Abstract:This work studies the problem of jointly estimating unknown parameters from Kronecker-structured multidimensional signals, which arises in applications like intelligent reflecting surface (IRS)-aided channel estimation. Exploiting the Kronecker structure, we decompose the estimation problem into smaller, independent subproblems across each dimension. Each subproblem is posed as a sparse recovery problem using basis expansion and solved using a novel off-grid sparse Bayesian learning (SBL)-based algorithm. Additionally, we derive probabilistic error bounds for the decomposition, quantify its denoising effect, and provide convergence analysis for off-grid SBL. Our simulations show that applying the algorithm to IRS-aided channel estimation improves accuracy and runtime compared to state-of-the-art methods through the low-complexity and denoising benefits of the decomposition step and the high-resolution estimation capabilities of off-grid SBL.
What problem does this paper attempt to address?
This paper attempts to solve the problem of jointly estimating unknown parameters from multi - dimensional signals with Kronecker structure, especially in application scenarios such as intelligent reflecting surface (IRS) - assisted channel estimation. Specifically, the author proposes a novel off - grid sparse Bayesian learning (SBL) algorithm. By decomposing the estimation problem into multiple independent sub - problems and using the Kronecker structure to reduce complexity and improve estimation accuracy.
### Main problems and solutions
1. **High - complexity problem**:
- **Deficiencies of existing methods**: Although the existing Kronecker - SBL (KroSBL) algorithm can be applied to such problems, it directly estimates the high - dimensional vector \( \mathbf{x} \), resulting in high computational complexity.
- **Solution**: This paper proposes a decomposition - based algorithm, which decomposes the original high - dimensional problem into multiple low - dimensional sub - problems, thereby significantly reducing the computational complexity. By making full use of the Kronecker structure, each sub - problem can be independently solved in a lower dimension.
2. **Grid - mismatch problem**:
- **Deficiencies of existing methods**: Traditional methods based on predefined grids may lead to grid - mismatch problems, that is, the true parameters are not on the predefined grid points, thus affecting the estimation performance.
- **Solution**: This paper introduces a discrete - grid SBL algorithm, which optimizes the grid position in each iteration and avoids the grid - mismatch problem. Specifically, the parameter estimation values and grid positions are dynamically adjusted through alternating minimization within the expectation - maximization (EM) framework.
### Method overview
1. **Decomposition algorithm**:
- Decompose the measurement vector \( \mathbf{\bar{y}} \) into multiple low - dimensional measurement vectors \( \hat{\mathbf{y}}_i \), with each sub - problem corresponding to one dimension.
- Use the higher - order singular value decomposition (HOSVD) or recursive SVD method to achieve low - complexity approximate decomposition.
2. **Discrete - grid SBL algorithm**:
- On each dimension, transform the parameter estimation problem into a sparse vector recovery problem.
- Update the parameter estimation values and grid positions through the EM algorithm to ensure the accuracy of the estimation results.
### Theoretical analysis and extension
- **Error - bound analysis**: Provide error - bound analysis of the decomposition step, quantifying the error between the decomposed signal and the true signal.
- **Denoising - effect analysis**: Prove that the decomposition step has a denoising effect, that is, the noise level of the decomposed signal is lower than that of the original signal.
- **Convergence - analysis**: Analyze the convergence of the discrete - grid SBL algorithm to ensure that the algorithm can converge stably.
### Applications and numerical results
- **IRS channel - estimation application**: Analyze in detail the signal model of the IRS - assisted wireless communication system and show the application of the proposed algorithm in the uplink - cascaded IRS channel estimation.
- **Numerical experiments**: Verify the effectiveness of the proposed algorithm through numerical experiments in three scenarios, including computational efficiency, high - resolution estimation ability, and improvement in IRS channel estimation.
In summary, this paper solves the high - complexity and grid - mismatch problems in Kronecker - structure signal parameter estimation by introducing decomposition techniques and the discrete - grid SBL algorithm, providing new ideas and methods for research in related fields.