Blind deconvolution for the generalized ar model

Wenyuan Xu, Mostafa Kaveh

Research output: Contribution to journalConference articlepeer-review


Applying the convex cost function Lc: to the blind deconvolution of general non-minimum phase AR(u) models is studied.A simple and realizable constraint is proposed for the L^ deconvolution.With this constraint, except for a gain.the model parameter is the unique solution of the L:o deconvolution.The strong consistency of the estimator of the model parameter defined by the sample version of L^ norm is presented.An algorithm is suggested for the iterative computation of the estimator.Simulation examples show the proposed approach works well for apprepriate blind equalization problems.

Original languageEnglish (US)
JournalEuropean Signal Processing Conference
StatePublished - 2015
Event8th European Signal Processing Conference, EUSIPCO 1996 - Trieste, Italy
Duration: Sep 10 1996Sep 13 1996

Bibliographical note

Funding Information:
∑∞ ck < ∞, which, in communications, describes the k=−∞ intersymbol interference, (ii) the input signal {xk} is a zero-mean i.i.d. random sequence. Denote c={ck,-∞<k<∞}. It is assumed that there exists d={dk, ∞<k<∞ },which is called the inverse filter of c, such that d⊗c=δ, where ⊗ means convolution and δ is a delta sequence. In the most general case, {ck} and the probability distribution of xk are unknown. The blind deconvolution problem for (1) is to estimate both of the input sequence {xn}, and the system response {ck}. Numerous approaches have been developed to solve the blind deconvolution problem. One approach is based on ________________________________________ This work was supported in part by the BMDO/IST program managed by the Office of Naval Reseach under Contract N00014-92-J-1911.

Publisher Copyright:
© 2015 European Signal Processing Conference, EUSIPCO. All rights reserved.


Dive into the research topics of 'Blind deconvolution for the generalized ar model'. Together they form a unique fingerprint.

Cite this