Wireless Train Spectrum Sensing Using CNN Communication Inference

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing wireless train control systems face challenges in achieving high-accuracy spectrum sensing for cognitive wireless communication due to interference from other systems using ISM bands, necessitating improved spectrum sensing techniques.

Innovation Solution

A machine learning model is generated using training data that includes radio wave environment and frequency-specific communication status information, utilizing a convolution neural network to improve accuracy in spectrum sensing by inferring the communication status of wireless base stations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If cognitive wireless communication is used to search for vacant channels, then the ability to avoid interference is improved, but the accuracy of spectrum sensing deteriorates due to difficulty in distinguishing actual vacancies from interference patterns

Engineering Contradiction:
Improveability to avoid interferenceVSAvoidaccuracy of spectrum sensing
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The system performs preliminary actions by collecting radio wave environment data and communication status information before making spectrum sensing decisions. Training data is gathered in advance under various radio wave environments, enabling the machine learning model to learn patterns and improve sensing accuracy before actual cognitive wireless communication operations begin.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system implements feedback mechanisms by using actual communication status information (such as successful communication outcomes) to refine and retrain the machine learning model. This feedback loop allows the system to continuously improve spectrum sensing accuracy based on real-world performance data, resolving the contradiction between adaptability and measurement precision.

Inventive Principle:
Principle #23Feedback

2Measurement precision

If machine learning models are trained with extensive radio wave environment data, then spectrum sensing accuracy is improved, but the time and computational resources required for model generation increase

Engineering Contradiction:
Improvespectrum sensing accuracyVSAvoidmodel generation time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary data collection and model training in advance, building a ready-to-use machine learning model before actual cognitive wireless communication operations. This preliminary action allows extensive training data to be processed beforehand, so that during actual operations, the model can make rapid spectrum sensing decisions without time-consuming computation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses partial training approaches by initially training with a representative subset of radio wave environment data to achieve sufficient accuracy, then using feedback from actual operations to incrementally improve the model. This avoids the need to process all possible training data upfront, reducing initial model generation time while still achieving high accuracy.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4068829B1Learning model generation method, inference device, and wireless train control system
Publication Date: 2026.02.25 KYOSAN ELECTRIC MFG CO LTD
  • EP4068829B1 patent drawingFigure 1
  • EP4068829B1 patent drawingFigure 2
  • EP4068829B1 patent drawingFigure 3

AI summary

A communication status inference apparatus (40) acquires a wireless signal as data indicating a radio wave environment of a wireless base station (10) installed along a railroad line when an onboard system (52) and the wireless base station (10) perform cognitive wireless communication, the wireless signal being received by a monitoring reception apparatus (20) in the vicinity of the wireless base station (10). The communication status inference apparatus (40) executes an inference process of inputting the acquired data indicating the radio wave environment to a communication status inference model and outputting frequency-specific communication status information based on a signal-to-noise ratio (SNR) at which the wireless base station (10) performed wireless communication under the radio wave environment. The communication status inference model is a machine learning model that has undergone learning using training data with the data indicating the radio wave environment of the wireless base station (10) as an input and with the frequency-specific communication status information based on the SNR at which the wireless base station (10) performed wireless communication under the radio wave environment as an output.