Analysis of the Influence of Software and Hardware Noise on the Results of Digital Signal Processing

Abstract

The achievement of a higher quality signal processing occurs, among other things, due to the complication of its processing algorithms, i. e., an increase in the amount of calculations. This processing must be carried out in real time. One of the ways to overcome this problem is to split a complex algorithm into several simpler parts that must be executed in parallel and then combined. For this purpose, digital computing platforms are used, which are built on the basis of various discrete elements, such as programmable logic integrated circuits, microcontrollers or multi-core digital signal processors. Based on the results of the research, a mathematical expression was obtained that shows that the final signal-to-noise ratio significantly depends on the processing frequency. This ratio also depends on the element base on the basis of which digital processing algorithms are implemented. Using the characteristics of individual chips as an example, it is shown that the use of certain chip types is preferable in some cases because it allows the processing to be implemented in a wider clock frequency range.