Gradient-Descent Algorithm Performance With Reduced Set of Quantized Measurements

The quantization (digitalization) of measurements greatly affects the reconstruction performance, especially in algorithms based on the reconstruction in the measurement domain. However, it provides a significant advantage in the hardware implementation sense. In this paper, we analyze the performance of the gradient-based algorithm in the signal reconstruction based on a reduced set of digital measurements. This algorithm is considered as a powerful tool for the reconstruction of various types of signals. The paper investigates the accuracy of the algorithm using $B$-bit quantized measurements. The reconstruction performance is analyzed through numerical examples.