摘要:
A two-stage stacked high IIP3 LNA with low current consumption is presented. Low-impedance bias terminations and optimum inter-stage match are used for IIP3 enhancement. A new graphical design technique is introduced for optimizing the linearity trade-offs in two-stage amplifiers and for optimizing the on-chip inter-stage matching network. Also, novel active circuits for bias modulation suppression are discussed. The LNA has been fabricated in a commercial SiGe BiCMOS technology, and measurement results are presented.
摘要:
Within a receiver, the frequency of a comparison reference clock signal supplied to a fractional-N Phase-Locked Loop (PLL) is dynamically changed such that undesirable reciprocal mixing of reference spurs with known jammers (for example, transmit leakage) is minimized. As the transmit channel changes within a band, and as the transmit leakage frequency changes, a circuit changes the frequency of the comparison reference clock signal such that reference spurs generated by the PLL are moved in frequency so that they do not reciprocally mix with transmitter leakage in undesirable ways. In a second aspect, the PLL is operable either as an integer-N PLL or a fractional-N PLL. In low total receive power situations, the PLL operates as an integer-N PLL to reduce receiver susceptibility to fractional-N spurs. In a third aspect, jammer detect information is used to determine the comparison reference clock signal frequency.
摘要:
Certain aspects of the present disclosure provide methods and apparatus for spiking neural computation of general linear systems. One example aspect is a neuron model that codes information in the relative timing between spikes. However, synaptic weights are unnecessary. In other words, a connection may either exist (significant synapse) or not (insignificant or non-existent synapse). Certain aspects of the present disclosure use binary-valued inputs and outputs and do not require post-synaptic filtering. However, certain aspects may involve modeling of connection delays (e.g., dendritic delays). A single neuron model may be used to compute any general linear transformation x=AX+BU to any arbitrary precision. This neuron model may also be capable of learning, such as learning input delays (e.g., corresponding to scaling values) to achieve a target output delay (or output value). Learning may also be used to determine a logical relation of causal inputs.
摘要:
Certain aspects of the present disclosure provide methods and apparatus for spiking neural computation of general linear systems. One example aspect is a neuron model that codes information in the relative timing between spikes. However, synaptic weights are unnecessary. In other words, a connection may either exist (significant synapse) or not (insignificant or non-existent synapse). Certain aspects of the present disclosure use binary-valued inputs and outputs and do not require post-synaptic filtering. However, certain aspects may involve modeling of connection delays (e.g., dendritic delays). A single neuron model may be used to compute any general linear transformation x=AX+BU to any arbitrary precision. This neuron model may also be capable of learning, such as learning input delays (e.g., corresponding to scaling values) to achieve a target output delay (or output value). Learning may also be used to determine a logical relation of causal inputs.
摘要:
Certain aspects of the present disclosure support a technique for utilizing a memory in probabilistic manner to store information about weights of synapses of a neural network.
摘要:
The present disclosure proposes implementation of a three-memristor synapse where an adjustment of synaptic strength is based on Spike-Timing-Dependent Plasticity (STDP) with dopamine signaling.
摘要:
A communication channel has a highly linear switched current mixer that incorporates passive filtering (e.g., low pass, notch) for improved transmitting (Tx) and receiving (Rx) with adding external filtering components. A high IIP2 (input referenced second order intercept point) of the receiver at the Tx offset is essential to avoid corrupting the system's sensitivity performance, and a high triple beat (TB) is required to avoid sensitivity degradation due to transmitter leakage. Thanks to the embedded filtering in the mixer and the active post-distortion (APD) method in a low noise amplifier (LNA), the required high linearity is achieved with low noise figure and power consumption, overcoming transmitter power leakage without the use of a SAW (surface acoustic wave) filter.
摘要:
Certain aspects of the present disclosure present a technique for primary visual cortex (V1) cell training and operation. The present disclosure proposes a model structure of V1 cells and retinal ganglion cells (RGCs), and an efficient method of training connectivity between these two layers of cells such that the proposed method leads to an autonomous formation of feature detectors within the V1 layer. The proposed approach enables a hardware-efficient and biological-plausible implementation of image recognition and motion detection systems.
摘要:
This disclosure describes techniques for reducing adverse effects of transmit signal leakage in a full-duplex, wireless communication system. The disclosure describes techniques for reducing adverse effects of second order distortion and cross-modulation distortion of transmit signal leakage from a transmitter via a duplexer. The techniques may be effective in rejecting at least a portion of a transmit leakage signal, thereby reducing or eliminating distortion. The adaptive filter may include an estimator circuit that generates a transmit leakage signal estimate. A summer subtracts the estimate from the received signal to cancel transmit leakage and produce an output signal. The estimator circuit generates the transmit leakage signal estimate based on a reference signal and feedback from the output signal. The reference signal approximates the carrier signal used to generate the transmit signal in the transmitter. The reference signal may be provided by the same oscillator used to produce the transmit carrier signal.
摘要:
Certain embodiments of the present disclosure support implementation of a digital neural processor with discrete-level synapses and probabilistic synapse weight training.