A/D converter resolution enhancement using neural networks

Abstract

In this paper, we propose a neural network-based A/D converter resolution enhancement scheme. The employed neural network is 'delayless' nonlinear filter, which reduces the quantization noise of the low-resolution A/D converter. Hence, the output resolution can be enhanced after the quantization noise has been reduced. The operating principle, structure, and algorithm of the resolution enhancement scheme are described in detail. We demonstrate the effectiveness of this approach by using both simulated and practical signals. Theoretical analysis and simulation experiments show that our method can improve the band-limited resolution of A/D converters.