Smart geo-intelligence systems consisting of mobile information carriers, driven by an information field, often face geo-intelligence problems possessing two aspects which need to be addressed to institute actions. First is estimation of the location state of a mobile information carrier where the dynamical law for state change is unknown but where the information field responsible for state change is known. The second is the characterization of the anomalous statistical structure in captured signals at predicted locations. A two-tier statistical approach is adopted here for the purpose of demonstrating optimal location estimation and anomaly quantification addressing the dual problem of where a mobile information carrier is and what anomalous structure exists in an acquired signal.In the first tier the least absolute shrinkage and selection operator (LASSO) along with Gaussian process (GP) nonlinear regression is used to optimally estimate the location of a mobile information carrier, a buoy, drifting due to a multivariate driving information field. This field emanates from a surface current and wavefield. The use of an ensemble of estimators emanating from the multivariate data is envisioned as the optimal estimator of location where estimator coupling is the ultimate means for reducing error producing noise. The multivariate data used in modelling the relationship between latitude and longitude, and the information field and then making future predictions of location is buoy data emanating from a SOFAR spotter buoy moving in the mid-Atlantic Ocean around the mean latitude and longitude of 38º and 320º (-40º) respectively. Two location variables (latitude and longitude) and seven information variables were measured by the SOFAR spotter buoy. The information field variables consist of the mean wave directional spread, mean wave direction, mean wave period, wave peak directional spread, wave peak direction, wave peak period, and significant wave height. The two location variables and the seven measured information field variables were used in the LASSO-GP nonlinear regression location estimator consisting of a LASSO-based feature extraction stage, a GP training stage, and a GP prediction stage aimed at optimal forecasting of location based only on data and utilizing no dynamical constraints.LASSO-GP estimation is a nonparametric Bayesian approach where the measured location and multivariate information fields are modelled as stochastic processes where latitude and longitude are the predictands and the information field is the predictor. LASSO results for latitude and longitude allow for extraction of three dominant oceanic information field variables, which are different for each location variable, as the most important for location estimation. Nonlinear fits of latitude and longitude data with mean wave directional spread demonstrate rough fits over a training data set (priors) of 300 points. Extrapolation over 50 future data points for location using these wave information field variables yield erratic location...
Deep learning can identify different signals and extract a range of useful features or track a signal source. Semi and selfsupervised learning techniques can be used to teach networks the underlying dynamics of a problem and broaden generalizability. We demonstrate preliminary results on machine learning software capable of identifying the source of a target and extracting key pieces of information to help resolve or identify the source including angle of arrival. A Ushaped convolutional network may be trained to classify signals based on IQ samples according to modulations or other select features while reconstructing the clean signal. Use of semi-supervised learning training schedule including Barlow Twins on the generated latent space was demonstrated on combinations of real and synthetic radiofrequency (RF) signals. These signals were augmented under various common signal obfuscations such as Raleigh fading, reflections, varying noise and background signals. Group structure of the signals may be displayed through latent space visualizations. Classification accuracy on unseen test sets was used as the primary measurement of performance under varying levels of obfuscation. From this base, we attempted to combine this network with directional sensitivity in order to enable beam steering or identifying the source. A similar augmentation route enhanced by similar semi and selfsupervised techniques was deployed to improve tracking accuracy under realistic conditions. Statistical techniques may be used to identify frequency regions of interest during the prototyping of this signal identification network. This Deep network framework may be applied across a variety of domains and regimes for sensing and tracking.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.