Signal Processing Method for Noise-Resistant Time-Frequency Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current signal processing methods, such as traditional time-frequency analysis and wavelet analysis, face limitations in achieving simultaneous good resolution in time and frequency, especially in noisy environments, and struggle to accurately detect transitions in non-stationary signals.
Innovation Solution
A method involving frequency selective analysis with undersampling operations, complex frequency translations, and Sliding Fourier Transforms to generate wideband analysis signals that provide instantaneous amplitude, phase, and frequency information, allowing for precise detection of signal modulations and noise rejection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional time-frequency analysis or wavelet analysis is used, then frequency resolution can be improved, but time resolution deteriorates and transitions between stationary periods cannot be detected accurately
Solution Approach 1:
The signal is divided into multiple stationary periods separated by transitions. For each stationary period, a specific frequency is identified, and transitions are detected by analyzing changes between consecutive periods. This segmentation allows simultaneous achievement of good frequency resolution (within each period) and time resolution (at transition points).
Solution Approach 2:
The analysis method adapts dynamically to signal characteristics by identifying stationary periods and transitions. The algorithm adjusts its analysis window and processing approach based on whether the signal is in a stationary state or undergoing transition, optimizing both time and frequency resolution for different signal conditions.
2Loss of information
If audio filter banks or sliding windowed Fourier Transforms are used for speech recognition, then frequency band information can be extracted, but transitions between stationary periods are not visible
Solution Approach 1:
The signal is segmented into stationary periods and transitions. Frequency band information is extracted for each stationary period, while transitions are identified separately by detecting changes in frequency content between consecutive periods. This ensures both frequency information and transition accuracy are preserved.
Solution Approach 2:
The analysis continuously processes the signal by maintaining a running identification of stationary periods and transitions. This continuous analysis ensures no frequency information is lost while simultaneously detecting all transitions, resolving the contradiction between information retention and transition detection reliability.
3Reliability
If a good noise rejection is achieved through accurate frequency analysis, then signal-to-noise ratio improves, but simultaneous good time and frequency resolution cannot be achieved
Solution Approach 1:
By segmenting the signal into stationary periods, the method can apply accurate frequency analysis within each period to reject noise, while transition detection between periods provides temporal precision. This segmentation enables both good noise rejection and simultaneous time-frequency resolution.
Solution Approach 2:
The analysis parameters (window size, frequency bins) are optimized for stationary periods to achieve good noise rejection, while transition detection uses different parameter settings focused on temporal changes. This parameter adaptation allows both noise rejection and time-frequency resolution to be achieved simultaneously for different signal segments.
Data Source
AI summary
A method for processing an initial signal includes a useful signal and added noise, which comprises a step of frequency selective analysis providing starting from initial signal a plurality of wideband analysis signals corresponding to one of the analyzed frequencies, and comprising the following actions: zero or more complex frequency translations, one or more undersampling operations, computation of the instantaneous Amplitude, of the instantaneous Phase, and of the instantaneous Frequency of the wideband analysis signals. This information then allow to detect modulations of signals included in high levels of noise and to detect with a good probability the presence of a signal in a high level of noise.