Recovery Probability Analysis for Sparse Signals via OMP

PDF

Authors
  1. Niu, M.
  2. Salari, S.
  3. Kim, I.-M.
  4. Chan, F.
  5. Rajan, S.
Corporate Authors
Defence Research and Development Canada, Ottawa Research Centre, Ottawa ON (CAN)
Abstract
In the context of compress sensing (CS), it is known that use of a random measurement (sensing) matrix usually results in good recovery performance via Orthogonal Matching Pursuit (OMP). The goal of this paper is to provide the probability ensuring the recovery of sparse signals using OMP for the case where all entries of the measurement matrix are independently selected from a Gaussian distribution. Our analysis relies on the mutual coherence (MC) property of the sensing matrix. The direct calculation of the probability of perfect recovery is very difficult. Therefore, we will apply a bounding technique to approximate the MC, which makes the probability analysis tractable. The accuracy of the analysis is well demonstrated by our extensive numerical experiments.
Keywords
compressive sensing;orthogonal matching pursuit OMP;mutual coherence;probability of recovery
Report Number
DRDC-RDDC-2016-P008 — External Literature
Date of publication
21 Jan 2016
Number of Pages
6
DSTKIM No
CA041920
CANDIS No
803078
Format(s):
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: