Abstract:We consider the problem of recovering two unknown vectors, $\boldsymbol{w}$ and $\boldsymbol{x}$, of length $L$ from their circular convolution. We make the structural assumption that the two vectors are members of known subspaces, one with dimension $N$ and the other with dimension $K$. Although the observed convolution is nonlinear in both $\boldsymbol{w}$ and $\boldsymbol{x}$, it is linear in the rank-1 matrix formed by their outer product $\boldsymbol{w}\boldsymbol{x}^*$. This observation allows us to recast the deconvolution problem as low-rank matrix recovery problem from linear measurements, whose natural convex relaxation is a nuclear norm minimization program. We prove the effectiveness of this relaxation by showing that for "generic" signals, the program can deconvolve $\boldsymbol{w}$ and $\boldsymbol{x}$ exactly when the maximum of $N$ and $K$ is almost on the order of $L$. That is, we show that if $\boldsymbol{x}$ is drawn from a random subspace of dimension $N$, and $\boldsymbol{w}$ is a vector in a subspace of dimension $K$ whose basis vectors are "spread out" in the frequency domain, then nuclear norm minimization recovers $\boldsymbol{w}\boldsymbol{x}^*$ without error. We discuss this result in the context of blind channel estimation in communications. If we have a message of length $N$ which we code using a random $L\times N$ coding matrix, and the encoded message travels through an unknown linear time-invariant channel of maximum length $K$, then the receiver can recover both the channel response and the message when $L\gtrsim N+K$, to within constant and log factors.

Blind deconvolution using maximum time delay slice of third order moment

Blind Deconvolution of Spotted Image Based on Slice Selection of Third-Order Moment

Multi-frame Blind Deconvolution Using Sparse Priors

Blind Deconvolution Using an Improved L0 Sparse Representation

Blind Image Deconvolution Using the Zero-Lag Slice of Higher-Order Statistics

Blind channel estimation using zero-lag slice of third-order moment

High-resolution multi-path time delay estimation using 2D blind deconvolution in third order cross-moment domain

Lp (1≤p≤∞) BLIND ADAPTIVE DECONVOLUTION AND ITS APPLICATIONS

Multi-frame Partially Saturated Images Blind Deconvolution

Identifiability of Multichannel Blind Deconvolution and Nonconvex Regularization Algorithm

A Blind Deconvolution Method for Attenuative Materials Based on Asymmetrical Gaussian Model.

Blind Deconvolution for Symmetric Point-spread Functions

Improving Time Estimation by Blind Deconvolution: with Applications to TOFD and Backscatter Sizing

Multichannel Sparse Blind Deconvolution on the Sphere

Algorithm for Multichannel Blind Deconvolution Based on Subspace Decomposition

A Robust Multi-Frame Image Blind Deconvolution Algorithm Via Total Variation

Blind Image Deblurring Using Spectral Properties of Convolution Operators

Audio Signal Blind Deconvolution Based on the Quotient Space Hierarchical Theory.

Blind Deconvolution using Convex Programming

Blindly Deconvolving Super-noisy Blurry Image Sequences

Blind single-frame deconvolution by tangential iterative projections (TIP)