Implementation of the OMP algorithm for sparse recovery tasks in ODFM systems

This paper presents several aspects of the implementation of the OMP algorithm, which is a very popular choice for solving sparse reconstruction problems. We consider the implementation for a class of such problems that occur in OFDM systems, which allows for instance the usage of an FFT for parts the calculations. Different methods for the solution of an occurring least squares sub-problem are presented and compared.We give an overview of the overall computational complexity for the different methods and discuss the topics of convergence and error propagation.