Abstract

This paper presents an innovative method for creating a flexible chronotropic frequency counter for processing endless data streams. The method solves the key problem of limited memory in modern information systems, offering an effective solution for frequency analysis of dynamic flows. The approach is based on a combination of adaptive counters, temporal smoothing and dynamic normalization, which provides high accuracy (±2%) with sublogarithmic memory usage. Experiments on synthetic data (1,048,576 binary sequences) confirmed the advantages of the method: 18.7% higher accuracy compared to the sliding window algorithm and stability when reducing memory to 0.01% of the original volume. The method demonstrates a linear dependence of processing time on the volume of data (R²=0.98) and rapid adaptation to changes in the flow (12.4±3.1 iterations). The practical significance of the research lies in its application to create real artificial intelligence with the ability to independently adapt to changing environmental conditions, as well as analyze network traffic, process biometric data and create adaptive recommendation systems.

Keywords: streaming data, chronotropic frequencies, flexible counters, adaptive algorithms, frequency analysis, dynamic normalization, real-time processing