Research Of Receiving Technologies For Communication Systems With Low-precision Quantization

In wireless communications, great effort has been dedicated to improve the power efficiency. The power consumption here consists of radiation power and processing power. As the spatial multiplexing increases, the communication distance trends to be shorter, thus making the processing power, especially the processing power at the receiver, becomes the dominant component compared with radiation power. The pro-cessing power of the receiver mainly includes two components, the power consumption of the analog-to-digital conversion (ADC), and the power consumption of the baseband processing. The power consumption of ADC is proportional to sampling frequency, and exponential to the bit width of ADC. The power consumption of baseband processing is proportional to the square of ADC bit width. In the future, communication system requires large bandwidth to achieve high data rates, rendering conventional high-speed high-resolution ADC a bottleneck due to its high power consumption and massive da-ta needed to be processed by the baseband. Thus, low-precision ADC has been in-troduced into the digital receiver with large-bandwidth, as a promising solution with low-complexity and low power consumption.This dissertation focuses on the receiving technologies with low-precision ADC, including three aspects:receiver design and performance analysis, impact of system non-ideal factors, and impact of inter-symbol interference (ISI). The main results are summarized as follows:For the first aspect, we consider the receiver design and performance analysis for quadrature modulation system with low-precision ADC. The optimal monobit receiver under Nyquist sampling is obtained and its performance is analyzed. Then, a subopti-mal but low-complexity receiver is proposed. Numerical simulations show that the low-complexity suboptimal monobit receiver suffers3dB signal-to-noise-ratio (SNR) loss in additive white Gaussian noise (AWGN) channels and only1dB SNR loss in multi-path channels compared with matched-filter monobit receiver with perfect channel state information (CSI). Further, the optimal receiver for QPSK with Z-level quantization is obtained, and the bit log-likelihood ratio (LLR) is derived. Approximating the bit LLR as Gaussian variable, the bit error rate (BER) of the optimal receiver is analyzed, and the relation between BER performance and ADC precision is given. Theoretical and simulation results show that the performance gain provided by increasing quantization level is less than0.1dB when quantization level is large than8. Analysis of phase offset, quantization threshold, and performance gain of increasing quantization level shows the effectiveness of the proposed analysis method. Besides, a minimum mean square error (MMSE) based practical piece-wise linear receiver is proposed to reduce implementa-tion complexity and increase robust.For the aspect of non-ideal factors, the impact of imbalances between In-phase (I) and Quadrature (Q) branches is carefully examined for monobit sampling. It is demon-strated that the amplitude imbalance has essentially no effect on monobit receivers, and phase imbalance greatly affects the receiving performance. To combat the performance loss caused by IQ imbalances, monobit receivers based on double training sequences are proposed, which is shown to efficiently compensate for the SNR loss without com-plexity increase for AWGN and sparse multipath channels. Then, a constellation rota-tion scheme at the transmitter side is proposed for coded setup, which can offer about0～3dB SNR gain for convolutional code with rates1/2～7/8. Furthermore, an eight-sector phase quantization receiver is proposed, which can almost completely eliminate the SNR loss caused by IQ imbalances. In dense multipath channels, the effect of im-balances is slight.For the aspect of the ISI impact, we focus on the cases with high date rate, e.g.100mega-bits per second (Mbps) or more. For implementation simplicity, the interference from other symbols is approximated as Gaussian noise, and thus the monobit receiv-er for the ISI case is obtained. To boost receiving performance, high autocorrelation training sequence is employed. The output SNR loss and BER of the proposed mono-bit receiver is analyzed, based on the typical Saleh-Valenzuela (S-V) model of indoor channels and the Gaussian-approximated interference. The simulation and theoretical results demonstrate that the proposed receiver has only about1dB SNR loss compared with the one without ISI when the data rate is up to250Mbps, showing that there is no need for complex equalization in this case. For the case with much more severe ISI, a time-domain equalization algorithm based on decision feedback equalization and iterative demodulation is proposed. The simulation results show that such algorithm has only about1.2dB SNR loss compared with that without ISI, when the data rate is500Mbps.