AI-assisted communication circuit optimization
AssigneeMediatek Inc.
InventorsPo-Yu Chen, Yen-Liang Chen, Chi-Tsan Chen, Chao-Wei Wang
Life of this patent
Estimating the term…
Abstract
A radio frequency (RF) communication assembly includes an RF communication circuit and a compensator apparatus. The compensator apparatus receives an input including an I-component of a pre-compensated signal, a Q-component of the pre-compensated signal, and encoded operating conditions of the RF communication circuit. The RF communication circuit includes RF circuit components causing signal impairments. The compensator apparatus perform neural network computing on the input, and the RF communication assembly generates a compensated output signal that compensates for at least a portion of the signal impairments.
Claims20
A method of training a compensator apparatus to compensate for signal impairments caused by a radio frequency (RF) transmitter (Tx) circuit, comprising:
- performing a first training stage to train an emulator neural network to emulate a plurality of Tx circuits in different operating conditions, wherein, for each of the Tx circuits, performing the first training stage further comprises:
- sending a first input to the Tx circuit and the emulator neural network to receive respective output signals;
- and
- updating weights and biases of the emulator neural network based on a first loss function, the first loss function calculated from a first difference between the respective output signals of the Tx circuit and the emulator neural network;
- and
- performing a second training stage subsequent to the first training stage to train a neural network in the compensator apparatus while keeping the weights and the biases of the emulator neural network fixed, wherein performing the second training stage further comprises:
- sending a given signal to the neural network, which causes the compensator apparatus to generate an input signal to the emulator neural network;
- and
- updating weights and biases of the neural network based on a second loss function, the second loss function calculated from a second difference between the given signal and a compensated output signal from the emulator neural network.
A method of training a compensator apparatus to compensate for signal impairments caused by a radio frequency (RF) receiver (Rx) circuit, comprising:
- performing a first training stage to train an emulator neural network to emulate a plurality of Rx circuits in different operating conditions, wherein, for each of the Rx circuits, performing the first training stage further comprises:
- sending a first input to the Rx circuit and the emulator neural network to receive respective output signals;
- and
- updating weights and biases of the emulator neural network based on a first loss function, the first loss function calculated from a first difference between the respective output signals of the Rx circuit and the emulator neural network;
- and
- performing a second training stage subsequent to the first training stage to train a neural network in the compensator apparatus while keeping the weights and the biases of the emulator neural network fixed, wherein performing the second training stage further comprises:
- sending a given signal to the emulator neural network, which generates an input signal to the neural network;
- and
- updating weights and biases of the neural network based on a second loss function, the second loss function calculated from a second difference between the given signal and a compensated output signal from the compensator apparatus.
Dependencies are inferred from the claim text.