Abstract:The conventional Cohen's distribution can't meet the requirement of additive noises jamming signals high-performance denoising under the condition of low signal-to-noise ratio, it is necessary to integrate the metaplectic transform for non-stationary signal fractional domain time-frequency analysis. In this paper, we blend time-frequency operators and coordinate operator fractionizations to formulate the definition of the metaplectic Wigner distribution, based on which we integrate the generalized metaplectic convolution to address the unified representation issue of the convolution type of metaplectic Cohen's distribution (CMCD), whose special cases and essential properties are also derived. We blend Wiener filter principle and fractional domain filter mechanism of the metaplectic transform to design the least-squares adaptive filter method in the metaplectic Wigner distribution domain, giving birth to the least-squares adaptive filter-based CMCD whose kernel function can be adjusted with the input signal automatically to achieve the minimum mean-square error (MSE) denoising in Wigner distribution domain. We discuss the optimal symplectic matrices selection strategy of the proposed adaptive CMCD through the minimum MSE minimization modeling and solving. Some examples are also carried out to demonstrate that the proposed filtering method outperforms some state-of-the-arts including Wiener filter and fixed kernel functions-based or adaptive Cohen's distribution in noise suppression.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is the high - performance denoising problem of additive noise - interfering signals in the time - frequency analysis of non - stationary signal fractional domains under low signal - to - noise ratio (SNR) conditions. Specifically, the traditional Cohen distribution cannot meet the need for efficient denoising of additive noise - interfering signals under low SNR conditions. Therefore, it is necessary to introduce the metaplectic transformation to deal with the unified representation problem in the time - frequency analysis of non - stationary signal fractional domains.
### Core problems and solutions of the paper
1. **Problem description**:
- The traditional Cohen distribution cannot effectively remove additive noise under low SNR conditions.
- A new method is required to improve the feature extraction and analysis capabilities of non - stationary signals in complex environments.
2. **Solutions**:
- **Introducing Metaplectic transformation**: By combining the blocking of time - frequency operators and coordinate operators with the metaplectic Wigner distribution, the convolution - type metaplectic Cohen distribution (CMCD) is defined, which solves the unified representation problem of the convolution - type metaplectic Cohen distribution.
- **Designing an adaptive filter**: Combining the Wiener filtering principle and the fractional - domain filtering mechanism, a least - squares adaptive filtering method is designed to achieve minimum mean - square - error (MSE) denoising in the Wigner distribution domain.
- **Optimally selecting symplectic matrices**: Through minimizing MSE modeling and solving, an optimal symplectic matrix selection strategy is proposed, thereby further improving the denoising performance.
### Specific steps and technical details
1. **Defining the Metaplectic Wigner distribution**:
- Combining the high - dimensional CICFWD and the symplectic Wigner distribution, the definition of the metaplectic Wigner distribution is proposed.
- Using the formula:
\[
W(M, M_1, M_2)f(x, u)=\mu_{y, 2}(M)T_P(\mu(M_1)f\otimes\mu(M_2)f)(x, u)
\]
2. **Constructing the convolution - type Metaplectic Cohen distribution**:
- Combining the metaplectic Wigner distribution with the generalized metaplectic convolution, the convolution - type metaplectic Cohen distribution (CMCD) is defined.
- Using the formula:
\[
(f\Theta_{M_4, M_5, M_6}g)(x)=e^{-\pi ixB_6^{-1}A_6x^T}\left(f(x)e^{\pi ixB_4^{-1}A_4x^T}*g(x)e^{\pi ixB_5^{-1}A_5x^T}\right)
\]
3. **Designing the least - squares adaptive filter**:
- Based on the metaplectic Wigner distribution domain, a least - squares adaptive filtering method is designed so that the kernel function can be automatically adjusted according to the input signal to achieve minimum mean - square - error (MSE) denoising.
- Using the formula:
\[
\mu(M_6)(f\Theta_{M_4, M_5, M_6}g)(u)=\epsilon_{M_4, M_5, M_6}(u)\mu(M_4)f(u(B_6^{-1})^T B_4^T)\mu(M_5)g(u(B_6^{-1})^T B_5^T)
\]
where,
\[
\epsilon_{M_4, M_5, M_6}(u)=\sqrt{-\frac{\det(B_4)\det(B_5)}{\det(B_6)}}e^{\pi i\left[uD_6B_6^{-1}u^T - u(B_6^{-1})^T B_4^T D_4B_6^{-1}u^T - u(B_