Analog Comparator-Based Binary Classifier for Low-Power Sensor Data

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing machine-learning classifiers for analog sensor data consume high energy due to instrumentation amplifiers and digital multiply and accumulate operations, requiring a reduction in circuitry complexity and power consumption.

Innovation Solution

A binary classifier system using weighting amplifier stages and comparators to generate classification outputs directly from analog signals, with Error Adaptive Classifier Boosting (EACB) and Constrained Resolution Regression (CRR) to correct misclassifications and reduce energy consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Use of energy by moving object

If traditional classification architecture with instrumentation amplifiers, ADC and digital MAC operations is used, then classification functionality is achieved, but energy consumption increases significantly

Engineering Contradiction:

Improveenergy consumptionVSAvoidcircuitry complexity

Core Design Contradiction:

Use of energy by moving objectVSDevice complexity

Solution Approach 1:

The patent extracts and removes the energy-consuming components (instrumentation amplifiers, ADC, and digital MAC operations) from the traditional classification architecture. By taking out these components and replacing them with a comparator-based analog classification system, the invention achieves significant energy reduction while maintaining classification functionality.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent substitutes digital processing mechanisms (ADC and digital MAC operations) with an analog comparator-based system. This replacement eliminates the need for analog-to-digital conversion and extensive digital computation, directly reducing energy consumption associated with these operations.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Device complexity

If traditional classification architecture with ADC and digital MAC operations is used, then classification accuracy is maintained, but device complexity increases

Engineering Contradiction:

Improvecircuitry complexityVSAvoidclassification accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The patent creates an analog copy of the classification process using comparators that directly process analog sensor data. Instead of converting to digital and processing through complex algorithms, the system uses analog comparators to perform classification, simplifying the hardware while preserving the essential classification function.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent changes the operational parameters from digital domain (after ADC conversion) to analog domain (direct comparator processing). By operating entirely in the analog domain with comparators, the system reduces hardware complexity while maintaining classification accuracy through direct analog signal processing.

Inventive Principle:

Principle #35Parameter changes

3Power

If instrumentation amplifiers and digital MAC operations are used, then classification output is generated, but power consumption increases

Engineering Contradiction:

Improvepower consumptionVSAvoidclassification performance

Core Design Contradiction:

PowerVSProductivity

Solution Approach 1:

The patent inverts the traditional approach by not converting analog signals to digital first. Instead, it processes analog signals directly through comparators and generates digital output only when needed. This inversion of the signal processing flow eliminates the power-consuming ADC and digital MAC operations while maintaining classification performance.

Inventive Principle:

Principle #13The other way round (Inversion)

Data Source

PatentUS10853737B2Machine-learning classifier based on comparators for direct inference on analog sensor data

Publication Date: 2020.12.01 THE TRUSTEES OF PRINCETON UNIV

AI summary

A weak binary classifier configured to receive an input signal for classification and generate a classification output is disclosed. The weak binary classifier includes a plurality of weighting amplifier stages, each weighting amplifier stage being configured to receive the input signal for classification and a weighting input derived from a classifier model and generate a weighted input signal, the plurality of weighting amplifier stages being configured to generate a plurality of positive weighted input signals coupled to a positive summing node and a plurality of negative weighted input signals coupled to a negative summing node. The weak binary classifier also includes a comparator having a non-inverting input coupled to the positive summing node and an inverting input coupled to the negative summing node and being configured to generate a weak classification output based on the plurality of weighted input signals.