Design of Intelligent Channelizer for Extracting Signals of Arbitrary Centre Frequency and Bandwidth

Abstract

There are two standard approaches to the problem of wideband signal channelization, namely those based upon the use of a digital down conversion (DDC) unit and those based upon the use of a polyphase discrete Fourier transform (DFT) or PDFT. There are clear advantages and disadvantages with both approaches, however, in that: a) with the DDC approach, optimal performance and flexibility is obtained but at the expense of a heavy computational load; whereas b) with the PDFT approach, a sub optimal and less flexible performance is obtained but at a greatly reduced computational cost through the exploitation of a suitably defined fast Fourier transform (FFT). An intelligent channelizer is described herein which possesses a flexible design able to exploit the merits of both approaches for the case where the input data comprises real-valued samples. The two key design features are: a) optimal setting of the PDFT parameters to ensure that for every signal of interest there is at least one channel completely containing it; and b) simultaneous computation of two real-data FFTs: the first as required by the PDFT and the second, a high-resolution FFT with high update rate, able to accurately direct the application of the low-rate DDC units to the relevant PDFT channel outputs.