Noise Reduction in Loose-Fitting Body Sensor Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Constrained body-sensor networks (cBSNs) cause discomfort and restrict motion, leading to reduced accuracy in human gesture and activity recognition (HAR), while loose-fitting body-sensor networks (fmBSNs) introduce motion artifacts that conventional techniques struggle to handle effectively.
Innovation Solution
A facility that processes fmBSN output using an artifact-reduction neural network, such as a DSTSAE or DCLSTM, to suppress noise and improve accuracy, allowing for more comfortable and effective HAR.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensors are tightly bound to the human body using straps or snugly-fitting clothing, then measurement precision is improved, but ease of operation deteriorates due to discomfort and restricted motion
Solution Approach 1:
The patent introduces an intermediary processing layer (noise reduction facility with autoencoder neural network) between the loose-fitting sensor network and the gesture recognition system. This intermediary suppresses motion artifacts in the sensor data, allowing loose-fitting sensors to achieve measurement precision previously only attainable with constrained sensors, while maintaining user comfort.
2Ease of operation
If loose-fitting clothing is used to bind sensors, then ease of operation is improved, but measurement precision deteriorates due to motion artifacts
Solution Approach 1:
The patent converts the harmful motion artifacts generated by loose-fitting sensors into beneficial training data for the autoencoder neural network. The network learns to distinguish between motion artifacts and genuine gesture signals during training, then automatically suppresses artifacts during operation, transforming the previously problematic loose-fitting approach into an accurate measurement system.
Solution Approach 2:
The patent replaces the mechanical constraint system (straps and snugly-fitting clothing) with a computational constraint system (neural network-based noise reduction). Instead of mechanically preventing sensor movement, the system computationally removes the effects of sensor movement from the data, achieving measurement stability without physical restraint.
3Measurement precision
If constrained body-sensor networks are used, then measurement precision is improved, but device complexity increases due to straps and securement mechanisms
Solution Approach 1:
The patent replaces complex mechanical securement systems (straps, clips, snugly-fitting garments) with a simpler computational processing system. The sensor network itself becomes mechanically simple and easy to attach, while the complexity is shifted to the software-based noise reduction facility that processes the sensor data.
Data Source
AI summary
A facility for processing output from a network of mechanical sensors is described. The facility accesses time-series data outputted by the network of sensors. The facility applies to the accessed time-series data a trained autoencoder to obtain a version of the accessed time-series data in which noise present in the accessed time-series data is at least partially suppressed. The facility stores the obtained version of the accessed time-series data, such as in order to perform human activity recognition against the obtained version of the accessed time-series data.