Year: 2025
Type: article
Abstract: The current work outlines an in-depth methodology for designing microcontroller-based signal processing systems with significant performance improvements through the use of integrated hardware-software co-optimized techniques. The proposed architecture employs an ARM Cortex-M4F processor running at 168 MHz with optimized peripherals, including a 16-bit SAR ADC and a 12-channel DMA controller, carefully designed to support real-time signal acquisition and processing while ensuring maximum CPU efficiency. The use of cascaded integrator-comb decimation filters in conjunction with adaptive applications of normalized least mean squares algorithms achieves a significant improvement in signal-to-noise ratio, at 48.3 dB, while ensuring computational efficiency through the adoption of block processing schemes. Realization of the four-layer printed circuit board incorporates electromagnetic interference suppression mechanisms and uses differential routing schemes, resulting in a measured noise floor of -96 dBV across the operating frequency range. Empirical testing confirms a 42% reduction in power consumption compared to conventional digital signal processing solutions, with processing latencies consistently restricted to under 50 microseconds for real-time application demands. The system achieves a computational efficiency of 3.8 GFLOPS/W while supporting multiple signal processing channels simultaneously, thus supporting its relevance for resource-restricted embedded environments in industrial control and the Internet of Things.The analysis validates that the adopted design achieves a balanced trade-off among performance metrics, energy consumption, and cost for next-generation embedded signal processing systems.
DOI: 10.54097/hddrb865
Author: Xinrun Song
Institution: Weifang University
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4412461500