0195136047.pdf

656 SIGNAL PROCESSING

works as a phase modulator for high-frequency components, a better overall system performance

results compared to each system (FM or PM) alone. This is the idea behind preemphasis and

deemphasis filtering techniques.

Figure 14.3.9(a) shows a typical noise power spectrum at the output of the demodulator in the

frequency interval|f|<Wfor PM, whereas Figure 14.3.9(b) shows that for FM. The preemphasis

and deemphasis filter characteristics (i.e., frequency responses) are shown in Figure 14.3.10.

Due to the high level of noise at high-frequency components of the message in FM, it is

desirable to attenuate the high-frequency components of the demodulated signal. This results in

a reduction in the noise level, but causes the higher frequency components of the message signal

to be attenuated also. In order to compensate for the attenuation of the higher components of the

message signal, one can amplify these components at the transmitter before modulation. Thus, at

the transmitter we need a high-pass filter, and at the receiver we must use a low-pass filter. The

net effect of these filters is to have a flat frequency response. The receiver filter should therefore

be the inverse of the transmitter filter. The modulator filter, which emphasizes high frequencies,

is called the preemphasis filter, and the demodulator filter, which is the inverse of the modulator

filter, is called the deemphasis filter.

If the signal in question is a constant whose value we seek, as is the case sometimes in simple

measurement systems, the measurement accuracy will be enhanced by a low-pass filter with the

smallest bandwidthB. Low-pass filtering, in a sense, carries out the operation of averaging, since

a constant corresponds to the average value (or dc component) and since noise usually has zero

average value. However, some noise will get through the filter and cause the processed signal

z(t) to fluctuate about the true valuex, as shown in Figure 14.3.11. Allowing any sample to fall

somewhere betweenx−εandx+ε, therms error Gis defined by

Noise-power

spectrum

f

N 0

Ac^2

−W W

(b)

(a)

Noise-power

spectrum

f

−W W

Figure 14.3.9Noise power spectrum at de-

modulator output.(a)In PM.(b)In FM.