Page 57 - From GMS to LTE
P. 57

Global System for Mobile Communications (GSM)  43


                                           Signal forming
                Voice,           Determination of  Determination of
                PCM coded        filter parameters  filter parameters   Coding of
                20 ms,            for the vocal    for the period       the ‘rest
                160 values,        tract filter       filter             signal’
                1280 bit

                                       36 Bit          36 Bit               188 Bit
                 Sender
                                      Frame (260 bits) for transmission every 20 milliseconds
                 Receiver

                                          Signal forming
                Voice,
                PCM coded          Vocal tract                          Excitation
                20 ms,               filter        Period filter          signal
                160 values,
                1280 bit


               Figure 1.34  Complete transmission chain with the transmitter and receiver of the GSM FR codec.


               1.7.6  Channel Coder and Interleaver in the BTS
               When a 260‐bit data frame from the TRAU arrives at the base station every 20 milliseconds,
               it is further processed before being sent over the air as shown in Figure 1.35. In the
               reverse direction, the tasks are performed in the mobile device.
                In the first step, the voice frames are processed in the channel coder unit, which adds
               error detection and correction information to the data stream. This step is very important
               as the transmission over the air interface is prone to frequent transmission errors due to
               the constantly changing radio environment. Furthermore, the compressed voice infor-
               mation is very sensitive and even a few bits that might be changed while the frame is
               transmitted over the air interface create an audible distortion. To prevent this, the chan-
               nel coder separates the 260 bits of a voice data frame into three different classes as
               shown in Figure 1.36.
                Fifty of the 260 bits of a speech frame are class Ia bits and are extremely important for
               the overall reproduction of the voice signal at the receiver side. Such bits are, for exam-
               ple, the higher order bits of the filter parameters. To enable the receiver to verify the
               correct transmission of those bits, a three‐bit cyclic redundancy check (CRC) check-
               sum is calculated and added to the data stream. If the receiver cannot recreate the
               checksum with the received bits later on, the frame is discarded.
                The other 132 bits of the frame are also quite important and are thus put into class Ib.
               However, no checksum is calculated for them. To generate the exact amount of bits that
               are necessary to fill a GSM burst, four filler bits are inserted. Afterward, the class Ia bits,
               checksum, class Ib bits and the four filler bits are treated by a convolutional coder that
               adds  redundancy  to  the  data  stream.  For  each  input  bit,  the  convolutional  decoder
   52   53   54   55   56   57   58   59   60   61   62