RFID Learning Device Using EM Simulation for Robust Tag Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
RFID systems face challenges in accurately identifying chipless-RFID tags or RFID sensors when the position, orientation, or environmental conditions of the identification target change, leading to erroneous identification due to changes in the reflected wave spectrum.
Innovation Solution
A learning apparatus and reader system that utilizes electromagnetic field analysis simulation to generate and train on varied reflected wave spectra, allowing for robust attribute identification through machine learning, even under changing conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the position or orientation of an identification target changes with respect to a reader, then the reflected wave spectrum changes, but identification accuracy deteriorates
Solution Approach 1:
The system performs preliminary actions by generating multiple reflected wave spectra through electromagnetic field analysis simulation under various position and orientation conditions before actual identification occurs. These pre-generated spectra serve as training data for machine learning, enabling the system to adapt to position and orientation changes without degrading identification accuracy
Solution Approach 2:
The system changes parameters by varying position and orientation parameters in electromagnetic field analysis simulation to generate diverse reflected wave spectra. This parameter variation enables the machine learning model to learn robust features that maintain identification accuracy across different positions and orientations
2Adaptability or versatility
If objects other than an identification target are present around the identification target, then the reflected wave spectrum changes, but identification accuracy deteriorates
Solution Approach 1:
The system performs preliminary actions by incorporating environmental variations including surrounding objects into the electromagnetic field analysis simulation before actual use. This pre-training with diverse environmental conditions enables the system to maintain identification accuracy when objects are present around the identification target
Solution Approach 2:
The machine learning model acts as an intermediary that processes reflected wave spectra and extracts features robust to environmental interference. This intermediary processing layer separates the identification task from environmental variations, maintaining accuracy despite surrounding objects
3Adaptability or versatility
If the ambient environment of an identification target changes, then the reflected wave spectrum changes, but identification accuracy deteriorates
Solution Approach 1:
The system performs preliminary actions by simulating various ambient environmental conditions in electromagnetic field analysis before deployment. This pre-training with diverse environmental conditions enables the system to adapt to temperature, humidity, and other environmental changes without losing identification accuracy
Solution Approach 2:
The system changes parameters by varying environmental condition parameters in simulation to generate reflected wave spectra under different ambient conditions. This enables the machine learning model to learn environmental invariance while maintaining identification accuracy
4Adaptability or versatility
If the reflected wave spectrum changes due to various factors, then the system becomes more adaptable to real-world conditions, but the complexity of identification processing increases
Solution Approach 1:
The system substitutes traditional mechanical or rule-based identification processing with machine learning-based processing. This substitution enables the system to handle complex reflected wave spectrum variations caused by real-world conditions while managing processing complexity through learned feature extraction rather than explicit rule management
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system achieves high accuracy and robustness in identifying RFID tags and sensors by constructing a learning model that can handle variations in position, orientation, and environmental changes, ensuring reliable identification performance.
Implementation Method 1
a plurality of resonators having resonance frequencies different from each other is formed on a base material of the tag, and identification information is expressed by a combination of the resonators
Implementation Method 2
a reflected wave spectrum of the identification target may change and identification itself of the attribute of the identification target may become difficult
Data Source
AI summary
This learning device is provided with: a simulation execution unit that, by using electromagnetic field analysis simulation, determines a reflected wave spectrum obtained when electromagnetic waves are emitted from a reader to an identification target; and a machine learning unit that, by using training data in which the reflected wave spectrum calculated by the simulation execution unit and an attribute thereof are defined as a set, performs a training process on a learning model by machine learning. The simulation execution unit generates a plurality of the reflected wave spectra belonging to the same attribute by variously changing various parameters related to the identification target from reference parameters. The machine learning unit performs a training process on the learning model by machine learning by using, as training data, the plurality of reflected wave spectra obtained for each attribute.