Abstract: An electronic system for a firearm includes a power source, one or more electrical conductors electrically connected to receive power from the power source, and a plurality of electronic devices. Each electronic device has an electrical input configured to receive power from the one or more electrical conductors to power the electronic device. A communication device is configured for data communication across the one or more electrical conductors. In some embodiments data can be communicated at rates greater than 10 Mbps. Some embodiments utilize multi-frequency encoding, such as orthogonal frequency-division multiplexing. Some embodiments include a data communication device with a lower data communication rate and another data communication device with a higher data communication rate, and can selectively communicate through either data communication device.

Abstract: A method comprising determining a plurality of digital pre-distortion engines, determining signals for the pre-distortion engines, determining terms for a matrix and filter the matrix, based on the filtered matrix, determining correlation matrixes, obtaining pre-distorted signals from the digital pre-distortion engines, wherein the pre-distorted signals are pre-distorted based on the determined correlation matrixes, and combining the pre-distorted signals to a combined pre-distorted signal.

Abstract: An electronic device may include wireless circuitry with a baseband processor, a digital transmitter, a digital-to-analog-converter (DAC), and an antenna. The baseband processor may produce baseband signals. The digital transmitter may generate self-interference-compensated signals based on the baseband signals. The DAC may generate radio-frequency signals for transmission by the antenna based on the self-interference-compensated signals and square-wave local oscillator waveforms. The digital transmitter may include a self-interference canceller that generates the self-interference-compensated signals. The self-interference-compensated signals may mitigate the creation of self-interferer repetition replicas that land on the carrier frequency of the radio-frequency signals. This may allow the radio-frequency signals to be free from error vector magnitude degradation and spectral regrowth that would otherwise be produced due to self-interference in the radio-frequency signals output by the DAC.

Abstract: Various embodiments provide for deep learning-based architectures and design methodologies for an orthogonal frequency division multiplexing (OFDM) receiver under the constraint of one-bit complex quantization. Single bit quantization greatly reduces complexity and power consumption in the receivers, but makes accurate channel estimation and data detection difficult. This is particularly true for OFDM waveforms, which have high peak-to average (signal power) ratio in the time domain and fragile subcarrier orthogonality in the frequency domain. The severe distortion for one-bit quantization typically results in an error floor even at moderately low signal-to-noise-ratio (SNR) such as 5 dB. For channel estimation (using pilots), various embodiments use novel generative supervised deep neural networks (DNNs) that can be trained with a reasonable number of pilots. After channel estimation, a neural network-based receiver specifically, an autoencoder jointly learns a precoder and decoder for data symbol detection.

Abstract: Disclosed is an artificial intelligence or machine learning algorithm that may be applied to a plurality of machine learning devices in a 5G environment connected to perform the Internet of things. A machine learning method by a first learning machine according to one embodiment of the present disclosure may include obtaining input data; determining, from among a plurality of clusters, a cluster to which the input data belongs, by using a first artificial neural network; transmitting a plurality of sample features associated with the determined cluster to a second learning device using a second artificial neural network; receiving a label for the plurality of sample features from the second learning device, in response to the transmission; and associating the received label with the determined cluster.

Abstract: A method and an apparatus for adapting feature data in a convolutional neural network. The method includes selecting a plurality of consecutive layers; determining an expected number of subdata blocks and a layout position, width and height of each subdata block in an output feature data of a last layer; determining, for each current layer, a layout position, width, and height of each subdata block of an input feature data for the current layer according to the layout position, width, and height of each subdata block of the output feature data for the current layer; determining an actual position of each subdata block of the input feature data for a first layer in the input feature data for the first layer; and obtaining the expected number of subdata blocks of the input feature data for the first layer according to the actual position, width and height of each subdata block of the input feature data for the first layer.

Abstract: Systems, devices, and methods related to interpolation are provided. An example apparatus includes a slope calculator to calculate a slope value based on a first value and a second value associated with a function. The apparatus further includes a compander to compand the slope value to provide a companded slope value having a smaller bit-width than the calculated slope value. The apparatus further includes a multiplier to multiply the companded slope value by a third value to provide a correction value. The apparatus further includes an adder to add the correction value to the first value or the second value to provide an interpolated value associated with the function. Companding the slope value can reduce a bit-width of the multiplier, and thus may reduce power consumption and/or area.

Abstract: Embodiments of the present disclosure relate to systems and methods to generate a signal in a communication network. The method comprises filtering a discrete Fourier transform spread orthogonal frequency division multiplexing (DFT-s-OFDM) data signal, and one of a DFT-S-OFDM and orthogonal frequency division multiplexing (OFDM) reference signal (RS) using a data filter and a RS filter respectively, to produce filtered data signal and filtered RS. The RS filter has one to one relationship with the data filter. Thereafter, port mapping the filtered RS to a corresponding port assigned to the transmitter to obtain port mapped filtered RS, wherein the port mapped filtered RS comprises a first subset of non-zero locations comprising of the filtered RS values and a second subset of zero locations comprising of zero values.

Abstract: A method of multi-sensor data fusion includes determining a plurality of first data sets using a plurality of sensors, each of the first data sets being associated with a respective one of a plurality of sensor coordinate systems, and each of the sensor coordinate systems being defined in dependence of a respective one of a plurality of mounting positions for the sensors; transforming the first data sets into a plurality of second data sets using a transformation rule, each of the second data sets being associated with a unified coordinate system, the unified coordinate system being defined in dependence of at least one predetermined reference point; and determining at least one fused data set by fusing the second data sets.

Abstract: A light-triggered transponder includes one or more of photo cells, a clock recovery circuit and a reverse antenna system. The clock recovery circuit (CRC) includes a photoconductor with a source terminal, a drain terminal for receiving a voltage, the photoconductor resistance varying with received light intensity. The CRC is configured to generate a recovered clock. A reverse antenna system connected to at least one photo cell and configured to transmit data. The photoconductor configured to produce a modulated voltage signal from an incident modulated light incident. The CRC can include an amplifier coupled to the source terminal of the photoconductor via a capacitor for receiving the modulated voltage signal and outputting an analog signal generated from the voltage signal. The CRC can include an inverter coupled to the amplifier and configured to digitize the analog signal of the amplifier.

Abstract: Embodiments of this application provide a data compression method, including: performing first processing on ?/2-binary phase shift keying BPSK modulated data with a length of M, to obtain first frequency domain data with a length of M, where the first processing includes Fourier transform, and M is an even number; performing second processing on second frequency domain data with a length of Q, to obtain time domain data, where data in the second frequency domain data is included in the first frequency domain data, Q is a positive integer, M is greater than Q, Q is greater than or equal to M/2, and the second processing includes inverse Fourier transform; and sending the time domain data on one time domain symbol.

Abstract: A transceiver architecture supports high-speed communication over a signal lane that extends between a high-performance integrated circuit (IC) and one or more relatively low-performance ICs employing less sophisticated transmitters and receivers. The architecture compensates for performance asymmetry between ICs communicating over a bidirectional lane by instantiating relatively complex transmit and receive equalization circuitry on the higher-performance side of the lane. Both the transmit and receive equalization filter coefficients in the higher-performance IC may be adaptively updated based upon the signal response at the receiver of the higher-performance IC.

Abstract: A radio-frequency module includes a module substrate having a first major surface and a second major surface, a receive filter, a low-noise amplifier, an antenna switch, a first matching circuit disposed on the input side of the receive filter, a second matching circuit disposed on the output side of the receive filter, and a control circuit. The receive filter and the first and second matching circuits are arranged at the first major surface. The low-noise amplifier, the antenna switch, and the control circuit are arranged at the second major surface. When the module substrate is viewed in plan view, the receive filter is positioned between the first and second matching circuits, the control circuit is positioned between the antenna switch and the low-noise amplifier, and respective footprints of the second matching circuit and the low-noise amplifier coincide with each other.

Abstract: Systems, methods, and apparatuses, including electronic devices and computer-readable storage media, for adaptively switching wireless connections between antennas of a wearable electronic device and a host electronic device. One device includes a wearable electronic device with a first and second housing, each housing including two or more antennas. The wearable electronic device is configured to establish and monitor a wireless cross-link between two antennas in different housings, or between antennas in one housing and antennas of a host electronic device. The wearable electronic device can monitor the integrity of the wireless cross-link, and establish an updated cross-link in response to the wireless cross-link not meeting a predetermined integrity threshold. The wearable electronic device can monitor a wireless cross-head link between housings of a wearable electronic device at the same time as a wireless cross-body link between the wearable electronic device and the host electronic device.

Abstract: Presented are systems and methods for automatically creating and labeling training data for training-based radio, comprising receiving, at a receiver, a frame that comprises a modulated radio frequency (RF) signal comprising a set of waveforms that correspond to payload data. The payload data comprises a sequence of random bits. In embodiments, until a stopping condition is met one or more steps are performed, comprising detecting the frame; demodulating the modulated RF signal to reconstruct the sequence of random bits; using the reconstructed sequence to determine whether the payload data has been correctly received; in response to determining that the payload data has not been correctly received, discarding it and, otherwise, accepting the sequence of random bits as a training label; associating the training label with the modulated RF signal to generate labeled training data; and appending the labeled training data to a labeled training data set.

Abstract: An arithmetic processing apparatus includes: a first determiner that determines, when a given learning model is repeatedly learned, an offset amount for correcting a decimal point position of fixed-point number data used in the learning in accordance with a degree of progress of the learning; and a second determiner that determines, based on the offset amount, the decimal point position of the fixed-point number data to be used in the learning. This configuration avoids lowering of the accuracy of a learning result of a learning model.

Abstract: This patent concerns novel technology for detecting backdoors of neural network, particularly deep neural network (DNN), classifiers. The backdoors are planted by suitably poisoning the training dataset, i.e., a data-poisoning attack. Once added to input samples from a source class (or source classes), the backdoor pattern causes the decision of the neural network to change to a target class. The backdoors under consideration are small in norm so as to be imperceptible to a human, but this does not limit their location, support or manner of incorporation. There may not be components (edges, nodes) of the DNN which are dedicated to achieving the backdoor function. Moreover, the training dataset used to learn the classifier may not be available.

Abstract: Certain aspects of the present disclosure provide techniques for training a digital pre-distorter (DPD) using real-time over-the-air transmissions and receptions by a user equipment (UE). A method for training the DPD generally includes transmitting a signal, generated by a transmitter front end, via a first port; sampling the signal, received over the air, at a second port; performing signal processing cleaning (e.g., synchronization, linear over-the-air channel estimation and equalization); calculating coefficients for a DPD; and configuring the DPD with the coefficients, for use in digitally pre-distorting sub sequent transmissions.

Abstract: Presented are systems and methods for automatically decoding and demodulating radio signals in a communication network. Embodiments utilize a one-dimensional (1D) convolutional neural network (CNN) as the key architecture of a decoder that utilizes one or more 1D convolution windows to perform convolution operations on to-be-decoded or demodulated input signals received by the communication network. In embodiments, this may be achieved by receiving, at a CNN correlator implemented in a decoder, an input signal that comprises unknown data and applying to the input signal, in the discrete time domain, a convolution to obtain a CNN correlator output to which an activation may be applied to decode the input signal and output the decoded signal.

Abstract: Maximum expressivity can be received representing a ratio between maximum and minimum input weights to a neuron of a neural network implementing a weighted real-valued logic gate. Operator arity can be received associated with the neuron. Logical constraints associated with the weighted real-valued logic gate can be determined in terms of weights associated with inputs to the neuron, a threshold-of-truth, and a neuron threshold for activation. The threshold-of-truth can be determined as a parameter used in an activation function of the neuron, based on solving an activation optimization formulated based on the logical constraints, the activation optimization maximizing a product of expressivity representing a distribution width of input weights to the neuron and gradient quality for the neuron given the operator arity and the maximum expressivity. The neural network of logical neurons can be trained using the activation function at the neuron, the activation function using the determined threshold-of-truth.