Signal Classification via Eigenvalue Ratio Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Energy detection in cognitive radios is vulnerable to noise uncertainty, making it difficult to accurately classify received signals as data or noise due to uncertainties in noise power from non-linearity, thermal noise, and interference from other users.
Innovation Solution
A method that determines a covariance matrix and eigenvalue matrix of the received signal, using these to calculate different functions based on eigenvalues to classify the signal into data or noise, without requiring knowledge of noise power or channel information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If energy detection is used for signal classification, then the detection method is simple and does not require signal information, but the classification accuracy deteriorates due to noise uncertainty
Solution Approach 1:
The patent transforms the classification problem from direct energy comparison to eigenvalue ratio analysis. By changing the parameter from raw energy values to eigenvalue ratios, the method achieves robustness against noise uncertainty while maintaining operational simplicity. The eigenvalue ratio automatically normalizes the effect of noise power variations.
Solution Approach 2:
The patent introduces covariance matrix and its eigenvalues as intermediary elements between the received signal and the classification decision. These intermediaries transform the raw signal energy into a form that is invariant to noise power changes, thereby improving classification accuracy without complicating the overall detection process.
2Measurement precision
If noise power knowledge is required for accurate detection, then the detection accuracy improves, but the system becomes vulnerable to noise uncertainty from non-linearity, thermal noise, and interference
Solution Approach 1:
The patent makes the detection method self-sufficient by using the signal's own statistical properties (covariance matrix eigenvalues) for classification. The method does not require external noise power knowledge or calibration, as the eigenvalue ratio inherently compensates for noise effects. This self-service approach eliminates vulnerability to noise uncertainty.
Solution Approach 2:
Instead of using noise power knowledge to improve detection (the conventional approach), the patent inverts the approach by using signal statistical properties that are inherently independent of noise power. The eigenvalue ratio method works against the conventional wisdom by not requiring noise power information at all.
3Device complexity
If traditional energy detection is used, then the implementation is straightforward, but the false alarm rate increases under noise uncertainty
Solution Approach 1:
The patent changes the detection parameter from total energy to eigenvalue ratio. This parameter transformation fundamentally alters the detection statistic's behavior under noise, reducing false alarms while maintaining implementation feasibility. The eigenvalue ratio naturally adapts to noise conditions without requiring complex adaptive algorithms.
Data Source
AI summary
An embodiment of the invention provides a method for classifying a received signal. The method includes determining a covariance matrix of signal values of the received signal, and determining an eigenvalue matrix of the covariance matrix. The eigenvalue matrix includes the eigenvalues of the covariance matrix. A first function is determined from at least one eigenvalue of the eigenvalues of the covariance matrix. A second function is determined from at least one eigenvalue of the eigenvalues of the covariance matrix, wherein the second function is different from the first function. Dependent from a comparison between a value of the first function and a value of the second function, the received signal is classified into a signal comprising data or into a noise signal.