Perturbation-based regularization for signal estimation in linear discrete ill-posed problems

Mohamed A. Suliman*, Tarig Ballal, Tareq Y. Al-Naffouri

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

7 Scopus citations


Estimating the values of unknown parameters in ill-posed problems from corrupted measured data presents formidable challenges in ill-posed problems. In such problems, many of the fundamental estimation methods fail to provide meaningful stabilized solutions. In this work, we propose a new regularization approach combined with a new regularization-parameter selection method for linear least-squares discrete ill-posed problems called constrained perturbation regularization approach (COPRA). The proposed COPRA is based on perturbing the singular-value structure of the linear model matrix to enhance the stability of the problem solution. Unlike many regularization methods that seek to minimize the estimated data error, the proposed approach is developed to minimize the mean-squared error of the estimator, which is the objective in many estimation scenarios. The performance of the proposed approach is demonstrated by applying it to a large set of real-world discrete ill-posed problems. Simulation results show that the proposed approach outperforms a set of benchmark regularization methods in most cases. In addition, the approach enjoys the shortest runtime and offers the highest level of robustness of all the tested benchmark regularization methods.

Original languageEnglish (US)
Pages (from-to)35-46
Number of pages12
JournalSignal Processing
StatePublished - Nov 2018

Bibliographical note

Publisher Copyright:
© 2018


  • Ill-posed problems
  • Linear estimation
  • Linear least squares
  • Perturbed models
  • Regularization

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering


Dive into the research topics of 'Perturbation-based regularization for signal estimation in linear discrete ill-posed problems'. Together they form a unique fingerprint.

Cite this