Page 308 - Satellite Communications, Fourth Edition

288 Chapter Ten

and the symbol rate in terms of bit rate is

R b
R sym m (10.5)

For satellite transmission, the encoded message must be modulated
onto the microwave carrier. Before examining the modulation process,
we describe the way in which speech signals are converted to a digital
format through pulse code modulation.

10.3 Pulse Code Modulation

In the previous section describing baseband digital signals, the infor-
mation was assumed to be encoded in one of the digital waveforms
shown in Figs. 10.2 and 10.3. Speech and video appear naturally as
analog signals, and these must be converted to digital form for trans-
mission over a digital link. In Fig. 10.1 the speech and video analog sig-
nals are shown converted to digital form through the use of A/D
converters. The particular form of A/D conversion used is known as
pulse-code modulation (PCM). Commercially available integrated cir-
cuits known as PCM codecs (for coder-decoder) are used to implement
PCM. Figure 10.4a shows a block schematic for the Motorola MC145500
series of codecs suitable for speech signals. The analog signal enters at
the Tx terminals and passes through a low-pass filter, followed by a high-
pass filter to remove any 50/60-Hz interference which may appear on
the line. The low-pass filter has a cutoff frequency of about 4 kHz, which
allows for the filter rolloff above the audio limit of 3400 Hz. As shown
in connection with single-sideband systems, a voice channel bandwidth
extending from 300 to 3400 Hz is considered satisfactory for speech.
Band limiting the audio signal in this way reduces noise. It has another
important consequence associated with the analog-to-digital conversion
process. The analog signal is digitized by taking samples at periodic
intervals. A theorem, known as the sampling theorem, states in part that
the sampling frequency must be at least twice the highest frequency in
the spectrum of the signal being sampled. With the upper cutoff fre-
quency of the audio filter at 4 kHz, the sampling frequency can be stan-
dardized at 8 kHz.
The sampled voltage levels are encoded as binary digital numbers in
the A/D converter following the high-pass filter. The binary number
which is transmitted actually represents a range of voltages, and all
samples which fall within this range are encoded as the same number.
This process, referred to as quantization, obviously will introduce
some distortion (termed quantization noise) into the signal. In a prop-
erly designed system, the quantization noise is kept well within
acceptable limits. The quantization steps follow a nonlinear law, with