Quantization Methods Based on Norms for Complex Gaussian-Like Signals in SDR Applications

Abstract

This paper deals with quantization and clipping methods for complex Gaussian-like signal such as multicarrier waveforms in an software define radio (SDR) context. The process involving the quantization of the real and imaginary parts of any signal independently is a basic of digital signal processing, therefore plenty of papers about these techniques are available in the literature. In this paper, we rather consider the quantization of Gaussian complex signals as a whole. To this end, we base our quantization and clipping methods on norms such as Euclidean or infinity norms, allowing the quantized signal to keep the complex argument of the clipped signal samples unchanged and to reduce the peak to average power ratio (PAPR). Furthermore, we derive the corresponding performance analysis in terms of error vector magnitude (EVM), bit error rate (BER), and complexity. We show that simulations match the theoretical results, and that the suggested quantization methods outperform the usual one and reduces the PAPR of the signal, to the cost of an increase of complexity. Furthermore, we discuss the extension of the principle to any norm, as well as practical signals taken from standards such as 5G for instance.