Accelerometer Peak Detection With Non-Uniform Sampling for Travel Modes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for automatically identifying transportation modes using smartphone accelerometers require lengthy data collection, straining embedded systems with limited storage capacity, as they rely on long time windows to differentiate between modes like subway and tram journeys.
Innovation Solution
A method that compresses data by non-uniformly sampling accelerometer measurements, detecting local extrema, and calculating characteristics such as mean amplitude, mean duration, and frequency, allowing for efficient recognition of travel modes without deteriorating performance, using techniques like principal component analysis and adaptive thresholding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If long time windows are used to differentiate between transportation modes, then measurement precision is improved, but storage capacity is exceeded
Solution Approach 1:
The patent extracts only the essential information from continuous acceleration data by detecting local extrema (peaks and valleys) and computing their characteristics (amplitude, duration, frequency). This extraction process converts large volumes of raw time-series data into a compact set of feature descriptors that retain the discriminative power needed for transportation mode classification while dramatically reducing storage requirements.
Solution Approach 2:
The patent transforms the raw acceleration signal into a different parameter space by computing characteristics such as peak amplitude, duration, and frequency. This parameter transformation allows the system to represent complex temporal patterns in a compressed form that is both storage-efficient and discriminative for mode recognition, resolving the contradiction between precision and storage volume.
2Reliability
If regular uniform sampling is used, then data coverage is improved, but data compression efficiency deteriorates
Solution Approach 1:
The patent transitions from static uniform sampling to dynamic non-uniform sampling where sample points are adaptively selected based on the local characteristics of the acceleration signal. By detecting local extrema and sampling at these dynamically determined points rather than at fixed intervals, the system maintains data representativeness while significantly reducing the number of samples required, thus improving compression efficiency.
Solution Approach 2:
The patent applies different sampling densities to different regions of the signal based on their local characteristics. Regions with significant features (local extrema) are sampled with higher density, while regions with minimal variation are sampled less frequently or skipped entirely. This local quality approach ensures that critical information is preserved while minimizing overall data volume.
Data Source
AI summary
A method for determining information regarding the displacement of an object from measurements provided by an accelerometer associated with the object. The method comprises the steps of detecting acceleration peaks in the measurements, of calculating one or more characteristics of the acceleration peaks detected, and of determining a mode of travel of the object from the characteristics of the acceleration peaks. The measurements undergo non-uniform resampling upon the detection of local extrema.