Mixed-Signal Simulation for Power-Accuracy Parameter Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manufacturers of commercial electronics devices lack a reliable methodology to optimize power usage in their products, as they are limited by the high-power consuming components, such as MEMS microphones and ADCs, which are not always necessary for the specific event or activity recognition tasks.
Innovation Solution
A computer system and methodology that simulates power usage and accuracy levels based on parameters associated with MEMS sensors, ADCs, and machine learning models, allowing for the identification of optimal power usage and accuracy settings for specific applications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-quality high-power-consuming components (MEMS microphone and audio ADC with 16 bits/16 khz resolution/sample rate) are used, then recognition accuracy is improved, but power usage increases
Solution Approach 1:
The system dynamically adjusts ADC parameters (resolution and sample rate) based on the specific recognition task requirements. For simple keyword spotting, lower parameters (8 bits/8 khz) are used to reduce power consumption, while for complex telephony tasks, higher parameters (16 bits/16 khz) are applied to maintain accuracy. This parameter adaptation resolves the contradiction between accuracy and power usage.
Solution Approach 2:
The patent implements a dynamic simulation system that evaluates different component configurations and their power-accuracy trade-offs in real-time. The system can adaptively select optimal component settings based on current operational requirements, transforming the static high-power configuration into a dynamic, task-adaptive system that maintains accuracy only when necessary.
2Reliability
If high-resolution high-sample rate ADC is used, then data quality for complex telephony recognition is improved, but power consumption increases
Solution Approach 1:
The system changes ADC operating parameters (resolution and sample rate) according to the specific recognition task. For keyword spotting applications, lower parameters reduce power consumption while maintaining sufficient reliability. For complex telephony recognition, higher parameters are applied to ensure reliability, thus resolving the contradiction through task-specific parameter adaptation.
Solution Approach 2:
The patent applies partial action by using only the necessary level of ADC quality required for each specific task. Instead of always using maximum quality (excessive action), the system applies just enough resolution and sample rate to achieve the required recognition reliability, thereby reducing power consumption while maintaining adequate performance.
3Measurement precision
If default high-power consuming components are used in all applications, then recognition accuracy is maintained across all tasks, but power efficiency deteriorates
Solution Approach 1:
The system implements parameter changes by adapting ADC resolution and sample rate to match specific recognition task requirements. This eliminates the waste of using high-power components for simple tasks like keyword spotting, thereby improving power efficiency while maintaining recognition accuracy only when the task complexity requires it.
Solution Approach 2:
The patent introduces dynamics into the previously static component configuration by implementing a simulation system that evaluates and selects optimal settings based on task requirements. This dynamic adaptation improves power efficiency by avoiding unnecessary high-power consumption in applications that do not require maximum recognition accuracy.
Data Source
AI summary
Disclosed embodiments provide a computer system and methodology for simulating operations of mixed analog and digital systems having various electronics components (for example, MEMS sensors, ADCs, and neural networks) for various sets of parameters, and selecting parameters for these components that would support an optimum pairing of power usage and accuracy in recognizing an activity or event.