Table of Contents
Advertisement
MTT-14B e_Manual D07-00-083P RevA00
Sampling
Pulse Amplitude Modulation
Figure 57 Converting a Voice Signal
The Nyquist theorem requires that the signal be sampled at
twice the signal's maximum frequency in order for the signal to
be reproduced without a loss of information. For voice signals,
the maximum frequency is approximately 4000 Hz. This provides
adequate clarity for voice transmission bandwidth. Thus, we
must sample our 4000 Hz voice signal at a frequency of 8000
Hz (8000 samples/second).
The amplitude of the analog voice signal is sampled 8000 times
per second. Each amplitude value is expressed as an 8-bit code
'word'. These 8-bit words occurring 8000 times per second form
a 64 kbit/s digital bit stream.
The 8-bit code word is formed by comparing the amplitude of the
analog sample to a companding characteristic. This characteristic
is a formula which translates the amplitudes of the samples into
the 8-bit code words. Internationally, a companding characteristic
known as 'A-law' is used. The purpose of A-law is to provide
optimum signal-to-noise performance over a wide ranger of
transmission levels. Linear encoding provides a poorer signal-to-
noise ratio at the -20 dB level typical of speech. In North America,
the encoding is done according to the Mu-Law. Therefore, the
companding law used for encoding the voice signal must match
that for decoding, for distortion-free transmission.
4.2.4 2.048 Mbit/s Data Rate
The E1 signal (bitstream) is transmitted at a rate of 2.048 Mbit/s
(2 048 000 bits per second). This transmission rate is achieved
by combining 32 individual 64 kbit/s bitstreams:
64 (kbit/s /Channel) x 32 (Channels) = 2048 kbit/s = 2.048 Mbit/s
This 2.048 Mbit/s signal is the overall E1 transmission rate.
Voice Signal
Time
Quantization
00101 101
101 101 10
Time
11000100
Page 83 of 101
Advertisement
Table of Contents
