Implementation of GPU-based Iterative Shrinkage-thresholding Algorithm in sparse microwave imaging

In this paper, we present the implementation of Iterative Shrinkage-thresholding Algorithm (ISTA) based on Graphic processing unit (GPU) parallel computation for sparse microwave imaging. First we introduce the theory of sparse microwave imaging and the mathematical model of Lq-norm regularization. Then taking the fast speed advantage of GPU on large-scale computation, we implement the ISTA with parallel computation via CUDA and apply it into sparse microwave imaging. The experiment simulations show that GPU has the same ability in signal reconstruction as CPU, which has less execution time and higher efficiency.