Researches On Sub-nyquist Sampling Methods For Impulse-UWB Communications

Digitizing analog signals is the fundamental element of modern communication theory. To sample Ultra-Wideband (UWB) pulses with extremely wide bandwidth (greater than 500MHz), the digital UWB receivers require Analog-to-Digital Converter (ADC) with sampling frequency in excess of several gigahertz. Unfortunately, it's very difficult and expensive to implement such high-rate, high-precision, low-power ADC with present semiconductor technology. Therefore, study of the sub-Nyquist rate sampling and reconstruction methods, among with the design of digital UWB receivers are of great importance.We first review and summarize the reseach status of UWB wireless communications, then the challenges in designing UWB communication systems and methods to reduce the ADC sampling rate are investigated. Based on the impact of UWB pulse's characteristic on the sampling, reconstruction and design methods for digital receiver, the main research motivation and startpoint of this thesis are indicated.Ultra-wideband transmission has recently received great attention in both academia and industry for applications in wireless communications. The most difficult job in implementing digital UWB receiver is to design a several GHz sampling rate ADC. The disadvantage of present ADC device and shortcomings of previous sub-Nyquist sampling methods are presented. Then the proposed sub-Nyquist sampling methods and their improvements are discussed in detail.A sinc sampling kernel is presented for non-bandlimited impulse signals to reduce the sampling rate. Theoretical analysis shows that the impulse signals can be sampled at or above the innovation rate, and then Least Squares Annihilating Filter (LSAF) reconstruction method is provided. When the number of impulses increases and the noise is present, the LSAF method suffers from performance loss. Then Singular Value Decomposition (SVD) method is proposed to improve the reconstruction performance. Analysis and simulation results show that the proposed sampling and reconstruction methods can achieve good reconstruction results in the presence of noise at much lower sampling rate. Next, sub-Nyquist sampling, parameter estimation and waveform reconstruction methods of UWB pulses are studied. Firstly, the spectrum of the UWB signals is shown and a sub-Nyquist bandpass sampling method is proposed on the basis of lowpass sampling of non-bandlimited signals with finite innovation rate. Sampling rate of the proposed method is related to the signal innovation rate and characteristic of the bandpass filter, which is about one tenth of the Nyquist rate. Secondly, two parameter estimation methods are proposed for the sub-Nyquist sampling UWB signals. The first is the Total Least Squares (TLS) method; and the other is the Estimation of Signal Parameters via Rotational Invariance Techniques (ESPRIT) method. It is shown that both methods can accurately estimate the parameters of the UWB pulses when SNR is high, and ESPRIT method outperforms TLS method when SNR is low. The waveform of UWB signals can be reconstructed by convolving Gaussian Monocycle with the estimated impulse signals. Experiment results of the UWB test bed show that the proposed sampling, reconstruction, and parameter estimation methods can achieve good performance.Finally, digital UWB receiver based on the frequency-domain sampling architecture is proposed. Frequency-domain ADCs that composed of analog filter banks followed by lower speed ADCs are presented for UWB signals with very large bandwidth. The number of ADCs necessary to fully sample the signal bandwidth is derived from theoretical analysis. The quantization noise and optimal bit allocation of the frequency-domain ADCs are discussed in detail. A fully digital UWB receiver is proposed based on the frequency-domain sampling signals. Frequency-domain synchronization, channel estimation and demodulation algorithms are presented. Analysis and computer simulation show that with much lower sampling speed, the frequency-domain UWB receiver can achieve BER performance comparable with conventional high sampling rate digital UWB receiver.