Multichannel digital processing of wideband signals

The feasibility of processing wideband signals by discretization with high speed narrow band converters is examined, conventional analog to digital converters not being suitable for processing signals more than 20 MHz wide. Subdividing the entire frequency range into N narrow channels is proposed as an alternative, with discretization in each channel to be followed by reconstruction of signal readings from readings in all those channels. For illustration, the signal is assumed to have a finite spectrum. Narrow channels are formed by means of quadrature conversion at the corresponding center frequencies with the aid of low-pass filters, whereupon analog to digital conversion is performed at the discretization frequency. The amplitude-frequency charactristics of these filters are trimmed and matched by means of a discrete Fourier transformation with the use of apertures such as Hemming or Kaiser apertures. The output signals from the filter array are subsequently processed by conventional two-dimensional methods for one-dimensional sequences.