IMPROVED METHOD FOR GENERATING STOCHASTIC SIGNALS USING VOLTERRA SERIES

Abstract

The article presents an improved method for forming and nonlinearly processing stochastic signals in command-and-control information systems, based on second-order Volterra series with adaptive kernel parameter tuning. A mechanism for adjusting the kernel structure in accordance with the statistical properties of the input signal is proposed, enabling increased model stability under varying spectral content.

To evaluate the effectiveness, an experimental comparison was conducted between the improved method and the baseline Volterra-series approach, analyzing indicators such as effective spectral width, spectral-efficiency gain, impact on high-frequency components, and performance on structured (OFDM) and noise-like signals. The optimization reduced the effective spectral width by up to 8.6% for structured signals and ensured a consistent spectral-efficiency improvement of 1.9–5%, while preventing the uncontrolled spectral expansion characteristic of the basic model when processing noise realizations. The experimental results confirm the feasibility of integrating the adaptive kernel-tuning mechanism into the Volterra-based processing pipeline to improve its efficiency for real-time operation under limited frequency-resource conditions.