty in spatial point-process models, such as those that might arise in modelling ship traffic. We consider a doubly stochastic Poisson point process where the intensity function is uncertain. To assess the role of uncertainty, we conduct a large set of numerical trials where we estimate a doubly stochastic Poisson point-process model from historical target data, and the evaluate the model by assessing the target detection performance of a set of sensors whose locations are selected using the model. Our work is motivated by seabed sensors that detect ship traffic, and we conduct numerical trials using historical ship traffic data near the mouth of the Chesapeake Bay, Virginia, USA, that was recorded by the Automated Identification System.
  • Single-Image 3D Human Digitization with Shape-guided Diffusion
    Albahar, Badour; Saito, Shunsuke; Tseng, Hung-Yu; Kim, Changil; Kopf, Johannes; Huang, Jia-Bin (ACM, 2023-12-10)
    We present an approach to generate a 360-degree view of a person with a consistent, high-resolution appearance from a single input image. NeRF and its variants typically require videos or images from different viewpoints. Most existing approaches taking monocular input either rely on ground-truth 3D scans for supervision or lack 3D consistency. While recent 3D generative models show promise of 3D consistent human digitization, these approaches do not generalize well to diverse clothing appearances, and the results lack photorealism. Unlike existing work, we utilize high-capacity 2D diffusion models pretrained for general image synthesis tasks as an appearance prior of clothed humans. To achieve better 3D consistency while retaining the input identity, we progressively synthesize multiple views of the human in the input image by inpainting missing regions with shape-guided diffusion conditioned on silhouette and surface normal. We then fuse these synthesized multi-view images via inverse rendering to obtain a fully textured high-resolution 3D mesh of the given person. Experiments show that our approach outperforms prior methods and achieves photorealistic 360-degree synthesis of a wide range of clothed humans with complex textures from a single image.
  • Towards Energy-Efficient Spiking Neural Networks: A Robust Hybrid CMOS-Memristive Accelerator
    Nowshin, Fabiha; An, Hongyu; Yi, Yang (ACM, 2024)
    Spiking Neural Networks (SNNs) are energy-efficient artificial neural network models that can carry out data-intensive applications. Energy consumption, latency, and memory bottleneck are some of the major issues that arise in machine learning applications due to their data-demanding nature. Memristor-enabled Computing-In-Memory (CIM) architectures have been able to tackle the memory wall issue, eliminating the energy and time-consuming movement of data. In this work we develop a scalable CIM-based SNN architecture with our fabricated two-layer memristor crossbar array. In addition to having an enhanced heat dissipation capability, our memristor exhibits substantial enhancement of 10% to 66% in design area, power and latency compared to state-of-the-art memristors. This design incorporates an inter-spike interval (ISI) encoding scheme due to its high information density to convert the incoming input signals into spikes. Furthermore, we include a time-to-first-spike (TTFS) based output processing stage for its energy-efficiency to carry out the final classification. With the combination of ISI, CIM and TTFS, this network has a competitive inference speed of 2?s/image and can successfully classify handwritten digits with 2.9mW of power and 2.51pJ energy per spike. The proposed architecture with the ISI encoding scheme can achieve ~10% higher accuracy than those of other encoding schemes in the MNIST dataset.
  • Dynamics of mid-latitude sporadic-E and its impact on HF propagation in the North American sector
    Kunduri, Bharat Simha Reddy; Erickson, Philip; Baker, Joseph; Ruohoniemi, John (American Geophysical Union, 2023-09-16)
    Sporadic-E (Es) are thin layers of enhanced ionization observed in the E-region, typically between 95 and 120 km altitude. Es plays an important role in controlling the dynamics of the upper atmosphere and it is necessary to understand the geophysical factors influencing Es from both the scientific and operational perspectives. While the wind-shear theory is widely accepted as an important mechanism responsible for the generation of Es, there are still gaps in the current state of our knowledge. For example, we are yet to determine precisely how changes in the dynamics of horizontal winds impact the formation, altitude, and destruction of Es layers. In this study, we report results from a coordinated experimental campaign between the Millstone Hill Incoherent Scatter Radar, the SuperDARN radar at Blackstone, and the Millstone Hill Digisonde to monitor the dynamics of mid-latitude Es layers. We report observations during a 15-hr window between 13 UT on 3 June 2022 and 4 UT on 4 June 2022, which was marked by the presence of a strong Es layer. We find that the height of the Es layer is collocated with strong vertical shears in atmospheric tides and that the zonal wind shears play a more important role than meridional wind shears in generating Es, especially at lower altitudes. Finally, we show that in the presence of Es, SuperDARN ground backscatter moves to closer ranges, and the height and critical frequency of the Es layer have a significant impact on the location and intensity of HF ground scatter.
  • Building a Statewide Experiential Learning Portfolio in Cybersecurity
    DaSilva, Luiz A.; Durant, Liza Wilson; Mason, Jordan; Hayes, Sarah (2023-06-25)
    The growing workforce gap in cybersecurity, with an estimated 770 thousand job openings across the country, poses economic and national security risks. Meanwhile, women, African Americans, Native Americans, and Latinos are significantly underrepresented in the cyber workforce. With these two challenges in mind, and informed by research findings that experiential learning opportunities correlate with multiple positive job outcomes, we have built a statewide experiential learning portfolio open to students in more than 40 two-year and four-year colleges and universities across Virginia. Programs in our experiential learning portfolio generally fall under one of five categories: transdisciplinary experiential learning; internships; traineeships; cybersecurity competitions; and intensive training coupled with professional development activities. In this paper, we describe the structure of these programs and associated metrics. Early results indicate very high interest by students and employers, high retention rates in cybersecurity careers, and gains in participation by underrepresented groups.
  • STAMINA: Implementation and Evaluation of Software-Defined Millimeter Wave Initial Access
    Santos, Joao F.; Fathalla, Efat; Da Silva, Aloizio P.; Da Silva, Luiz A.; Kibilda, Jacek (IEEE, 2023-01-01)
    In this paper, we present a framework for experimentation in next-generation Initial Access (IA) procedures for Millimeter Wave (mmWave) and Terahertz (THz) communications called SofTwAre-defined Mmwave INitial Access (STAMINA). The IA procedure is one of the essential components for communication systems in high frequencies, enabling directional transmitters and receivers to acquire each other's relative orientation before data transmission. While effective in establishing communication, the existing IA procedure standardized by 3GPP consumes a significant amount of radio resources. Many research efforts have proposed enhancements over the current-generation IA procedure, e.g., leveraging non-uniform beam sweep sequences or adaptive codebooks. However, no existing experimental mmWave platforms support modifications in their standard-compliant IA procedures, preventing their utilization for conducting experimental research on next-generation IA procedures. Our software-defined mmWave framework addresses this gap by combining the flexibility of Software-defined Radios (SDRs) with the directionality of mmWave front-ends to perform customizable IA procedures. We demonstrate STAMINA's ability to control mmWave frontends correctly, its increased performance over traditional static experiments, and its flexibility to customize the IA parameters to achieve different objectives. Our results show that STAMINA provides experimenters with a flexible platform for performing experiments on next-generation IA procedures.
  • Parallelizable synthesis of arbitrary single-qubit gates with linear optics and time-frequency encoding
    Henry, Antoine; Raghunathan, Ravi; Ricard, Guillaume; Lefaucher, Baptiste; Miatto, Filippo; Belabas, Nadia; Zaquine, Isabelle; Alleaume, Romain (American Physical Society, 2023-06-26)
    We propose methods for the exact synthesis of single-qubit unitaries with high success probability and gate fidelity, considering both time-bin and frequency-bin encodings. The proposed schemes are experimentally implementable with a spectral linear-optical quantum computation (S-LOQC) platform, composed of electro-optic phase modulators and phase-only programmable filters (pulse shapers). We assess the performances in terms of fidelity and probability of the two simplest three-component configurations for arbitrary gate generation in both encodings and give an exact analytical solution for the synthesis of an arbitrary single-qubit unitary in the time-bin encoding, using a single-tone rf driving of the electro-optic modulators. We further investigate the parallelization of arbitrary single-qubit gates over multiple qubits with a compact experimental setup, both for spectral and temporal encodings. We systematically evaluate and discuss the impact of the rf bandwidth, which conditions the number of tones driving the modulators, and of the choice of encoding for different targeted gates. We moreover quantify the number of high-fidelity Hadamard gates that can be synthesized in parallel, with minimal and increasing resources in terms of driving rf tones in a realistic system. Our analysis positions spectral S-LOQC as a promising platform to conduct massively parallel single-qubit operations, with potential applications to quantum metrology and quantum tomography.
  • Reservoir based spiking models for univariate Time Series Classification
    Gaurav, Ramashish; Stewart, Terrence C.; Yi, Yang (Frontiers, 2023-06-08)
    A variety of advanced machine learning and deep learning algorithms achieve state-of-the-art performance on various temporal processing tasks. However, these methods are heavily energy inefficient—they run mainly on the power hungry CPUs and GPUs. Computing with Spiking Networks, on the other hand, has shown to be energy efficient on specialized neuromorphic hardware, e.g., Loihi, TrueNorth, SpiNNaker, etc. In this work, we present two architectures of spiking models, inspired from the theory of Reservoir Computing and Legendre Memory Units, for the Time Series Classification (TSC) task. Our first spiking architecture is closer to the general Reservoir Computing architecture and we successfully deploy it on Loihi; the second spiking architecture differs from the first by the inclusion of non-linearity in the readout layer. Our second model (trained with Surrogate Gradient Descent method) shows that non-linear decoding of the linearly extracted temporal features through spiking neurons not only achieves promising results, but also offers low computation-overhead by significantly reducing the number of neurons compared to the popular LSM based models—more than 40x reduction with respect to the recent spiking model we compare with. We experiment on five TSC datasets and achieve new SoTA spiking results (—as much as 28.607% accuracy improvement on one of the datasets), thereby showing the potential of our models to address the TSC tasks in a green energy-efficient manner. In addition, we also do energy profiling and comparison on Loihi and CPU to support our claims.
  • Portable, low-cost samplers for distributed sampling of atmospheric gases
    Hurley, James; Caceres, Alejandra; McGlynn, Deborah; Tovillo, Mary; Pinar, Suzanne; Schuerch, Roger; Onufrieva, Ksenia; Isaacman-VanWertz, Gabriel (2023-10-13)
    Volatile organic compounds (VOCs) contribute to air pollution both directly, as hazardous gases, and through their reactions with common atmospheric oxidants to produce ozone, particulate matter, and other hazardous air pollutants. There are enormous ranges of structures and reaction rates among VOCs, and consequently a need to accurately characterize the spatial and temporal distribution of individual identified compounds. Current VOC measurements are often made with complex, expensive instrumentation that provides high chemical detail, but is limited in its portability and requires high expense (e.g., mobile labs) for spatially resolved measurements. Alternatively, periodic collection of samples on cartridges is inexpensive but demands significant operator interaction that can limit possibilities for time-resolved measurements or distributed measurements across a spatial area. Thus, there is a need for simple, portable devices that can sample with limited operator presence to enable temporally and/or spatially resolved measurements. In this work, we describe new portable and programmable VOC samplers that enable simultaneous collection of samples across a spatially distributed network, validate their reproducibility, and demonstrate their utility. Validation experiments confirmed high precision between samplers as well as the ability of miniature ozone scrubbers to preserve reactive analytes collected on commercially available adsorbent gas sampling cartridges, supporting simultaneous field deployment across multiple locations. In indoor environments, 24-hour integrated samples demonstrate observable day-to-day variability, as well as variability across very short spatial scales (meters). The utility of the samplers was further demonstrated by locating outdoor point sources of analytes through the development of a new mapping approach that employs a group of the portable samplers and back projection techniques to assess a sampling area with higher resolution than stationary sampling. As with all gas sampling, the limits of detection depend on sampling times and the properties of sorbent and analyte. Limit of detection of the analytical system used in this work is on the order of nanograms, corresponding to mixing ratios of 1-10 pptv after one hour of sampling at the programmable flow rate of 50-250 sccm enabled by the developed system. The portable VOC samplers described and validated here provide a simple, low-cost sampling solution for spatially and/or temporally variable measurements of any organic gases that are collectable on currently available sampling media.
  • Properties of a Random Bipartite Geometric Associator Graph Inspired by Vehicular Networks
    Pandey, Kaushlendra; Gupta, Abhishek K.; Dhillon, Harpreet S.; Perumalla, Kanaka Raju (MDPI, 2023-12-04)
    We consider a point process (PP) generated by superimposing an independent Poisson point process (PPP) on each line of a 2D Poisson line process (PLP). Termed PLP-PPP, this PP is suitable for modeling networks formed on an irregular collection of lines, such as vehicles on a network of roads and sensors deployed along trails in a forest. Inspired by vehicular networks in which vehicles connect with their nearest wireless base stations (BSs), we consider a random bipartite associator graph in which each point of the PLP-PPP is associated with the nearest point of an independent PPP through an edge. This graph is equivalent to the partitioning of PLP-PPP by a Poisson Voronoi tessellation (PVT) formed by an independent PPP. We first characterize the exact distribution of the number of points of PLP-PPP falling inside the ball centered at an arbitrary location in R2 as well as the typical point of PLP-PPP. Using these distributions, we derive cumulative distribution functions (CDFs) and probability density functions (PDFs) of kth contact distance (CD) and the nearest neighbor distance (NND) of PLP-PPP. As intermediate results, we present the empirical distribution of the perimeter and approximate distribution of the length of the typical chord of the zero-cell of this PVT. Using these results, we present two close approximations of the distribution of node degree of the random bipartite associator graph. In a vehicular network setting, this result characterizes the number of vehicles connected to each BS, which models its load. Since each BS has to distribute its limited resources across all the vehicles connected to it, a good statistical understanding of load is important for an efficient system design. Several applications of these new results to different wireless network settings are also discussed.
  • Exact and Paraxial Broadband Airy Wave Packets in Free Space and a Temporally Dispersive Medium
    Besieris, Ioannis M.; Saari, Peeter (MDPI, 2024-01-21)
    A question of physical importance is whether finite-energy spatiotemporally localized (i.e., pulsed) generalizations of monochromatic accelerating Airy beams are feasible. For luminal solutions, this question has been answered within the framework of paraxial geometry. The time-diffraction technique that has been motivated by the Lorentz invariance of the equation governing the narrow angular spectrum and narrowband temporal spectrum paraxial approximation has been used to derive finite-energy spatiotemporally confined subluminal, luminal, and superluminal Airy wave packets. The goal in this article is to provide novel exact finite-energy broadband spatio-temporally localized Airy solutions (a) to the scalar wave equation in free space; (b) in a dielectric medium moving at its phase velocity; and (c) in a lossless second-order temporally dispersive medium. Such solutions can be useful in practical applications involving broadband (few-cycle) wave packets.
  • Room-Temperature Intrinsic and Extrinsic Damping in Polycrystalline Fe Thin Films
    Wu, Shuang; Smith, David A.; Nakarmi, Prabandha; Rai, Anish; Clavel, Michael; Hudait, Mantu K.; Zhao, Jing; Michel, F. Marc; Mewes, Claudia; Mewes, Tim; Emori, Satoru (2021-09-08)
    We examine room-temperature magnetic relaxation in polycrystalline Fe films. Out-of-plane ferromagnetic resonance (FMR) measurements reveal Gilbert damping parameters of $\approx$ 0.0024 for Fe films with thicknesses of 4-25 nm, regardless of their microstructural properties. The remarkable invariance with film microstructure strongly suggests that intrinsic Gilbert damping in polycrystalline metals at room temperature is a local property of nanoscale crystal grains, with limited impact from grain boundaries and film roughness. By contrast, the in-plane FMR linewidths of the Fe films exhibit distinct nonlinear frequency dependences, indicating the presence of strong extrinsic damping. To fit our in-plane FMR data, we have used a grain-to-grain two-magnon scattering model with two types of correlation functions aimed at describing the spatial distribution of inhomogeneities in the film. However, neither of the two correlation functions is able to reproduce the experimental data quantitatively with physically reasonable parameters. Our findings advance the fundamental understanding of intrinsic Gilbert damping in structurally disordered films, while demonstrating the need for a deeper examination of how microstructural disorder governs extrinsic damping.