A Limit-Aware Sparse Frequency-Domain Decision Engine for EMI Risk Feedback in Resource-Constrained Systems.

Area of Science:

• Electrical Engineering
• Signal Processing
• Embedded Systems

Background:

• Resource-constrained systems necessitate efficient electromagnetic interference (EMI) management.
• Fast Fourier Transform (FFT)-based full-spectrum processing is computationally intensive for EMI risk feedback.

Purpose of the Study:

• To propose a limit-aware sparse frequency-domain decision engine for internal EMI risk feedback.
• To reduce computational and hardware overhead in EMI management for resource-constrained environments.

Main Methods:

• Redefines EMI analysis from spectrum reconstruction to selective exceedance verification.
• Employs randomized spectral reordering, flat-window bucket aggregation, and folded sampling for spectral compression.
• Compares bucket-level amplitude envelopes with local limit envelopes to exclude risk-negative buckets.

Main Results:

• Achieves an average decision latency of 6.031 ms at 100 MHz.
• Reduces BRAM usage by up to 97.59% and dynamic power by up to 83.0% compared to FFT baselines.
• Demonstrates applicability through degradation experiments and in situ GaN power-converter measurements.

Conclusions:

• The proposed decision engine significantly reduces hardware overhead for frequency-domain EMI risk feedback.
• Enables efficient EMI management in resource-constrained systems.
• Offers substantial improvements in latency, power consumption, and hardware resource utilization.