796 Discrete Hilbert Transforms Chap. 11
11.4.2 Representation of Bandpass Signals
Many of the applications of analytic signals concern narrowband communication. In
such applications, it is sometimes convenient to represent a bandpass signal in terms of
a lowpass signal. To see how this may be done, consider the compfex lowpass signal
x[n] = x-[n]} + jxiln],
where x; [71] is the Hilbert transform of x,[n] and
Xe) =0, -n sw <0.
‘The Fourier transforms X,(e/*) and jX;(e*) are depicted in Figures 11.10(a) and
11.10(b), respectively, and the resulting transform X(e#) = X,(el) + jXi(e!) is
shown in Figure 11.10(c). (Solid curves are real parts and dashed curves are imaginary
parts.) Now, consider the sequence
s{n] je" = s,[n] + jsiln), (11.71)
where s,[n] and s;[7] are real sequences. The corresponding Fourier transform is
Sei”) = Xe”), (11.72)
which is depicted in Figure 11.10(d). Applying Eqs. (11.59) to S(e/) leads to the equa-
tions
S,(el#) = 3[S(el”) + S*(e/*)], (11.73a)
jSi(e) = 3[S(ei) - S*(e-/*)). (11.73b)
For the example of Figure 11.10, S,(e”) and j S,(e/”) are illustrated in Figures 11.10(e)
and 11.10(f), respectively, I: is straightforward to show that if X,(e/*) = 0 for Aw <
lol < x, and if, + Aw < 1, then S(e/”) will be a one-sided bandpass signal such that
S(ei®) = 0 except in the interval w, < @ < w: + Aw. As the example of Figure 11.10
illustrates, and as can be shown using Eqs. (111.58) and (11.59), S,(e/*) = H{e"*)S,(e!”),
or, in other words, si[n] is the Hilbert transform of s, [n].
An alternative representation of a complex signal is in terms of magnitude and
phase; ice, x[n] can be expressed as
x[n] = A[nje/#", (11.74a)
where
An] = (x2[n] + x?[n))'? (11.74b)
and
o(n] = arctan 34) (11.740)
‘Therefore, from Eqs. (11.71) and (11.74), we can express s[n] as
sin] = Gln] + jxilnpelo" (11.75a)
= Alnje/errol (11.75b)