A novel method based on zoom fast Fourier transform（FFT） is proposed for minimizing the burden processing of cross-ambiguity functions without affecting performance. The low-pass anti-aliasing filter in zoom FFT is ...A novel method based on zoom fast Fourier transform（FFT） is proposed for minimizing the burden processing of cross-ambiguity functions without affecting performance. The low-pass anti-aliasing filter in zoom FFT is realized by using the multistage filtering technique and the weighting processing is employed in the first stage filter to get rid of the redundancy of the computation. In practical systems, the input data is divided into overlapped data frames to avoid loss of detection ability which results in the rapid increase of computational complexity. A segment technique is also proposed in which CAF calculation of overlapped data frames is viewed as slide window FFT to decrease the computational complexity. The experimental results show that compared to the conventional methods, the proposed method can lower computational complexity and is consistent with the real time implementation in existing high-speed digital processors.展开更多
基金Sponsored by the Excellent Young Scholar Research Fund of Beijing Institute of Technology (000Y01-5)BIT(UBF 200501F4208.4)
文摘A novel method based on zoom fast Fourier transform（FFT） is proposed for minimizing the burden processing of cross-ambiguity functions without affecting performance. The low-pass anti-aliasing filter in zoom FFT is realized by using the multistage filtering technique and the weighting processing is employed in the first stage filter to get rid of the redundancy of the computation. In practical systems, the input data is divided into overlapped data frames to avoid loss of detection ability which results in the rapid increase of computational complexity. A segment technique is also proposed in which CAF calculation of overlapped data frames is viewed as slide window FFT to decrease the computational complexity. The experimental results show that compared to the conventional methods, the proposed method can lower computational complexity and is consistent with the real time implementation in existing high-speed digital processors.
