{"title":"Tracking a Moving Object for Tomographic Below Ground Imaging","authors":"Abdulhakim Daluom, M. Wicks","doi":"10.1109/NAECON.2018.8556813","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556813","url":null,"abstract":"In this paper, we present a Ground Penetrating Radar (GPR) technique currently under development to image a moving object. This GPR system localizes and tracks a moving spherical object that could be in a plastic pipe under the ground for a given bistatic and multi-static radar geometry. Two different approaches (two dyadic Green’s functions) are applied to image a moving object using Maxwell’s equations, and analyzing the comparison between two dyadic functions is considered. The object is assumed to be inside a region of interest (ROI). Due to the excellent geometric results in [6], the applied GPR system can localize and track a moving object underground by scanning the region of interest in t intervals. During each scan, the GPR system detects and calculates the exact location of the object using a Threshold Technique Method (TTM) based target detection system by analyzing the electromagnetic fields. The 3D FEKO Electromagnetic (EM) simulation software tool is used to model and simulate the GPR system. Simulation results achieve an excellent and accurate performance in the absence of external and unknown disturbances.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126622237","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Microscopic Blood Cell Classification Using Inception Recurrent Residual Convolutional Neural Networks","authors":"Md. Zahangir Alom, C. Yakopcic, T. Taha, V. Asari","doi":"10.1109/NAECON.2018.8556737","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556737","url":null,"abstract":"Deep Learning (DL) approaches have been explored in different modalities of biomedical image analysis, and they provide superior performance against alternative machine learning approaches. In this paper, we have evaluated the performance of a deep learning model called the Inception Recurrent Residual Convolutional Neural Network (IRRCNN) for White Blood Cell (WBC) and Red Blood Cell (RBC) classification. We have tested the performance of the IRRCNN approach on two publicly available blood cell datasets for both RBC and WBC classification obtained from the Yale School of medicine and CellaVision respectively. The experimental results show almost 100% recognition accuracy for the WBC dataset and 99.94% testing accuracy for RBC classification. This is approximately a 1.4% and 2.35% improvement when compared to existing deep learning-based approaches.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128520462","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Harmonic Distortion Estimate for Damage Detection","authors":"P. Burrascano, S. Laureti, M. Ricci","doi":"10.1109/NAECON.2018.8556701","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556701","url":null,"abstract":"The ultrasonic techniques of flaws detection and evaluation usually rely on linear models. High mechanical stress generates non-linear phenomena. Models suited to fit the input-output characteristics and the existence of non-linear behaviors, if any, allow improved sensitivity and early detection of the onset of cracks. We propose a technique of nonlinear characterization of the response, adequate for a real time industrial implementation.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128993287","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
M. McCaskey, Emily Kukura, Robert Corrigan, K. Bhasin, David T. Chelmins
{"title":"Machine Learning Applied to an RF Communication Channel","authors":"M. McCaskey, Emily Kukura, Robert Corrigan, K. Bhasin, David T. Chelmins","doi":"10.1109/NAECON.2018.8556708","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556708","url":null,"abstract":"In this paper we introduce a model of a node-to-node communication channel as an auto-encoder neural network where the inputs and outputs represent the transmitted and received bits respectively. With some simple assumptions, the encoding layer of the auto-encoder can be interpreted as a phase-modulated radio frequency (RF) signal to be transmitted across the communication channel. When the auto-encoder is trained to minimize bit errors between the transmitter and receiver nodes, it effectively learns the best modulation scheme for that particular communication channel. We show that an auto-encoder trained with Additive White Gaussian Noise (AWGN) has promising bit error rates compared to traditional modulation schemes in a simulated environment. Implementing the auto-encoder generated modulation scheme in GNURadio, we send auto-encoder modulated messages between two Universal Software Radio Peripherals (USRPs) and experimentally calculate the corresponding bit error rate. Preliminary results show bit error rates that are comparable to those of traditional modulation schemes measured in similar experimental conditions. Finally, we develop the methodology for incorporating the GNURadio flowcharts with the auto-encoder training program so that the auto-encoder can be trained on the actual communication channel itself and discuss how that can improve the auto-encoder results.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129690838","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Accelerating Inference In Long Short-Term Memory Neural Networks","authors":"Thomas C. Mealey, T. Taha","doi":"10.1109/NAECON.2018.8556674","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556674","url":null,"abstract":"Long Short-Term Memory (LSTM) is a powerful neural network algorithm that has been shown to provide state-of-the-art performance in various sequence learning tasks, including natural language processing, video classification, and speech recognition. Once an LSTM model has been trained on a dataset, the utility it provides comes from its ability to then infer information from completely new data. Due to the large complexity of LSTM models, the inference stage of LSTM can require significant computing power and memory resources in order to keep up with a real-time workload. Many approaches have been taken to accelerate inference, from offloading computations to GPUs or other specialized hardware, to reducing the number of computations and memory footprint required by compressing model parameters. This work takes a two-pronged approach to accelerating LSTM inference. First, a model compression scheme called binarization is identified to both reduce the storage size of model parameters and to simplify computations. This technique is applied to training LSTM for two separate sequence learning tasks, and it is shown to provide prediction performance comparable to the uncompressed model counterparts. Then, a digital processor architecture, called Binary Recurrent Unit (BRU), is proposed to accelerate inference for binarized LSTM models. Specifically targeted for FPGA implementation, this accelerator takes advantage of binary model weights and on-chip memory resources in order to parallelize LSTM inference computations. The BRU architecture is implemented and tested on a Xilinx Z7020 device clocked at 200 MHz. Inference computation time for BRU is evaluated against the performance of CPU and GPU inference implementations. BRU is shown to outperform CPU and GPU by as much as 39 times and 3.8 times, respectively.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"284 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121407841","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Improving Below Ground Radar Imagery of Targets Obscured by Metal Plates","authors":"Abdunaser Abdusamad, Muftah Akroush, M. Wicks","doi":"10.1109/NAECON.2018.8556654","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556654","url":null,"abstract":"Improving target detection and imagery in the presence of metal plates for any scanning device is a very important for two reasons: First, side plates are part of the scanning devices cannot be eliminated. Second, designing side plates using materials that absorb unwanted energy is very complicated and expensive. In this paper, we investigate the effect of metal side plates on a scanning radar operating at high frequency. Our goal is to improve the target image in the presence of these metal side plates. Mathematically, we can reduce the reflected waves from these surfaces and improve the resulting target image if we know the scattering caused by these plates (or at least a part of them).","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115803616","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
William Mitchell, Stefan Westberg, Anthony Reiling, T. Taha, E. Balster, Kerry L. Hill
{"title":"Generalized Power Modeling for Deep Learning","authors":"William Mitchell, Stefan Westberg, Anthony Reiling, T. Taha, E. Balster, Kerry L. Hill","doi":"10.1109/NAECON.2018.8556639","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556639","url":null,"abstract":"Convolutional neural networks (CNNs) have gained prominence with image classification and other related tasks, often outperforming humans. However, many CNNs require significant power to function. Power modeling tells how much power a particular CNN consumes on a particular hardware platform. This can be particularly helpful for deciding which CNN to deploy on a mobile platform where the power budgets is limited. Many existing power models are hardware specific and function at a fine grained level. That is, they describe the power consumption of basic kernel operations. Using these models to gauge the power consumption of a CNN, which uses more complicated kernel computations, is not straight forward. Also, comparing one CNN power consumption requires running multiple architecture specific models. Therefore it is beneficial to have one model that is hardware agnostic. This paper presents a statistics based approach to developing hardware agnostic general purpose power estimator for CNNs. In this paper, the model is validated for the Movidius Neural Compute Stick (MNCS) [1] to predict the power usage for a variety of CNNs. The model shows over 90% accuracy with predicting the power consumption of several popular CNNs.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"150 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134011027","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ayesha Zaman, E. Shin, C. Yakopcic, T. Taha, G. Subramanyam
{"title":"Experimental Study of Memristors for use in Neuromorphic Computing","authors":"Ayesha Zaman, E. Shin, C. Yakopcic, T. Taha, G. Subramanyam","doi":"10.1109/NAECON.2018.8556749","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556749","url":null,"abstract":"Memristor devices have the potential to drive a new class of specialized low power embedded hardware. The unique characteristics of these non-volatile and nanoscale devices allow them to perform parallel analog computing with extreme efficiency. To help facilitate the design of such systems, this paper describes the fabrication and characterization process used to develop memristors that are strong candidates for use in neuromorphic systems. In this work two different types of memristor devices, those with a GeTe switching layer, and those with a VO2 switching layer, are characterized and analyzed. These results are used to determine device suitability for use in neuromorphic computing applications through the properties of symmetry, reliability, stability, and programmability. In short, repeatable multi-level resistive switching has been investigated and the results have been summarized.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"220 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116166489","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A Real-Time Automatic Stability Optimization Loop (SOL) for MEMS-Referenced Oscillators","authors":"Mohammad S. Islam, S. Mandal","doi":"10.1109/NAECON.2018.8556793","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556793","url":null,"abstract":"The frequency stability of local oscillators (LOs) based on micro- and/or nano-mechanical resonators integrated with CMOS-based circuitry is a key performance limiter for ultra-low-power reconfigurable wireless architectures. This paper proposes a software-defined automatic stability optimization loop (SOL) that optimizes the stability of such LOs by finding and tracking their maximum stability points (MSPs). The technique enables automatic minimization of either frequency- or time domain oscillator stability metrics such as phase noise or Allan deviation (ADEV) in the presence of external perturbations. The SOL accomplishes this goal by using a perturb-and-observe (P & O) algorithm to lock the operating point of the oscillator at one of its MSPs. The proposed approach was experimentally verified on a MEMS-referenced oscillator realized using a single-crystal silicon (SC-Si) comb-drive resonator connected in feedback around a highly tunable single-chip CMOS sustaining amplifier.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121979208","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A. Brettin, Cobey L. McGinnis, Farzaneh Abolmaali, N. Limberopoulos, D. Walker, A. Urbas, L. Poffo, A. Maslov, V. Astratov
{"title":"Superresolution imaging of fluorescent nanospheres by using high-index microspheres embedded in slabs with illumination provided by plasmonic array","authors":"A. Brettin, Cobey L. McGinnis, Farzaneh Abolmaali, N. Limberopoulos, D. Walker, A. Urbas, L. Poffo, A. Maslov, V. Astratov","doi":"10.1109/NAECON.2018.8556690","DOIUrl":"https://doi.org/10.1109/NAECON.2018.8556690","url":null,"abstract":"It is shown that the resolution of virtual images of dye-doped dielectric nanospheres obtained through dielectric microspheres can be increased beyond the classical diffraction limit by decreasing the period of nanoplasmonic array used for localized plasmonic structured illumination of these objects. In contrast to far-field superlens design, the superresolution is obtained without using metallic layers and gratings after superlens. In addition, it is shown that the postimaging processing, which represents an intrinsic part of SIM, is not required for achieving the super-resolved images. This observation is interpreted based on a concept that the radiation of objects placed at the surface of nanoplasmonic arrays with sufficiently short periods is initially provided into folded dispersions of nanoplamonic array, so that the diffraction orders responsible for super-resolution are more efficiently coupled to dielectric microspherical antenna compared to that for the propagating modes within the cutoff frequency.","PeriodicalId":237271,"journal":{"name":"NAECON 2018 - IEEE National Aerospace and Electronics Conference","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127091848","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}