Abstract: The invention relates to a device for linearising a power amplifier by employing digital predistortion, comprising: a digital predistortion module, configured to infer a polar domain predistortion to be applied to a signal, and comprising a first neural network and a second neural network respectively configured to correct amplitude and phase distortion produced by the amplifier; an optimisation module of each of said neural networks configured to implement meta-learning, using: a meta-initialisation providing a prior initialisation of the initial weights of each of said neural networks; a meta-matching of the initial weights into optimal weights of each of said neural networks.

Abstract: A system filters out-of-band radio frequency emissions generated by wireless transmission of a first signal by a first transmitter from a wireless transmission of a second transmitter. The first signal is communicated from the first transmitter to the second transmitter via one or more wired connections. An out-of-band portion of a modulated format of the first signal is inverted to generate an inverted out-of-band component signal. The inverted out-of-band component signal is combined with a second signal of the second transmitter to create a filtering second signal. The filtering second signal is wirelessly transmitted from the second transmitter concurrently with wireless transmission of the first signal by the first transmitter, wherein the wireless transmission of the inverted out-of-band component signal in the filtering second signal by the second transmitter is synchronized with the wireless transmission of the first signal by the first transmitter.

Abstract: Aspects of this disclosure relate to user equipment assisted multiple-input multiple-output (MIMO) downlink configuration. Features are described for a user equipment determination of a desired transmission mode and/or active set of serving nodes for wireless communication service(s). The user equipment may submit a request for the desired mode and/or nodes to a network controller such as a baseband unit. The user equipment may subsequently receive a configuration for the requested wireless communication service(s).

Abstract: This patent concerns novel technology for detecting backdoors in neural network, particularly deep neural network (DNN) classification or prediction/regression models. The backdoors are planted by suitably poisoning the training dataset, i.e., a data-poisoning attack. Once added to an input sample from a source class of the attack, the backdoor pattern causes the decision of the neural network to change to the attacker's target class in the case of classification, or causes the output of the network to significantly change in the case of prediction or regression. The backdoors under consideration are small in norm so as to be imperceptible to a human or otherwise innocuous/evasive, but this does not limit their location, support or manner of incorporation. There may not be components (edges, nodes) of the DNN which are specifically dedicated to achieving the backdoor function. Moreover, the training dataset used to learn the classifier or predictor/regressor may not be available.

Abstract: Systems and methods for mitigating an effect interference. The methods comprise: receiving, by a device, a signal comprising a plurality of signal components; determining whether each signal component has a sufficient reconstructability; reconstructing each said signal component that was determined to have sufficient reconstructability using the received signal or an at least partially clean signal with other signal component(s) removed from the received signal; and using the reconstructed signal components to generate a modified received comprising the received signal with the signal components removed therefrom that (i) are devoid of a signal of interest and (ii) have sufficient reconstructability.

Abstract: A method that may include training a student ODNN to mimic a teacher ODNN. The training may include calculating a teacher student detection loss that is based on a pre-bounding-box output of the teacher ODNN. The pre-bounding-box output of the teacher ODNN is a function of pre-bounding-box outputs of different ODNNs that belong to the teacher ODNN. The method may also include detecting one or more objects in an image, by feeding the image to the trained student ODNN; outputting by the trained student ODNN a student pre-bounding-box output; and calculating one or more bounding boxes based on the student pre-bounding-box output.

Abstract: An embedded GPU-based wideband parallel channelized receiving method includes: constructing an OpenCL platform; decimating a wideband signal read in the OpenCL platform at an interval indicated by the number of channels; assigning data in each row to one of multiple work groups for processing; filtering data on each of channels based on a coefficient of a polyphase filter on a branch; multiplying the filtered data by a factor; and performing an FFT on the formed two-dimensional matrix by columns to obtain data outputted from each of the channels.

Abstract: A multi-PAM equalizer receives an input signal distorted by inter-symbol interference (ISI) and expressing a series of symbols each representing one of four pulse amplitudes to convey two binary bits of data per symbol. High-order circuitry resolves the most-significant bit (MSB) of each two-bit symbol, whereas low-order circuitry 115 resolves the immediate least-significant bit (LSB). The MSB is used without the LSB for timing recovery and to calculate tap values for both MSB and LSB evaluation.

Abstract: A method, system, and apparatus for applying dithering to waveforms in a transmitter such as a Bluetooth transmitter. A current waveform corresponding to a current bit of a bitstream is received where the current waveform has a nominal frequency deviation defined by a value of the current bit. Based on the determination that the current waveform and an immediately previous bit of the bitstream are associated with different bit values, a first dithered signal is output which is defined by a first frequency offset pseudorandomly selected from a first set of frequency offsets. A subsequent waveform to the current waveform is received corresponding to a subsequent bit of the bitstream. Based on the subsequent bit and the current bit being associated with bits of the same value, a second dithered signal is output which is defined by a second frequency offset pseudorandomly selected from a second set of frequency offsets.

Abstract: Power management for remote units in a wireless distribution system. Power can be managed for a remote unit configured to power modules and devices that may require more power to operate than power available to the remote unit. For example, the remote unit may be configured to include power-consuming remote unit modules to provide communication services. As another example, the remote unit may be configured to provide power through powered ports in the remote unit to power-consuming devices. Depending on the configuration of the remote unit, the power-consuming remote unit modules and/or power-consuming devices may demand more power than is available at the remote unit. In this instance, the power available at the remote unit can be distributed to the power-consuming modules and devices based on the priority of services desired to be provided by the remote unit.

Abstract: Systems, devices, and methods related to performing digital predistortion in radio frequency (RF) systems are provided. A digital predistortion (DPD) arrangement includes a DPD actuator circuit to predistort, using DPD coefficients, at least a portion of an input signal, the DPD coefficients associated with a characteristic of a nonlinear component. The DPD arrangement further includes a DPD capture circuit to perform, based on a capture cycle timing, multiple captures of a feedback signal, the feedback signal indicative of an output of the nonlinear component; compute, based on one or more characteristics of the multiple captures, one or more criteria for a subsequent capture of the feedback signal; and perform, based on the one or more criteria, the subsequent capture of the feedback signal. The DPD arrangement circuit further includes a DPD adaptation circuit to update the DPD coefficients based at least in part on the subsequent capture.

Abstract: Radio frequency front-end systems with filter reuse. In certain embodiments, a front-end system includes a filter, a low noise amplifier (LNA), and a switch interposed between the filter and an input to the LNA. In a first state of the switch, the filter serves to filter a radio frequency signal that is amplified by the LNA. The front-end system further includes a power amplifier that is coupled to the switch. Additionally, in a second state of the switch, the filter serves to filter an amplified radio frequency transmit signal provided by the power amplifier. Accordingly, a filter along a receive path of the front-end system is reused for transmit.

Abstract: A method for automatically detecting a packet mode in a wireless communication system supporting a multiple transmission mode includes: acquiring at least one of data rate information, packet length information and channel bandwidth information from a transmitted frame; and determining the packet mode on the basis of the phase rotation check result of a symbol transmitted after a signal field signal and at least one of the data rate information, the packet length information and the channel bandwidth information acquired from the transmitted frame.

Abstract: Embodiments are disclosed for facilitating an end-to-end link channel with one or more lookup tables for equalization. An example system includes a first transceiver and a second transceiver. The first transceiver includes a clock data recovery (CDR) circuit configured to receive communication data from a switch and to manage a lookup table associated with equalization of the communication data. The first transceiver also includes a first driver circuit communicatively coupled to the CDR circuit and configured to generate an electrical signal associated with the communication data. The second transceiver includes a second driver circuit, communicatively coupled to the first transceiver, that is configured to receive the electrical signal from the first transceiver and to modulate a laser source based on the electrical signal to generate an optical signal via the laser source.

Abstract: Systems, methods, apparatuses, and computer program products for array antenna adaptive digital pre-distortion with Bayesian observation analysis are provided. One method may include selecting a plurality of patch elements from an array antenna of a network element. The method may also include determining an accuracy confidence value for each patch element. A set of coefficients of the antenna array may be generated. In addition, an ensemble of non-linear forward models may be generated using the accuracy confidence value and the set of coefficients. Further, an array of pre-distortion signals may be generated using the ensemble of non-linear forward models, and each antenna of the array antenna may be configured with a corresponding pre-distortion signal from the array of pre-distortion signals.

Abstract: Methods, systems, and devices for wireless communications are described. A network entity may map a demodulation reference signal (DMRS), a truncated sequence, and data in a delay-Doppler domain in accordance with a control signal. The network entity may apply a Fourier transform on the mapped DMRS, the truncated sequence, and the data to generate a signal in the time domain. The network entity may output, and a user equipment (UE) may receive, the signal in the time domain, including the DMRS, the truncated sequence, and the data. The UE may apply a Fourier transform on the received signal in the time domain to generate a mapping of the DMRS and the data in the delay-Doppler domain. The UE may perform channel estimation based on applying the Fourier transform on the received signal.

Abstract: Methods, systems, and devices for wireless communication are described. A communication device, such as a user equipment (UE) may receive a set of demodulation reference signal (DMRS) samples including a first subset of DMRS samples associated with a first power level and a second subset of DMRS samples associated with a second power level. The UE may perform a digital post distortion operation based on the first subset of DMRS samples associated with the first power level and the second subset of DMRS samples associated with the second power level. The UE may receive the wireless communication based on performing the digital post distortion operation.

Abstract: A transmitting apparatus is provided.

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: A wireless device may receive one or more messages comprising configuration parameters for a primary cell and a secondary cell. The configuration parameters may indicate a deactivation timer for the secondary cell and a bandwidth part (BWP) inactivity timer for the primary cell. The wireless device may start the deactivation timer in response to activating the secondary cell. The wireless device may start the BWP inactivity timer in response to switching an active BWP of the primary cell. The wireless device may stop the BWP inactivity timer of the primary cell based on initiating a random-access procedure for the secondary cell. In response to the deactivation timer of the secondary cell expiring during the random-access procedure, the wireless device may restart the BWP inactivity timer of the primary cell.