B. P. Lathi, Zhi Ding - Modern Digital and Analog Communication Systems-Oxford University Press (2009)

6.4 Digital Multiplexing 287
Figure 6.23
DMl /2
multiplexer
format.
Mo [48] CA [48] Fo [48] CA [48] CA [48] F1 [48]
M1 [48] CB [48] Fo [48] CB [48] CB [48] F1 [48]
M1 [48] C c [48] Fo [48] C c [48] C c [48] F1 [48]
M1 [48] Co [48] Fo [48] Co [48] Co [48] F1 [48]
wrong sequence. The presence ofMoM1 M1 M1 provides verification of the genuine FoF1 FoF1
sequence. The C bits are used to transmit additional information about bit stuffing, as discussed
later.
In the majority of cases, not all incoming channels are active all the time: some transmit
data, and some are idle. This means the system is underutilized. We can, therefore, accept more
input channels to take advantage of the inactivity, at any given time, of at least one channel.
This obviously involves much more complicated switching operations, and also rather careful
system planning. In any random traffic situation we cannot guarantee that the number of
transmission channels demanded will not exceed the number available; but by taking account
of the statistics of the signal sources, it is possible to ensure an acceptably low probability of
this occurring. Multiplex structures of this type have been developed for satellite systems and
are known as time division multiple-access (TOMA) systems.
In TDMA systems employed for telephony, the design parameters are chosen so that any
overload condition lasts only a fraction of a second, which leads to acceptable performance
for speech communication. For other types of data and telegraphy, transmission delays are
unimportant. Hence, in overload condition, the incoming data can be stored and transmitted
later.
6.4.2 Asynchronous Channels and Bit Stuffing
In the preceding discussion, we assumed synchronization between all the incoming channels
and the multiplexer. This is difficult even when all the channels are nominally at the same
rate. For example, consider a 1000 km coaxial cable carrying 2 x 10 8 pulses per second.
Assuming the nominal propagation speed in the cable to be 2 x 10 8 mis, it takes 1/200 second
of transit time and 1 million pulses will be in transit. If the cable temperature increases by 1 °F,
the propagation velocity will increase by about 0.01 %. This will cause the pulses in transit
to arrive sooner, thus producing a temporary increase in the rate of pulses received. Because
the extra pulses cannot be accommodated in the multiplexer, they must be temporarily stored
at the receiver. If the cable temperature drops, the rate of received pulses will drop, and the
multiplexer will have vacant slots with no data. These slots need to be stuffed with dummy
digits (pulse stuffing).
DSl signals in the North American network are often generated by crystal oscillators
in individual channel banks or other digital terminal equipment. Although the oscillators are
quite stable, they will not oscillate at exactly the same frequency, leading to another cause of
asynchronicity in the network.
This shows that even in synchronously multiplexed systems, the data are rarely received
at a synchronous rate. We always need a storage (known as an elastic store) and pulse stuffing
(also known as justification) to accommodate such an situation. Obviously, this method of an
elastic store and pulse stuffing will work even when the channels are asynchronous.
Three variants of the pulse stuffing scheme exist: (1) positive pulse stuffing, (2) negative
pulse stuffing, and (3) positive/negative pulse stuffing. In positive pulse stuffing, the multiplexer