A Hodgkin-Huxley Interpretation of Power Amplifiers Memory Effects in Stochastic Computing

Paper in proceeding, 2025

This paper reports an interpretation of the memory effects of radiofrequency (RF) power amplifiers (PA) from the perspective of stochastic neural networks. The nonlinearity of RF PA is found biomimicking and can enhance neuromorphic computing capacities of a joint communication and computation (J2C) network. The memory effects are also sources of distortion caused by the PA. Our current study identifies a behavioral similarity between these memory effects and stochastic neurons in the neuromorphic computation when the input of the PA is a modulated signal. By solving differential equations in the stochastic Hodgkin-Huxley model with Bayesian estimation, this study demonstrates that the memory effects of RF PA are helpful when the required computation in the J2C involves a stochastic process. Overall, our work advances the integration of artificial intelligence (AI) and automation into communication networks, which is crucial to 6 G technology.