Communication Systems

Mobile Communication Systems

Lecture (3-2) :- The Digital Signal Processing for
Speech signal in the mobile station.
As we know that the main specification of the second generation of mobile
system depend on the digital technique to transmit the speech signal between all
parts of the system, this technique provide an additional properties such as
compression, ciphering ,error detection and other techniques, the bandwidth for the
speech signal that can be recognized has been limited (from telephony system )
between 300Hz- to 3400Hz , this signal sampled by8000 bit per second and
converted to PCM (digital form) and passed through an additional functional blocks
as shown in Figure(1) below

From figure (1) above one can be see that the speech signal pass through the
following functional Processing blocks in the transmission side:-
1 – Speech Coding
2 – Channel Coding
3 – Interleaving
4- Ciphering
5 – Modulator

1 – Speech coding :-
While the PCM algorithm digitizes analog volume levels by statically mapping
them to digital values, the GSM speech digitization is much more complex to reach
the desired compression rate. At the first the speech signal has been sampled by
8kb/sec an then each sample converted to the 13 bit PCM signal, these digital codes
then compressed to 8-bit per sample, after that another compression is achieved by
emulating the human vocal system. This is done by using a source-filter model as
shown in Figure (2)

The speech forming is simulated by using two time-invariant filters. The period filter
creates the periodic vibrations of the human voice while the vocal tract filter
simulates the envelope. The filter parameters are generated from the human voice,
which is the input signal into the system. The system is simplified by generating a
pair of filter parameters for an interval of 20 milliseconds. As an input to the
algorithm, a speech signal is used that has previously been converted into an 8- from
13-bit PCM codec. As the PCM algorithm delivers 8000 values per second, the FR
codec requires 160 values for a 20 ms interval to calculate the filter parameters. As
eight bits are used per value, 8 bits × 160 values = 1280 input bits are used per 20 ms
interval. For the period filter, the input bits are used to generate a filter parameter
with a length of 36 bits. Afterwards, the filter is applied to the original input
signal. The resulting signal is then used to calculate another filter parameter with a
length of 36 bits for the vocal tract filter. Afterwards, the signal is again sent through
the vocal tract filter with the filter parameter applied. The signal, which is thus
created, is called the ‘rest signal’ and coded into 188 bits.
Once all parameters have been calculated, the two 36-bit filter parameters and the rest
signal, which is coded into 188 bits, are sent to the receiver. Thus, the original
which was coded in 1280 bits, has been reduced to 260 bits. In the receiver, the filter
procedure is applied in reverse order on the rest signal and thus the original signal is
recreated. As the procedure uses a lossy compression algorithm, the original signal
and the recreated signal at the other end are no longer exactly identical. For the
human ear, however, the differences are almost inaudible.

2 – Channel Coding :-
Before a 260-bit data frame is transmitted over the air interface every 20 ms, it
traverses a number of additional functional blocks . In a first step, the voice frames
are processed in the channel coder unit, which adds error detection and correction
information to the data stream. This channel coder process are shown in figure (3)
below.
The channel coder separates the 260 bits of a voice data frame into three different
classes as shown in Figure (3) above.
 - Fifty (50) bits from the 260 bits of a speech frame are class Ia bits and
extremely important for the overall reproduction of the voice signal, such bits
are for example the higher order bits of the filter parameters, a three-bit CRC
checksum is calculated and added to the data stream.
 - Another (132) bits from the frame are also quite important and are thus put
into class Ib. However, no checksum is calculated for them. In order to
generate the exact amount of bits that are necessary to fill a GSM burst, four
(4) filler bits are inserted.
  - The remaining (78) bits from the original 260-bit data frame belong to the
third class which is called class Ic. These are not protected by a checksum and
no redundancy is added for them. Errors which occur during the transmission
of these bits can neither be detected nor corrected.
Afterwards, the class Ia bits, checksum, class Ib bits, and the four filler bits which
represent (189 bits) are treated by a convolutional coder which adds redundancy to
the data stream. For each input bit, the convolutional decoder calculates two output
bits, the resultant outputs of the convolutionl channel coder are ( 378 bits).
The final output of the channel coder contain (456 bits) which are came from the
(378 bits) and the (78 bits ) that are classified as class II.

3 – Interleaving
As has been shown from the previous paragraph , the channel coder uses the
260-bit input frame to generate 456 bits on the output side. If several consecutive bits
are changed during the transmission over the air interface, the convolutional decoder
on the receiver side is not able to correctly reconstruct the original frame. This effect
is often observed as air interface disturbances usually affect several series bits.
In order to decrease this effect, the interleaver changes the bit order of a 456-bit data
frame in a specified pattern over eight bursts Figure (4).

The interleaving process can be illustrated as the following steps :-

 - Combine two channel coding outputs frames ( each contain 456 bits)
together to generate one frame that contain (912 bits) which represents 40
mSec from speech signal.
 - Divides the resultant frame to 114 segments each segment contain 8 bits .
 - Generate 8 segments from the above 114 segments by combining b0 from
each 114 segment to generate the anew segment (1), and combine b1 from
each 114 segments to generate the new segment (2) and so on.
4 – Ciphering :-
The next module of the transmission chain is the cipherer (Figure( 5)), which
encrypts the data frames it receives from the interleaver.

GSM uses, like most communication systems, a stream cipher algorithm. In order to
encrypt the data stream, a ciphering key (Kc) is calculated in the authentication
center and on the SIM card by using a random number (RAND) and the secret key
(Ki) as input parameters for the A8 algorithm. Together with the GSM frame number,
which is increased for every air interface frame, Kc is used as input parameter for the
A5 ciphering algorithm.
The A5 algorithm computes a 114-bit sequence which is XOR combined with the bits
of the original data stream. As the frame number is different for every burst, it is
ensured that the 114-bit ciphering sequence also changes for every burst which
further enhances security.
5 –Modulation :-
Before modulated the frames of the digital voiced ciphered frames they must
formatted into specified structure called traffic burst (Tch burst), in this stage each
(114) bits ciphered frame are divide to two part each contain (57 ) bits and assisted
with additional sections of bits as a tail, flags and training bits section, the burst is
encapsulated by two guard time in which no data is sent, each guard time with
duration equal (15.25) micro second lies in the start and the end of the burst data, the
overall duration of the burst in GSM is 577 micro second, the Tch burst structure are
shown in figure (6) below.

At the end of the transmission chain, the modulator maps the digital data onto
an analog carrier, which uses a bandwidth of 200 kHz. This mapping is done by
encoding the bits into changes of the carrier frequency. As the frequency change
takes a finite amount of time, a method called Gaussian minimum shift keying
(GMSK) is used, which smoothes the flanks created by the frequency changes.
GMSK has been selected for GSM as its modulation and demodulation properties are
easy to handle and implement into hardware and due to the fact that it interferes only
slightly with neighboring channels.
GMSK modulation is based on MSK, which is itself a form of continuous-phase
frequency-shift keying, CPFSK. One of the problems with standard forms of PSK is
that sidebands extend out from the carrier. To overcome this, MSK and its derivative
GMSK can be used.
There are two main ways in which GMSK modulation can be generated. The most
obvious way is to filter the modulating signal using a Gaussian filter and then apply
this to a frequency modulator where the modulation index is set to 0.5. This method
is very simple and straightforward but it has the drawback that the modulation index
must exactly equal 0.5. In practice this analogue method is not suitable because
component tolerances drift and cannot be set exactly.
A second method is more widely used. The quadrature modulator uses one signal that
is said to be in-phase and another that is in quadrature to this. In view of the in-phase
and quadrature elements this type of modulator is often said to be an I-Q modulator.
Using this type of modulator the modulation index can be maintained at exactly 0.5
without the need for any settings or adjustments. This makes it much easier to use,
and capable of providing the required level of performance without the need for
adjustments. For demodulation the technique can be used in reverse.