Abstract: Systems and methods include a reduced computation, computer-based management of point-of-care (POC) on-site ultrasound (ULS) imaging resources. At a digital computation resource of a POC site, a ULS tissue reflection sample data is received from a hand-held ULS scanning device. A digital computation resource of the POC site, applies a reduced computation tissue ULS reflection speckle noise (SN) physics model—blurring noise (BN) physics model based deep learning (DL) trained convolutional neural network (CNN) denoising process to the ULS tissue reflection sample data outputting an estimated denoised ULS tissue reflection image data. A visual rendering of the estimated denoised ULS tissue reflection image data is displayed on a display resource of the POC site.

Abstract: Examples include receiving and storing samples of target frequency bands filtered from EEG measurement of a subject's brain waves in an NFB training session. In an example, upon storing a time window of the samples, unsupervised adaptive adjusting an NFB reward threshold is automatic. The adjusting includes, in examples, determining neuromarker values in the time window, which indicate peak values of the target frequency bands over the time window. The adjusting computes the mean value of the neuromarker values and, utilizing same, automatically proceeds to unsupervised computing an adaptive adjusted reward threshold. The unsupervised computing, in examples, includes a multiplication product of a reward threshold adjustment factor, a training protocol value, and the computed mean value of the neuromarker values. Examples proceed to communicating the adaptive adjusted reward threshold to a controller for threshold based feedback reward to the NBF subject.

Abstract: Methods, systems and wearable devices for real-time mental stress assessment are provided. The methods and systems employ deep learning using a Gated Recurrent Unit (GRU) gating mechanism in a recurrent neural network with a sliding window approach applied to raw EEG data.

Abstract: A device, method, and non-transitory computer readable medium to perform a method of blind equalization implemented by a filter bank multi-carrier with offset quadrature amplitude modulation (FBMC-OQAM) transmission system. A first matrix W is obtained by dividing a second matrix V by a receiver waveform matrix G. The second matrix V is obtained by calculating a total objective function J until the total objective function J is either constant or below a threshold error margin. The calculation of the total objective function J includes iterating calculations with a plurality of combination of a frequency bin p and a time slot q. Weights of the obtained first matrix W are applied to an equalizer. A received signal in the equalizer is processed using the applied weights. The weights of the obtained first matrix W are configured to minimize a total outage probability Pout, TOTAL.

Abstract: Memristor based chaotic oscillators exhibit complex dynamics. They are often chosen for secure communication owing to their interesting feature. In chaos-based secure communication applications using the master-slave configuration, synchronization is a central issue. Most of the synchronization methods proposed in the literature are asymptotic. In practice, it is desirable that synchronization be established in a predefined time. This invention provides new developments in the design of high-gain observers with an unknown input dedicated for predefined-time synchronization of memristor based chaotic systems. The proposed predefined-time extended high gain observer is constructed on the basis of a time-dependent coordinates transformation based on modulating functions that annihilate the effect of initial conditions on the synchronization time. Both noise-free channel and noisy channel are considered. Simulations performed on a numerical example illustrated the efficiency of proposed approaches.

Abstract: A filter bank multi-carrier (FBMC)-offset quadrature amplitude modulation (OQAM) system is disclosed. The FBMC-OQAM system includes a processing circuitry which is configured to receive a signal over a transmission medium, equalize the signal by a convolution neural network (CNN) equalizer, wherein the CNN equalizer is configured to estimate the received signal without performing channel estimation, and output the estimated signal as a bit stream.

Abstract: A method is provided of optimizing multi-cell association for a downlink multi-user, multiple-input, multiple-output (MU-MIMO) system that includes a plurality of base stations corresponding to C number of cells which serve K user terminals. Each user terminal connects with at least two cells. The method includes determining which cells for each user to connect with based on minimizing an objective function that is dependent on at least weight parameters for each kth user connected to each cth cell. Based on the determination, the user terminals are connected to their respective cells and the weight parameters are applied toward a beamforming configuration for the plurality of base stations. A device is also provided having processing circuitry configured to perform the method.

Abstract: Beamformers and beamforming methods are disclosed which involve diagonal loading (regularization). Features may include the automatic determination of the regularization parameter using a linearly constrained minimum power (LCMP) bounded perturbation regularization (BPR) approach.

Abstract: A speaker identification system and method to identify a speaker based on the speaker's voice is disclosed. In an exemplary embodiment, the speaker identification system comprises a Gaussian Mixture Model (GMM) for speaker accent and dialect identification for a given speech signal input by the speaker and an Artificial Neural Network (ANN) to identify the speaker based on the identified dialect, in which the output of the GMM is input to the ANN.

Abstract: A method and system that includes a wearable device having a plurality of EEG sensors to detect an EEG signal in a window of a predetermined length, a bandpass filter to remove frequency bands of the EEG signal to obtain a combined signal of remaining frequency bands, and a wireless device connection for wireless transmission of information from the wearable device. The information including the EEG signal and the combined signal. A mobile device includes a communication device for receiving the transmitted information, at least one processor for processing a machine learning model. The machine learning model classifies the combined signal to obtain a classification result of mental stress or not mental stress, and a display device displays a mental stress assessment based on the EEG signal and the classification result.

Abstract: A method and system for set point control of a liquid level in a conical tank is disclosed. The method includes receiving set point signal, R(s), summing the set point signal, R(s), with a negative feedback signal, d{tilde over (f)}(s), and generating set point error signal, E(s), which is used as feedback at the input to drive the liquid level in the conical tank to a desired set point, thus controlling the liquid level of conical tank so that external disturbance, df(s), is minimized.

Abstract: A method and system that includes a wearable device having a plurality of EEG sensors to detect an EEG signal in a window of a predetermined length, a bandpass filter to remove frequency bands of the EEG signal to obtain a combined signal of remaining frequency bands, and a wireless device connection for wireless transmission of information from the wearable device. The information including the EEG signal and the combined signal. A mobile device includes a communication device for receiving the transmitted information, at least one processor for processing a machine learning model. The machine learning model classifies the combined signal to obtain a classification result of mental stress or not mental stress, and a display device displays a mental stress assessment based on the EEG signal and the classification result.

Abstract: Beamformers and beamforming methods involve diagonal loading (regularization). Significant features include the automatic determination of the regularization parameter using a linearly constrained minimum power (LCMP) bounded perturbation regularization (BPR) approach. More specifically, the method has the advantage of efficient beamforming when the desired signal is from a moving source. After determining an initial regularization parameter, each data window (i.e., snapshot) uses the regularization parameter of the prior window to recursively determine the next regularization parameter.

Abstract: A device, method, and non-transitory computer readable medium for identification of stress resilience. The method for identification of stress resilience includes stimulating a human subject by at least one of a plurality of stressful events in a virtual reality environment, acquiring multichannel real-time electroencephalograph (EEG) signals by an EEG monitor worn by a human subject, recording the real-time EEG signals received during the stressful event, transmitting the real-time EEG signals to a computing device. The computing device generates a plurality of filtered brain wave frequencies related to the stressful event by filtering the multichannel real-time EEG signals, classifies the brain wave frequencies by frequency level by applying the filtered brain wave frequencies to the deep learning model, applies each frequency level associated with the stressful event to the convolutional neural network, and identifies a level of stress resilience of the human subject associated with the stressful event.

Abstract: A method, system, and apparatus for a linear active disturbance rejection control with a fractional order integral (FOI-LADRC) action for set-point tracking of a process variable is disclosed. The FOI-LADRC method includes receiving and multiplying a reference signal by a first feedback signal, applying, by a set-point tracking controller, fractional order integration to the reference signal, amplifying and multiplying the reference signal by a series of second feedback signals and a third feedback signal, dividing the reference signal by a static gain, and generating a process control variable, inputting the process control variable and applying a disturbance to the plant to output an output signal, feeding back the output signal, as the first feedback signal, generating the series of second feedback signals and a third feedback signal by an extended state observer, ESO, and tuning the set-point tracking controller and the ESO to eliminate the disturbance, from the output signal.

Abstract: A method and system for set point tracking control of an nth order integer control system. A first feedback compensated input signal is generated from an input signal, r(t), and a first negative feedback signal. A fractional order controller generates a set point modified signal from the first feedback compensated input signal. A second feedback compensated input signal is generated from the set point modified signal and a second negative feedback signal. A process variable is generated from the second feedback compensated input signal and applied to a non-linear plant which is affected by a disturbance which carries over into an output signal, y(t). An extended state observer receives the process variable and the output signal and generates the first second negative feedback and the second negative feedback signal to cancel the disturbance.

Abstract: A method and a system of distributed estimation using q-diffusion least mean squares (qDiff-LMS) to modify adaptive filter weights in a decentralized wireless sensor network of N nodes is described. The method includes receiving, at each node, k, a local estimate of a previous time instance weight, wk(i?1), of an adaptive filter of each neighboring node, l, where l=1, 2, . . . , M, combining the local estimates of the previous time instance weights to generate a linear combination of global diffused weights, Øk(i?1), measuring, for each node k, an output, yk(i), of the adaptive filter of the node k, calculating, for each node k, a desired response, dk(i); generating, for each node k, an estimation error, ekCTA(i) by subtracting the output, yk(i) from the desired response, dk(i), and updating the global diffused weights by adding a portion of the estimated error to the global diffused weights.

Abstract: A device, method, and non-transitory computer readable medium for interactive 3D visualization of ultrasound images of a supraspinatus tendon injury. Ultrasound images are acquired of a region in which the supraspinatus tendon injury is suspected. The ultrasound images are preprocessed, and energy of the preprocessed ultrasound images is minimized. A set of supraspinatus tendon images are extracted from low energy preprocessed ultrasound images. A morphological operation is performed on the set of supraspinatus tendon images to generate a smoothed set of supraspinatus tendon images. A binary mask is applied to the smoothed set of supraspinatus tendon images to detect boundary points of the supraspinatus tendon and generate a set of segmented image frames. The set of segmented image frames are arranged based on spatial position of each segmented image frame with respect to the supraspinatus tendon. A 3D representation of the supraspinatus tendon is reconstructed and rendered on a display.

Abstract: Methods and apparatuses are provided for estimating a path of an aerial vehicle engaged in attacking network devices in a wireless communication network. A distance function corresponding to the aerial vehicle and a boundary node is determined based on an initial coordinate location of the aerial vehicle and an initial coordinate location of the boundary node. A function of jamming power received at the boundary node from the aerial vehicle is determined based at least on the first distance function and a transmission power of the boundary node. The function of jamming power represents a power associated with a jamming signal received from the aerial vehicle at the boundary node. A trajectory of the aerial vehicle at a plurality of time periods is estimated by the boundary node with an extended Kalman filter. The extended Kalman filter is determined based on the function of jamming power.

Abstract: Devices, methods, and non-transitory computer readable media perform steps for securing communications in a Multi-User (MU), Massive, Multiple Input Multiple Output (MIMO) network. The base station estimates an initial precoding matrix based on a first estimated value of a signal to interference plus noise ratio (SINR) and determines a kth user device outage probability value and an eavesdropper device outage probability value associated with the eavesdropper device. An optimized beamformer weight matrix is determined based on the user device outage probability value and the eavesdropper device outage probability value. The base station determines whether the optimized beamformer weight matrix satisfies at least one outage condition. When the outage condition is satisfied, the base station transmits the optimized beamformer weight matrix to each remote radio head, which transmits downlink signals using a kth beamformer vector of the optimized beamformer weight matrix.