Abstract:The blind deconvolution problem amounts to reconstructing both a signal and a filter from the convolution of these two. It constitutes a prominent topic in mathematical and engineering literature. In this work, we analyze a sparse version of the problem: The filter $h\in \mathbb{R}^\mu$ is assumed to be $s$-sparse, and the signal $b \in \mathbb{R}^n$ is taken to be $\sigma$-sparse, both supports being unknown. We observe a convolution between the filter and a linear transformation of the signal. Motivated by practically important multi-user communication applications, we derive a recovery guarantee for the simultaneous demixing and deconvolution setting. We achieve efficient recovery by relaxing the problem to a hierarchical sparse recovery for which we can build on a flexible framework. At the same time, for this we pay the price of some sub-optimal guarantees compared to the number of free parameters of the problem. The signal model we consider is sufficiently general to capture many applications in a number of engineering fields. Despite their practical importance, we provide first rigorous performance guarantees for efficient and simple algorithms for the bi-sparse and generalized demixing setting. We complement our analytical results by presenting results of numerical simulations. We find evidence that the sub-optimal scaling $s^2\sigma \log(\mu)\log(n)$ of our derived sufficient condition is likely overly pessimistic and that the observed performance is better described by a scaling proportional to $ s\sigma$ up to log-factors.

Blind Speech Deconvolution Via Pretrained Polynomial Dictionary and Sparse Representation

Multi-frame Blind Deconvolution Using Sparse Priors

Sparse Blind Speech Deconvolution with Dynamic Range Regularization and Indicator Function

Blind Deconvolution Using an Improved L0 Sparse Representation

Underdetermined Blind Source Separation of Speech Mixtures Unifying Dictionary Learning and Sparse Representation.

Toeplitz-based Blind Deconvolution of Underwater Acoustic Channels Using Wideband Integrated Dictionaries.

Joint L1-L2 Regularisation for Blind Speech Deconvolution.

Blind source separation by sparse decomposition in a signal dictionary

Blind DOA Estimation in a Reverberant Environment Based on Hybrid Initialized Multichannel Deep 2-D Convolutional NMF with Feedback Mechanism

Multichannel Sparse Blind Deconvolution on the Sphere

Approach Based on Colored Character to Blind Deconvolution for Speech Signals

Blind Channel Identification for Speech Dereverberation Using L1-Norm Sparse Learning

Guaranteed blind deconvolution and demixing via hierarchically sparse reconstruction

Blind Deconvolution Based on Temporal Predictability and PSO

Multichannel blind speech source separation with a disjoint constraint source model

Blind image deblurring based on sparse prior of dictionary pair

Blind Room Parameter Estimation Using Multiple-Multichannel Speech Recordings

Blind Convolutive Separation Algorithm for Speech Signals Via Joint Block Diagonalization

Convolutive Blind Speech Separation By Decorrelation

A Nonconvex Approach for Exact and Efficient Multichannel Sparse Blind Deconvolution

Adaptive Speech Enhancement Using Sparse Prior Information.