| As a basic approach of array signal processing, the beamforming technique has been widely used in many important applications including radar, sonar, and communications. It is mainly to realize two functions: 1) obtaining spatial processing gain of signals of interest against noises, interferences or reverberations, and 2) achieving high spatial resolution for more accurate target detection and localization. On the other hand, wide-band signals are frequently used in sonar, radar and communications systems. It is thus highly desired to develop wide-band beamforming techniques and investigate their practical performances.In this thesis, some basic concepts involving beamforming and space filtering are first introduced, and then some traditional methods are reviewed, including time-domain interpolation beamformer and frequency-domain beamformer. After that, a time-domain method applying the Polyphase FIR Filter is developed, and for the bandpass signal, a bandpass sampling beamforming approach based on the Polyphase FIR Filter is proposed. The proposed approach has the following features: 1) the quadrature demodulation is naturally realized at the output of bandpass sampling through a proper choice of sampling rate, thus eliminating analog components for quadrature demodulation; 2) a true time delay beamforming is achieved by a single FIR filter, which combines all the related operations including lowpass filtering, interpolation delaying, and phase rotating; 3) the interpolation coefficients are determined through the coefficients of the lowpass filter used to obtain the desired interpolation point of signals in the zero-padding interpolation beamforming method. This method can be applied specially to wide-band signals.In a practical multi-channel system, it is hard to maintain consistency in response across different channels, particularly for wide-band signals. Thus some channel calibration is needed in order to reduce the system performance loss due to the response mismatch. In this thesis, a channel mismatch model is developed, and techniques on channel calibration are investigated using the characteristic eigensubspace approach and the channel equalization approach.Computer simulations are conducted to evaluate the developed approaches. A complete time-domain beamforming procedure with channel calibration and matched filtering is implemented. Performances of the proposed time-domain approach are analyzed and compared to that of the conventional frequency-domain method. The results show no obvious difference in simulation performances, however, the time-domain beamforming approach proposed in this thesis can save resources required for hardware realization when the number of beams is small.Finally, the proposed bandpass sampling beamforming approach is realized in a real DSP-FPGA platform. A cascade multi-channel FIR filtering structure is developed, which fits particularly well to FPGA by efficiently using the relevant logic resources. Both the VHDL simulation and real signal test indicate that the VHDL code developed implements the designed function of wide-band beamforming while meeting the real timing constraints. |
