1: \begin{abstract}

2: For many practical applications in wireless communications, we need

3: to recover a structured sparse signal from a linear observation model

4: with dynamic grid parameters in the sensing matrix. Conventional expectation

5: maximization (EM)-based compressed sensing (CS) methods, such as turbo

6: compressed sensing (Turbo-CS) and turbo variational Bayesian inference

7: (Turbo-VBI), have double-loop iterations, where the inner loop (E-step)

8: obtains a Bayesian estimation of sparse signals and the outer loop

9: (M-step) obtains a point estimation of dynamic grid parameters. This

10: leads to a slow convergence rate. Furthermore, each iteration of the

11: E-step involves a complicated matrix inverse in general. To overcome

12: these drawbacks, we first propose a successive linear approximation

13: VBI (SLA-VBI) algorithm that can provide Bayesian estimation of both

14: sparse signals and dynamic grid parameters. Besides, we simplify the

15: matrix inverse operation based on the majorization-minimization (MM)

16: algorithmic framework. In addition, we extend our proposed algorithm

17: from an independent sparse prior to more complicated structured sparse

18: priors, which can exploit structured sparsity in specific applications

19: to further enhance the performance. Finally, we apply our proposed

20: algorithm to solve two practical application problems in wireless

21: communications and verify that the proposed algorithm can achieve

22: faster convergence, lower complexity, and better performance compared

23: to the state-of-the-art EM-based methods.

24: \end{abstract}