A Dynamical System with Fixed Convergence Time for Sparse Recovery

Abstract

The sparse recovery (SR) algorithm, under the premise that signals are sparse, can be divided into two categories. One is a digital discrete method implemented via lots of iterative computations and the other is a continuous method implemented via analog circuits, which is usually faster. In this paper, we focus on the continuous method and propose a fixed-time convergence dynamical system. Compared with the existing system, it dynamically allocates the exponent according to time-varying elements of the system state, avoiding possible mismatches between the fixed exponent and some elements.