| 摘要 |
RD-19,737 A MULTI-RATE SUPERRESOLUTION TIME SERIES SPECTRUM ANALYZER Parallel architectures preprocesses large matrices from sampled coherent time apertures receiving signals from distant sources to produce lower order matrices, derived from pseudo coherent time apertures, which are computationally less burdensome. The large matrices are processed by frequency shifting, low pass filtering with an FIR filter, and executing front-end decimation to create the pseudo coherent time apertures, each corresponding to different subbands of the temporal frequency spectrum. The signals representing the pseudo coherent time apertures are processed using matrix based superresolution spectral estimation algorithms such as the Tufts-Kumaresan (T-K) reduced rank modified covariance algorithm and the Linear Minimum Free Energy algorithms to produce an image of the sources.
|
