Motion Sensor Vector Data Analysis for Breathing Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing processes for analyzing multidimensional vector data from motion sensors, particularly for detecting breathing motion, require significant computing power and storage, and are inefficient in handling the influence of gravity, leading to high power consumption and potential inaccuracies.
Innovation Solution
A process that involves receiving and averaging multidimensional vector data over different time intervals to calculate medium-term and long-term vectors, determining mean-free vectors, and computing scalar products using unit vectors oriented in random directions, which allows for the extraction of rotation-invariant features and reduces computing power by using scalar values for motion identification and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If principal component analysis is used to analyze multidimensional vector data, then reliable feature extraction is achieved, but computing power and storage requirements increase significantly
Solution Approach 1:
The patent extracts only the essential gravitational component from the multidimensional vector data by using unit vectors oriented in random directions. Instead of performing full principal component analysis, the method extracts specific scalar projections that capture the necessary motion information while discarding redundant data, thereby reducing computing power and storage requirements while maintaining detection accuracy.
Solution Approach 2:
The patent transforms the multidimensional vector data into scalar values by computing dot products with randomly oriented unit vectors. This parameter transformation changes the data representation from high-dimensional vectors to simple scalars, significantly reducing the computational complexity and energy consumption while preserving the essential motion characteristics needed for breathing detection.
2Reliability
If full vector data is processed and stored, then complete motion information is preserved, but storage effort and computing complexity increase
Solution Approach 1:
The method extracts only the necessary motion information by projecting the vector data onto randomly oriented unit vectors. This extraction process isolates the relevant gravitational and motion components while eliminating redundant information, thereby reducing processing complexity and storage requirements while maintaining reliable motion detection capability.
Solution Approach 2:
Instead of analyzing the full multidimensional vector data directly, the patent inverts the approach by using randomly oriented unit vectors to project the data into scalar values. This inverted methodology simplifies the processing complexity while preserving the essential motion characteristics needed for accurate breathing detection.
3Speed
If multidimensional vector data is analyzed in real-time, then immediate motion detection is achieved, but power consumption increases
Solution Approach 1:
The patent changes the data parameters from high-dimensional vectors requiring complex matrix operations to simple scalar values that require only basic arithmetic operations. This parameter transformation enables real-time processing with minimal power consumption, as the simplified calculations can be performed quickly on low-power microcontrollers without sacrificing detection speed or accuracy.
Data Source
AI summary
A process for analyzing multidimensional vector data of a motion sensor for detecting a breathing motion, includes receiving and storing the multidimensional vector data of the motion sensor in a time series, calculating a plurality of medium-term vectors and of a plurality of long-term average vectors, calculating and storing a plurality of mean-free vectors depending on a difference between a respective medium-term vector and a respective long-term average vector and determining a plurality of unit vectors. The respective unit vector is oriented in a random direction. A plurality of scalar products are calculated from a respective mean-free vector and the unit vector assigned to the mean-free vector. A motion identification is calculated, which is an indicator of the breathing motion, based on the plurality of scalar products. An analysis signal is determined and output based on a comparison between the motion identification and a predefined motion threshold value.