Performance Evaluation of FISTA with Constant Inertial Parameter

Compressed sensing is a framework for reconstructing sparse signals from their underdetermined linear measurements. In this paper, we discuss compressed sensing algorithms appropriate for the implementation with optical circuits, which have attracted attention as a technology that can perform vector-matrix products at high speed with low power consumption. Although the fast iterative shrinkage thresholding algorithm (FISTA) is one of the basic compressed sensing algorithms, updating the inertial parameters is difficult to implement in optical circuits. We thus propose an algorithm by replacing the inertial parameter of FISTA with a constant and evaluate its performance via computer simulations. Simulation results show that the proposed algorithm with an appropriate inertial parameter can achieve comparable performance to the original FISTA.