Time–Frequency Analysis of Phase Encoding in Spiking Neural Networks Using Polar Coordinates
Lei Zhang · 2024
This paper presents the time–frequency analysis of impulse trains and delayed impulse signal in polar coordinates for neural encoding and signal reconstruction using Spiking Neural Network (SNN). The frequency of the impulse train is encoded by phase in frequency domain polar coordinates. The time delay of impulse signal is encoded by phase in time domain polar coordinates. The SNN architecture is inspired by the Discrete Fourier Transform (DFT). The proposed SNN leverages complex exponential spiking neurons to represent the frequency components of the DFT spectrum. This approach encodes the temporal delay of an impulse through phase increments of the DFT frequency coefficients, allowing for the reconstruction of the original impulse signal using the SNN. The primary research objective is to simplify the SNN architecture and improve the computational efficiency of the SNN training process.