Improving recovery of ECG signal with
deterministic guarantees using split signal
for multiple supports of matching pursuit
(SS-MSMP) algorithm
Israa Shaker Tawfic
a,
*, Sema Koc Kayhan
b
a
Electric and Electronic Engineering, Ministry of Science and Technology, Baghdad, Iraq
b
Electric and Electronic Engineering, University of Gaziantep, Gaziantep,Turkey
ARTICLE INFO
Article history:
Received 18 June 2016
Received in revised form
9 September 2016
Accepted 18 October 2016
ABSTRACT
Compressed sensing (CS) is a new field used for signal acquisition and design of sensor that
made a large drooping in the cost of acquiring sparse signals. In this paper, new algo-
rithms are developed to improve the performance of the greedy algorithms.
In this paper, a new greedy pursuit algorithm, SS-MSMP (Split Signal for Multiple Support
of Matching Pursuit), is introduced and theoretical analyses are given. The SS-MSMP is sug-
gested for sparse data acquisition, in order to reconstruct analog and efficient signals via a
small set of general measurements. This paper proposes a new fast method which depends
on a study of the behavior of the support indices through picking the best estimation of
the corrosion between residual and measurement matrix.
The term multiple supports originates from an algorithm; in each iteration, the best support
indices are picked based on maximum quality created by discovering correlation for a par-
ticular length of support.
We depend on this new algorithm upon our previous derivative of halting condition that
we produce for Least Support Orthogonal Matching Pursuit (LS-OMP) for clear and noisy signal.
For better reconstructed results, SS-MSMP algorithm provides the recovery of support
set for long signals such as signals used in WBAN.
Numerical experiments demonstrate that the new suggested algorithm performs well
compared to existing algorithms in terms of many factors used for reconstruction performance.
© 2016 Elsevier Ireland Ltd. All rights reserved.
Keywords:
Compressed Sensing
MSMP Algorithm
Least Support Orthogonal Matching
Pursuit
percentage root-mean square
difference PRD, Average
support - cardinality error ASCE
1. Introduction
Signal processing on its core is concerning about an effective
algorithm for extracting data of signals. In order to design such
algorithms for a specific problem, we must have precise models
for the concerned signal. These could be a form of obstetric
models, probabilistic models or even deterministic classes.
According to these challenges, a low dimensional model has
been suggested [1]. Such a low-dimensional model constructs
a mathematical structure for capturing important data from
those high-dimensional signals as long as most of their
* Corresponding author. Ministry of Science and Technology/ Information Technology, Baghdad, Iraq.
E-mail address: isshakeralani@yahoo.com (I.S. Tawfic).
http://dx.doi.org/10.1016/j.cmpb.2016.10.014
0169-2607/© 2016 Elsevier Ireland Ltd. All rights reserved.
computer methods and programs in biomedicine 139 (2017) 39–50
journal homepage: www.intl.elsevierhealth.com/journals/cmpb