| Antenna arraying techniques can effectively improve the quality of receivedsignals, which can improve the system stability and reliability as well. Smallantenna arraying is the trend at present, where the signal-to-noise ratio (SNR)of the received signals is very low. Therefore, research on signal processingmethods of low SNR signals has important significance.This dissertation studies the key technologies of signal combining in antennaarraying, which focuses on joint time delay estimation algorithm and implementof signal combining. A system based on FPGA and DSP is also designed formulti-antenna signal combining. The contributions obtained in this thesis can besummarized as follows.1The Cramer-Rao lower bound (CRLB) for time delay estimation is derived withantenna placement and modulation types of the signals totally lacking. Fouriertransform of the received signals is used to derive the probability densityfunction (PDF) in the frequency domain. Then, the CRLB is deduced as a functionof antenna number, SNR, etc.. It is remarkable that the estimation accuracyincreases as the number of antenna elements increases. Moreover, it is shown thatincreasing the number of antennas results in a small benefit at high SNR values.However, the benefit becomes more significant for low SNR conditions. Besides,there is a theoretical lower bound for the CRLB when the number of antennas isinfinite or the SNR in one sensor is much larger than that in the others.2To solve the problem of time delay estimation in signal combing, an efficienttime delay estimation algorithm based on multiple reference signals is presented.We designate multiple high-SNR signals rather than a single fixed signal asreferences. Each of these reference signals is correlated with all the others toderive time-delay vector. Each of the time delays to be estimated can be expressedby several linear expressions of multiple elements of the vector. Since randomerrors among these expressions are partly correlated, it is possible for us toimprove the estimating performance by adding proper weights to all theseexpressions. The estimating performance and computational complexity can becontrolled by increasing or decreasing the number of reference signals accordingto the requirement. The performance of the new algorithm is higher than thesemethods with a single fixed reference. And it is a more flexible and efficient method as compared to the optimal method. Both theoretical analysis and simulationexperiment show that the estimation performance increases as the number ofreference signals.3To solve the problem of signal combing for low SNR signals, parallel signalprocessing system of time delay estimation and signal combining is presented. Atfirst, the performance of different combining algorithms is compared. And thecombining algorithm based on multiple reference signals is selected as our basiccombining scheme. Both time delay estimation method and combining algorithm areparallel decomposed. The signal processing method based on point-by-point is usedto deal with time delay estimation, time delay compensation, and signal combining.This scheme is suitable for most parallel devices such as Graphic Processing Unit(GPU), Field-Programmable Gate Array (FPGA), and Multi-core processor, which makesit possible for real-time signal combining.4A multi-antenna combining system is designed and implemented on the hardwareplatform of FPGA and DSP. It is mainly composed of whole system design, analysisof chip resources, Nios II system design, and implement of main modules. At last,the system is tested which meets the requirement. |
PDF Full Text Request