Page 216 - Fundamentals of Radar Signal Processing

requirements compared to RF sampling. Furthermore, the intermediate

frequency (IF) is chosen so that required complex multiplications by functions
of the form exp(jω n) reduce to particularly simple forms. Second, they use a
0
combination of digital filtering and down sampling to obtain a final output
consisting only of the desired sideband of the original spectrum, sampled at or
near the appropriate Nyquist rate of β complex samples per second. Two
approaches are briefly described here.

The first method, which is particularly elegant, is described in (Rader,
1984.) The RF signal is assumed to have a bandpass spectrum with an
information bandwidth of β Hz. Figure 3.20 is a block diagram of the system,
and Fig. 3.21 sketches the signal spectrum at various points in the system. The
first step is an analog frequency shifting operation that translates this spectrum
to a low IF of β Hz. The bandpass filter rejects the double frequency terms
created by the mixer. The spectrum is therefore bandlimited to ±β/2 Hz, so the

Nyquist rate is 3β samples per second. However, for reasons that will become
clear shortly a higher sampling rate of 4β samples per second is used, giving a
discrete-time signal with the spectrum shown in Fig. 3.21c.

FIGURE 3.20 Architecture of Rader’s system for digital generation of in-phase
and quadrature signals. (After Rader, 1984.)