Generative Neural Network Sensor Data Translation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Different sensors have unique transfer functions due to variations in hardware and software processing, making it challenging to process data from one sensor using a process designed for another, and existing methods fail to effectively handle signal interference in sensor data transmission.

Innovation Solution

A generative neural network is trained to translate data from a first sensor to simulated data from a second sensor, which is then processed by a specific-task machine-learning model, allowing for the use of a wider range of data sets and improving data compatibility across sensors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a processing process designed for a specific sensor is used, then the processing accuracy for that sensor is improved, but the adaptability to process data from different sensors deteriorates

Engineering Contradiction:

Improveprocessing accuracyVSAvoidsensor compatibility

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent introduces a generative neural network as an intermediary component that translates sensor data from one sensor type to another. This mediator enables the specific-task machine-learning model to process data from different sensors by converting them into the expected data format, thus resolving the contradiction between processing accuracy and sensor compatibility

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The generative neural network transforms the parameters and characteristics of sensor data by learning the transfer function mappings between different sensors. By changing the data representation parameters through translation, the system maintains processing accuracy while adapting to different sensor types

Inventive Principle:

Principle #35Parameter changes

2Device complexity

If existing sensor data processing methods are used, then the processing pipeline remains simple, but the ability to handle signal interference and translate between different sensor transfer functions deteriorates

Engineering Contradiction:

Improveprocessing pipeline complexityVSAvoidsignal handling capability

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The system performs preliminary translation of sensor data using the generative neural network before it enters the specific-task machine-learning model. This preliminary action prepares the data by removing sensor-specific characteristics and signal interference, improving reliability without significantly increasing overall system complexity

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces traditional mechanical signal processing approaches with a neural network-based translation system. This substitution enables the system to handle complex signal interference and transfer function variations more effectively while maintaining a relatively simple processing architecture

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11403510B2Processing sensor data

Publication Date: 2022.08.02 NOKIA TECHNOLOGIES OY

AI summary

An apparatus comprising means for: using a generative neural network, trained to translate first sensor data to simulated second sensor data, to translate input first sensor data from a first sensor to simulated second sensor data; and providing the simulated second sensor data to a different, specific-task, machine-learning model for processing at least second sensor data.