Broadband Interferer Detection Using Spectral Kurtosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for automatically detecting and separating broadband interferers in a time-frequency area are computationally intensive and inefficient, particularly when dealing with linear and non-linear chirp signals, as they require significant processing power and manual intervention, limiting their applicability in fully automated systems.
Innovation Solution
A method involving spectral kurtosis to identify significant frequency bins, followed by time domain analysis and a two-step fit process using predefined criteria and fit functions to determine and suppress broadband interferers, allowing for automatic detection and removal of chirp signals without manual interaction, thereby enabling efficient evaluation of the time-frequency area.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If spectral kurtosis is used to identify significant frequency bins, then detection accuracy of broadband interferers is improved, but computational complexity increases
Solution Approach 1:
The detection process is divided into multiple stages: first using spectral kurtosis to identify significant frequency bins, then applying time domain analysis only to those specific bins, and finally using fit functions only on the remaining significant points. This segmentation allows the system to achieve high detection accuracy while limiting computational complexity by applying complex operations only where necessary.
Solution Approach 2:
The patent applies different levels of analysis to different parts of the signal data. Spectral kurtosis is applied to all frequency bins to identify candidates, but the computationally intensive time domain analysis and fit function evaluation are applied only to significant frequency bins and their significant points, respectively. This local quality approach ensures high detection accuracy for interferers while reducing overall computational burden.
2Reliability
If Radon transform is used to identify chirp signals, then detection capability is improved, but processing time and computational load increase prohibitively
Solution Approach 1:
The patent performs preliminary filtering using spectral kurtosis to identify significant frequency bins before applying time domain analysis. This preliminary action reduces the data volume that requires computationally intensive processing, thereby reducing processing time while maintaining detection capability. The fit functions are then applied only to significant points within these significant bins, further optimizing the balance between detection reliability and processing speed.
3Measurement precision
If image processing algorithms like Sobel filtering are used, then chirp signal identification is improved, but computational complexity grows prohibitively when time-frequency slope is not estimated beforehand
Solution Approach 1:
The patent changes the approach from using fixed filter tap lengths (which require prior slope estimation) to using adaptive fit functions that can handle varying slopes. The fit functions are applied to significant points with dynamically determined parameters based on the local signal characteristics, allowing accurate chirp identification without prohibitively increasing computational complexity.
4Ease of operation
If cluster algorithms are used to combine significant points, then signal grouping is improved, but ability to separate chirp signals from other signals deteriorates
Solution Approach 1:
The patent extracts and isolates significant points from significant frequency bins using time domain analysis before applying fit functions. This extraction process separates potential chirp signal points from other signal components based on their temporal characteristics within the significant bins, improving the ability to distinguish chirp signals from other signals while maintaining ease of operation through systematic processing.
Data Source
Figure 1~2
Figure 3
AI summary
A method for detecting broadband interferer is described wherein a time-frequency area of incoming signals is determined by means of a spectrogram. A spectral kurtosis is used such that a spectral kurtosis value is calculated for each frequency bin of the incoming signals. A time domain analysis is performed in order to identify significant points of the significant frequency bins which have been investigated by spectral kurtosis previously. The significant points are evaluated in order to determine whether these significant points correspond to at least one broadband interferer.