Adaptive Voltage Comparator for Fast Low-Power Anomaly Detection

The patent segments the anomaly detection task into two independent comparison operations: one for amplitude-based anomalies (comparing signal amplitude to threshold) and one for frequency-based anomalies (comparing frequency characteristics to expected values). This segmentation allows simple comparator circuits to replace complex ADC and FFT processing while maintaining detection accuracy for both anomaly types.

The patent extracts only the essential features needed for anomaly detection (amplitude and frequency characteristics) and processes them through dedicated comparator circuits. By taking out and processing only these critical features rather than performing full-spectrum FFT analysis, the system achieves anomaly detection with reduced computational complexity and lower power consumption.

The patent employs simple comparator circuits that are far less power-intensive than high-resolution ADCs. These comparators perform binary comparison operations that consume minimal power, enabling continuous monitoring without the high energy costs associated with precision ADC conversion and FFT processing. The system achieves adequate detection precision through this low-power approach.

The patent extracts and processes only the critical amplitude and frequency features through efficient comparator operations, avoiding the power-intensive full-spectrum analysis of FFT. By focusing computational resources only on the extracted essential features rather than processing entire signal spectra, the system maintains anomaly detection capability while dramatically reducing power consumption.

The patent segments the detection process into parallel amplitude comparison and frequency comparison operations. By dividing the anomaly detection task into these independent parallel pathways, the system eliminates the sequential processing bottleneck inherent in FFT-based methods, achieving faster detection without sacrificing the ability to detect both amplitude-based and frequency-based anomalies.

Instead of using FFT to extract frequency information from time-domain signals (the conventional approach), the patent inverts the approach by directly comparing frequency characteristics in the frequency domain and directly comparing amplitudes in the amplitude domain. This inverted methodology bypasses the computationally intensive FFT transformation step, significantly reducing detection time while maintaining comprehensive anomaly detection capability.