Optimization of the digital comb filters implementation based on modern microcontrollers for real-time signal processing tasks

The article presents the results of both research and practical implementation of digital comb filters using STM32 series microcontrollers based on the ARM Cortex-M4 core. The topic is important because it addresses the need to improve computational efficiency and reduce hardware costs in real-time digital signal processing systems. Enhancing the methodology for implementing comb filters makes it possible to optimize decimation processes in modern telecommunication standards and improve the suppression of intermittent interference under challenging noise conditions. The aim of the article is to investigate a combined software–hardware approach to the implementation of recursive and non-recursive comb filters based on the STM32 microcontroller, in order to ensure their real-time operation with minimal memory usage, as well as to evaluate their frequency characteristics during the processing of signals with a specified sampling rate. Mathematical models of recursive and non-recursive comb filters are considered, as well as the features of their adaptation to architectures with limited computing resources. A method for optimizing the algorithm through the use of ring buffers, direct access to memory and the CMSIS-DSP library is proposed. A comparative analysis of the CPU workload for calculations with floating-point and fixed-point arithmetic is carried out. The effectiveness of the proposed solutions for filtering periodic interference under real time conditions has been experimentally confirmed. The practical significance of the results includes reduced CPU utilization, the development of software templates based on the CMSIS-DSP library that shorten the development and debugging time of embedded systems based on CortexM cores, and the ability to implement high-order filters on resource-constrained microcontrollers without loss of sampling rate.