21.2 Digital time division multiplex structure
21.2.1 Frame organisation
For each digital channel there is assigned a group of b bits, called a word which corresponds to a digital message element (data). The groups belonging to the same channel are transmitted at a frequency fs, equal to the sampling frequency. Thus, the bit rate of channel i can be calculated as in Equation 21.2.
In the case of digital PCM telephony, these words are composed of 8 bits and called octets. Hence the bit rate of each channel equals 64000bit/s.
When n digital channels are assembled into a time division multiplex, the collection of the n words of b bits (and eventually auxiliary bits which are added to them), within a period given by Equation 21.3, constitutes a frame.
Although the structure of the frame is strictly repetitive, their content obviously is not because the channels contain variable digital information.
Two types of frame organisation can be considered, as shown in Figure 21.3:
Word interleaved, where the interleaving is performed on a character by character basis, i.e. a character from each data source is accumulated and combined as a word or frame for transmission. To do so, the frame is subdivided into у ≥ z time slots each containing b grouped bits corresponding to the same digital channel or auxiliary bits.
Bit interleaved, where the interleaving is performed on a bit by bit basis, i.e. one bit from each data source is accumulated and combined as a word or frame for transmission. To do so, the frame is subdivided into b groups each containing z bits of the same order belonging to each of the channels.
The word interleaved structure corresponds well to the mode of functioning of a PCM modulator, while the bit interleaved structure has advantages for time division digital switching because it suffers less throughput delay.
21.2.2 Frame alignment
The frame alignment or framing is typical of time division multiplexing. It consists of synchronising the receiving equipment both in frequency and in phase to the stream of symbols it receives. This operation is obviously necessary each time the receiver is switched on, but also during normal operation. Indeed, once aligned, the receiver needs a periodic time reference in order to check its isochronism and detects eventual shifts.
This necessary time reference consists of a particular pattern of several bits, called the framing pattern. When the receiver has lost frame alignment, it searches for this pattern in order to realign itself with as short a delay as possible. The framing pattern is periodically carried by the frame according to one of the following organisations:
Grouped framing pattern, which consists of a number of consecutive bits at the beginning of a frame.
Distributed framing pattern which, as the name suggests, is spread over the frame on a bit by it basis, or over several frames at one bit per frame.
There is of course a danger of simulating the framing pattern by chance combination of other information carrying bits. There are different ways to protect the system from this, such as:
One can choose a framing pattern with low autocorrelation, so that it is impossible to imitate by shifting and infringement on random neighbouring bits. Examples are the patterns 110 or 1110010 for for 3-bits or 7-bits grouped framing patterns respectively.
One can block all the channels of the frame when the framing is lost at reception. The frame would be replaced by the trans mission of a resynchronisation signal. This necessitates an announcement warning in the opposite direction, which is assumed to be correctly aligned.
One could confirm correct framing by a criterion other than the presence of the framing pattern, e.g. absence of the pattern in one frame out of two.
In order to avoid reacting to each transmission error, the reaction to an incorrect reception of the framing pattern in an aligned situation, i.e. during normal data transfer, must be delayed. The solution to the framing problems of both Tl and El are given in the respective sections later.