Multiplexing Technique

• Types of TDM
  
  • There are two types of TDM –

(a) Synchronous TDM (b) Asynchronous/Statistical TDM

(a) Synchronous TDM:

Definition

• • In Synchronous TDM, each device has given same time slot/time slice to transmit the data over the single high speed link, irrespective of ignoring other factors such as the device has any data to transmit or not. Hence the name Synchronous TDM.

• Features 
  • Here, each source device places its data onto the link when its time slot arrives.
  • Each device has given the time to access the link  turn by turn and they can use the link only their time slot has come.
  • If any device does not have data to send then its time slot remains empty.
  • In this, the various time slots are organized/grouped into frames and each frame consists of one or more time slots dedicated to each sending device. In another words, if there are n sending /source devices, there will be n slots in frame, if created i.e. one slot for each device.
  • Multiplexing is concerned with physical layer of the OSI reference model, not with data link layer.
• Advantages :

• • It is a simpler method of TDM to implement because the process of multiplexing and demultiplexing is easier.
  • In Synchronous TDM, the time slots are reserved for each transmission, irrespective of whether it has any data to transmit or not.
  • Here, we mainly transmit digital signals, although the actual transmission may be digital or analog. In the latter case, the composite signal has to be converted into analog data by passing it through a modem.
  • It supports data sources with differing data rates to transmit. This could be done by assigning fewer slots to slower sources and more slots to the faster ones.
  • Here, a complete cycle of time slots is called a frame. At the receiving end, the frame is decomposed into the constituent data streams by sending the data in each time slot to the appropriate buffer.
  • The sequence of time slots allotted to a single data source makes up a transmission channel.

• Disadvantages :
  • Synchronous TDM strictly follow the principle that the total speed of various input lines should not exceed but less or at least equal to the capacity of single high speed link/path otherwise transmission would be aborted.
  • This method can be inefficient because many time slots may have only silence/no data to transmit during their turn but that time slot is not used by other and finally wasted. 
  • As we know that in synchronous TDM data transmission can be quite wasteful. For example, as in a voice transmission, much of the bandwidth can be wasted because there are typically many periods of silence in a human conversation. Another example, could be, that of a time sharing system where many terminals are connected to a computer through the network. Here too, many of the time slots will not be used because there is no activity at the terminal. In spite of being comparatively simpler, synchronous TDM is therefore, often not an attractive option. 
  • We have already seen that the transmission in TDM must be synchronous because otherwise, all the data would be garbled and lost. [ To achieve synchronisation, at the physical layer, we have to ensure that each frame is synchronised to maintain the integrity of the transmission. One scheme to attain this is to add an extra bit of data to each frame. In succeeding frames, this extra bit forms a fixed pattern that is highly unlikely to occur naturally in the data. The receiver can then use this fact to synchronise the frames. Suppose each frame has n bits. Starting from anywhere, the receiver compares bit number 1, n+1, 2n+1 and so on. If the fixed data pattern is found, the frames are synchronised. Otherwise, the receiver commences to check bit number 2, n+2, 2n+2, … until it can determine whether the frames are in synchronization or otherwise. This continues till, it is able to synchronise the frames and can then start receiving the data from the various channels.
    Even after this, it is important to continue to monitor the fixed pattern to make sure that the synchronisation remains intact. If it is lost, the routine of re-establishing it needs to be repeated as before.
    The second problem in synchronising the transmission is the fact that there can be some variations between the different clock pulses from the different input streams.
    To take care of this, we can use the technique of pulse stuffing. Here, the multiplexed signal is of a rate that is a bit higher than that of the sum of the individual inputs. The input data is stuffed with extra pulses as appropriate by the multiplexer at fixed locations in the frame. This is needed so that the demultiplexer at the other, receiving, end can identify and remove these extra pulses. All the input data streams are thus synchronised with the single local, multiplexer clock. ] 

(b) Asynchronous/Statistical TDM:

• • Statistical TDM is normally used in place of synchronous TDM to overcome the disadvantage of it.
  • In this technique, time slots are not fixed as synchronous TDM rather here time slots are allocated dynamically to only those devices which have the data to send i.e. active workstations.
  • Here, total speed of the input lines can be greater than the capacity of the channel.
  • This multiplexing accepts the incoming data streams and creates a frame that contains only data with no empty slots.
  • In Asynchronous TDM, each slot contains an address part that identifies the source of the data.
  • This is a method where there are more devices than the number of time slots available. Here, each input data stream has a buffer associated with it. The multiplexer looks at the buffer of each device that provides the input data and checks to see if it has enough to fill a frame. If, there are enough, the multiplexer sends the frame. Otherwise, it goes to the next device in line and checks its input buffer. This cycle is continued indefinitely.

At the receiving end the demultiplexer decomposes the signal into the different data streams and sends them to the corresponding output line.

• • The multiplexer data rate is not higher than that of the sum of all the input devices connected to it. So statistical, TDM can make do with a lower transmission rate than needed by synchronous TDM, at the cost of more complexity.
  • They support higher throughput for the same available data rate of the multiplexer. This capability is easily realised during the normal, expected data transmission periods when the amount of data that the different input devices have available for transmission is, in fact, lower than the capacity of the multiplexing device. But the same attribute becomes a disadvantage during peak loads, where all or many of the devices may have data to transmit for a short time. In such a situation, the slack that was available to us for more efficient transmission is no longer present.
  • In this kind of scheme it is not known to us which input device will have data to send at a given time. So we cannot use a round robin positional scheme that was possible in synchronous TDM. Each frame has to be accompanied by information that will tell the demultiplexer which input device the frame belongs to, so that it can be delivered to the appropriate destination. This is the overhead of statistical TDM, besides increased complexity of equipment.
  • When we are sending input data from only one source at a time, the structure of a frame in it would need to have an address field for the source followed by the data for it. This is really the statistical TDM data frame. It would form the data part of a larger,
    enclosing HDLC frame if we are using HDLC as the transmission protocol. This frame would itself have various other fields. Such a scheme is also not as efficient as we can make it. This is because the quantum of data available from the source, in that time slot, may not be enough to fill the TDM sub frame. So, while it may be an adequate method if the load is not heavy, we also need to think of a method that can utilise available resources better.
  • It transfers more than one input data source transmit in a single TDM frame.
  • We would then need to have a more complex structure for the frame whereby, we would have to specify the different input devices in the frame followed by the length of the data field.
  • More sophisticated approaches could be used, in order to optimise the number of bits, we need to encode all this addressing and data information. For peak loads, there is need for some kind of buffering mechanism, so that, whenever there is excess input from the data sources that the multiplexer cannot immediately handle, it is stored until it can be sent. The size of the buffer needed will increase as the transmission capacity of the multiplexer decreases, as it will become more likely that an input data stream will not be transmitted immediately.
  • It will also depend on the average data rate of the input devices taken together. No matter what the buffer size we choose, there is always a non-zero probability that the buffer itself will overflow, leading to loss of data. If, the average data rate of all devices is close to the peak rate, then we are approaching a situation of synchronous TDM where we are not able to take advantage of periods of silence that is the basis of statistical TDM.