Analysis of the influence of fixed-point computing on quantization noise in software DDC
Journal Title: Вісник НТУУ КПІ. Серія Радіотехніка, Радіоапаратобудування - Year 2015, Vol 0, Issue 60
Abstract
Introduction. With the development of digital technology more attention is paid to making radio-receiving tracks based on SDR technology (Software Defined Radio), which is based on the processing of digitalized radio-frequency signals exclusively by means of software in real time. Moreover the construction ofdata-acquisition systems on one integral chip (SoC) is widely used right now. The main fundamental basis of SoC systems is the FieldProgrammable Gate Array (FPGA). One important step of receiving of radio-frequency signals is a digital down conversion, followed by filtration and repeated decimation. Since processing in DDC is made in a fixed calculation format, than before the immediate implementation of this structure on FPGA, it is necessary to analyze the effect offixed calculation bits on the quality of signal processing. Later it can be used to choose DDC internal calculation bits without sacrificing the set quality of signal processing and hardware resources on its implementation. In the well-known papers, analysis of the effect of signal quantization is reduced to two methods: deterministic and probability methods. Deterministic method gives the top-ofthe-line evaluation of quantizing noise at the output of digital system. Probability method is based on evaluation of quantizing noise dispersion at the output of the device. Evaluation of the signal and the quantization noise in the SDDC based on their PSD. In this section, mathematical expressions were obtained for the construction of PSD signal and the quantizing noise at the output of each of the subsystems of the digital DDC. Criterion for selecting SDDC bit is the relative level of PSD quantizing noise at the output of the digital system. Since the ADC is fixed, the choice of bit computing in SDDC should be such that the total quantizing noise in SDDC would not exceed the ADC quantizing noise at the output of the digital system. Example calculation and construction of PSD signal and quantization noise in Matlab for reception QPSK-signal. In this part of the article an example of the analysis of DDC structure for receiving QPSK- signal was considered. Mathematical Matlab's model has been developed which allows to calculate the PSD of the signal and the quantization noise depending on the internal structure of the SDDC bit based on the formulas of section 2 this paper. These expressions make it possible to construct a PSD noise separately from each SDDC source output, and separately estimate the contribution of each source of quantizing noise to the total noise at the output of the digital system. From these results a decision on the choice of a particular quantization step that depends on the selected calculating bit at each stage of processing was made. Conclusions. Quantizing noise dispersion is not an adequate quality evaluation of processing in multi-stage and multi-rate systems, which include software DDC under review. In such systems the sequential changes of the line, as well as the nature of the spectrum of interfering components, including quantization noise, happen. Therefore in this paper I propose a different approach based on the calculation of the power spectral density (PSD) of the quantizing noise at the output of the device, taking into account the filtering effect of all of its stages and repeated decimation.
Authors and Affiliations
N. Pavlenko
Keywords
Related Articles
- EP ID EP308753
- DOI 10.20535/RADAP.2015.60.13-24
- Views 94
- Downloads 0
Accuracy of trajectory estimation algorithms based on distance- measuring surveillance data using different linearization methods
Problem statement: The radar monitoring systems are used broadly for the moving object coordinates determination. Errors of the radar navigation systems are random in nature. The optimal estimation methods and algorithms...
Artificial neural networks as approximate procedure in wireless devices designing problems
Purpose. Present work is dedicated to the optimal configuration selection and training method of neural network (NN). This NN is architecture's element of modified NN ensemble accepted by authors for implementation of fr...
Analysis of sounding signal with the trapezoidal frequency intra-modulation
A signal with a trapezoidal linear frequency modulation and its properties are considered, figures of amplitude spectrum of signal are drawn, figures of ambiguity diagram, cross-correlation functions on different levels...
Comparison of fast-acting of Java on microcomputer Raspberry Pi
Introduction. The speed of a specific set of instructions is important working with highlevel programming language for microcomputers. The method of comparing execution times of two functionally different programs on dif...
Research of resisting of the biological active point for constant and alternative current
Is conducted research of resistance of biologically active point (BAT) on a direct and variable current. Research results are presented. The estimation of intercommunication between resistance of skin and by an electroma...