Agilent Technologies Switch/Measure Manual de usuario Pagina 27

  • Descarga
  • Añadir a mis manuales
  • Imprimir
  • Pagina
    / 36
  • Tabla de contenidos
  • MARCADORES
  • Valorado. / 5. Basado en revisión del cliente
Vista de pagina 26
27
GSM speech coding
To transmit actual voice data would require more bandwidth than a system can
practically afford. Most digital communication systems use some sort of voice
compression. GSM is no exception. The encoding goal is to represent speech
with the least amount of bits.
Human speech consists of an excitation signal (pitch, loudness) a filter (tongue,
teeth), and redundancy (long vowels and pauses). The GSM speech encoder
produces 260 bits every 20 ms that represent the human voice. The output bits
are separated into groups by their importance at reproducing an acceptable
representation of speech. The most important bits are the Class Ia bits. These
bits contain filter coefficients, and receive the most protection during the chan-
nel coding process. Next in importance are the Class Ib bits, and least impor-
tant are the Class II bits. Class II bits do not receive additional protection from
the channel coder.
During channel decoding, the CRC must be successful or the entire frame is
discarded. Once CRC is verified Class Ib and Class II bits are checked for errors.
The RBER test then becomes bit errors in accepted frames over the total
number of bits in the accepted frames.
FER
The frame erasure rate (FER) indicates how many of the received frames are
bad. The channel decoder erases frames when the CRC fails. FER is performed
on both speech and signaling frames. When a speech frame is discarded, the
system will interpolate. When a signaling frame is discarded, the mobile station
is instructed to resend. In the GSM standards, it is common to see FER limits
multiplied by correctional value α. This correction factor allows the system
designer to trade off between FER and RBER. If the system is designed to
discard a large number of frames, then the FER will be higher (multiply by α)
and RBER should be significantly lower (divide by α). Values for α range
between 1 and 1.6.
Mic
Filter
300 Hz to 4 kHz
A/D 8K
13 bit
104 kbps
Speach
encoder
(compression)
13 kbps
To
Modulator
Speach
encoder
(protection)
Class Ia CRC Class Ib Class II
50 bits 3 bits 132 bits 78 bits
Vista de pagina 26
1 2 ... 22 23 24 25 26 27 28 29 30 31 32 ... 35 36

Comentarios a estos manuales

Sin comentarios