Analog Neural Network Folded Circuit for Ultra-Low Power Wearable Sensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wearable sensors face challenges in continuous monitoring due to limited battery life, which hinders the analysis of physiological data and the early detection of diseases, as existing technologies consume high power for processing and communication.
Innovation Solution
An ultra-low power analog neural network architecture with a 'folded' circuit design that enables neural network processing at the sensor node with nano- or pico-watt consumption, allowing for continuous monitoring without batteries by harnessing energy from environmental sources like vibrations or thermal heat.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If digital devices are used to analyze physiological data from wearable sensors, then data analysis capability is improved, but energy consumption increases
Solution Approach 1:
The system segments the data processing function by separating analog preprocessing at the sensor node from final digital analysis at remote devices. The analog neural network handles initial signal processing locally, while digital devices perform higher-level analysis, dividing the computational workload to reduce energy consumption at the wearable node.
Solution Approach 2:
An analog neural network circuit serves as an intermediary between the sensor and digital processing devices. This intermediary performs preliminary data analysis and filtering in the analog domain, reducing the amount of data that needs to be transmitted and processed digitally, thereby lowering overall energy consumption.
2Duration of action of moving object
If wearable sensors perform data analysis locally, then continuous monitoring capability is improved, but device complexity increases
Solution Approach 1:
The patent replaces complex digital processing systems with simpler analog neural network circuits that can perform data analysis continuously without requiring powerful batteries or complex digital processors. The analog circuitry provides continuous monitoring capability while maintaining relatively simple device architecture.
3Duration of action of moving object
If battery size is increased to support continuous monitoring, then monitoring duration is improved, but wearable device size increases
Solution Approach 1:
The system uses periodic action through energy harvesting from environmental sources such as vibrations or thermal heat to power the monitoring device continuously without requiring large batteries. The analog neural network circuit consumes minimal power, allowing it to be sustained by periodic energy input from the environment.
Data Source
AI summary
An analog neural network circuit includes at least one fewer layers than a number of expected layers of a neural network such that at least two cycles of feeding back outputs and applying weights occur to complete all the expected layers of the neural network. A control circuit, for example implemented using an analog oscillator, provides timing signals to control signal paths, including a feedback signal path to reuse circuitry of a layer for the at least two cycles. An analog memory is coupled to store an output of the circuitry of the layer. The analog memory is controllably coupled as part of the feedback signal path to the circuitry of the layer.