Page 225 - Fundamentals of Radar Signal Processing
P. 225
that is retained. Discuss the spectrum recentering step in going from the
equivalent of Fig. 3.21d to the equivalent of Fig. 3.21e: can demodulation
by decimation be used and, if not, why not? If multiplication by complex
exponentials is used, do the multipliers assume a simplified form?
Summarize the pros and cons of this system versus Rader’s system.
_____________
1
The terms resolution bins and resolution cells are also sometimes used synonymously with range bins,
and in many cases are synonymous; but the sampling interval in range does not always equal the range
resolution, so caution should be used in interpreting the latter two terms.
2
In this text, the abbreviations “PRF” and “PRI” are used both as acronyms and as mathematical variables.
When used as an acronym, they are not italicized (PRF, PRI); when used as mathematical variables, they
are italicized (PRF, PRI).
3 Note that the velocity of portions of a vehicle may differ from the nominal velocity of the vehicle as a
whole. The Doppler frequency corresponding to the nominal velocity is frequently termed the skin return.
However, the lug nuts on the wheels on a wheeled vehicle have velocities ranging from zero to twice the
nominal velocity of the vehicle. Some laser radars even attempt to measure the Doppler shifts due to vehicle
vibration. The spectrum of Doppler shifts from different portions of a vehicle is sometimes called a
microDoppler spectrum.
4
Seven range bins is unrealistically few in most situations, but is used here for ease of illustration.
5 The precise value depends on the definition of mainlobe width that is used, and whether or not weighting
for sidelobe control is employed.
6
Not to be confused with the “Dirac delta” impulse function δ (·) used in continuous-time analysis.
D
7
Using a DFT size K that is strictly greater than the number of time samples M is often referred to as zero
padding. This term is a relic of traditional software for computing the fast Fourier transform, which often
returned the vector of frequency samples in the same variable used to input the time samples. When K > M,
the input vector was “padded” to the required length of K by adding K–M zeros at the end. The term has
nothing to do with the mathematics of the DFT.
8 Straddle loss is also called scallop loss by some authors, e.g., Harris (1978).
9 This same result is often derived in antenna literature by requiring that the antenna pattern not contain
grating lobes, which are replicas of the antenna pattern caused by sampling of the aperture of a phased
array antenna by the elements.