As we saw in the previous section, synchronous TDM does not guarantee that the full capacity of a link is used. In fact, it is more likely that only a portion of the time slots is in use at a given instant. Because the time slots are preassigned and fixed, whenever a connected device is not transmitting the corresponding slot is empty and that much of the path is wasted. For example, imagine that we have multiplexed the output of 20 identical computers onto a single line. Using synchronous TDM, the speed of that line must be at least 20 times the speed of each input line. But what if only l0 computers are in use at a time? Half of the capacity of the line is wasted.
Asynchronous time-division multiplexing, or statistical time-division multiplexing, is designed to avoid this type of waste. As with the term synchronous, the term asynchronous means something different in multiplexing than it means in other areas of data communications. Here it means flexible or not fixed.
Like synchronous TDM, asynchronous TDM allows a number of lower speed input lines to be multiplexed to a single higher speed line. Unlike synchronous TDM, however, in asynchronous TDM the total speed of the input lines can be greater than the capacity of the path. In a synchronous system, if we have n input lines, the frame contains a fixed number of at least n time slots. In an asynchronous system, if we have n input lines, the frame contains no more than m slots, with m less than n (see Figure 7.3-7). In this way, asynchronous TDM supports the same number of input lines as synchronous TDM with a lower capacity link. Or, given the same link, asynchronous TDM can support more devices than synchronous TDM.
The number of time slots in an asynchronous TDM frame (m) is based on a statistical analysis of the number of input lines that are likely to be transmitting at any given time. Rather than being preassigned, each slot is available to any of the attached input lines that has data to send. The multiplexer scans the input lines, accepts portions of data until a frame is filled, and then sends the frame across the link. If there are not enough data to all the slots in a frame, the frame is transmitted only partially filled; thus full link capacity may not be used l00 percent of the time. But the ability to allocate time slots dynamically, coupled with the lower ratio of time slots to input lines, greatly reduces the likelihood and degree of waste.
Figure 7.3-8 shows a system where five computers are sharing a data link using asynchronous TDM. In this example, the frame size is three slots. The figure shows how the multiplexer handles three levels of traffic. In the first case, only three of the five computers have data to send (the average scenario for this system, as indicated by the fact that a frame size of three slots was chosen). In the second case, four lines are sending data, one more than the number, of slots per frame. In the third case (statistically rare), all lines are sending data. In each case, the multiplexer scans the devices in order, from 1 to 5, filling time slots as it encounters data to be sent.
In the first case, the three active input lines correspond to the three slots in each frame. For the first four frames, the input is symmetrically distributed among all the communicating devices. By the fifth frame, however, devices 3 and 5 have completed their transmissions, but device 1 still has two characters to go. The multiplexer picks up the A from device 1, scans down the line without finding another transmission, and returns to device 1 to pick up the last A. There being no data to fill the final slot, the multiplexer then transmits the fifth frame with only two slots filled. In a synchronous TDM system, six frames of five time slots each would have been required to transmit all of the data---a total of 30 time slots. But only 14 of those slots would have been filled, leaving the line unused for more than half the elapsed time. With the asynchronous system shown here, only one frame is transmitted partially empty. During the rest of the transmission time, the entire capacity of the link is active.
In the second case, there is one more active input line than there are slots in each frame. This time, as the multiplexer scans from 1 to 5, it fills a frame before all he lines have been checked. The first frame, therefore, carries data from devices 1, 3, and 4, but not 5. The multiplexer continues its scan where it left off, putting the first portion of device 5's transmission into the first slot of the next frame, then moving back to the top of the line and putting the second portion of device 1's data into the second slot, and so on. As you can see, when the number of active senders does not equal the number of slots in a frame, the time slots are not filled symmetrically. Device 1 in this example, controls the first slot in the first frame, the second slot in the second frame, and so on.
In the third case, the frames are filled as above, but here all five input lines are active. In this example, device 1 controls the first slot in the first frame, the third slot in the second frame, and no slots at all in the third frame.
In cases 2 and 3, if the speed of the line is equal to three of the input lines, then the data to be transmitted will arrive faster than the multiplexer can put it on the link. In that case, a buffer is needed to store data until the multiplexer is ready for it.
Addressing and Overhead Cases 2 and 3 in the above example illustrate a major weakness of asynchronous TDM: how does the demultipiexer know which slot belongs to which output line? In synchronous TDM, the device to which the data in a time slot belong is indicated by the position of the time slot in the frame. But in asynchronous TDM, data from a given device might be in the first slot of one frame and in the third of the next. In the absence of fixed positional relationships, each time slot must carry an address telling the demultiplexer how to direct the data. This address, for local use only, is attached by the multiplexer and discarded by the demultiplexer once it has been read. In Figure 7.3-8, the address is specified by a digit.
Adding address bits to each time slot increases the overhead of an asynchronous system and somewhat limits its potential efficiency. To limit their impact, addresses usually consist of only a small number of bits and can be made even shorter by appending a full address only to the first portion of a transmission, with abbreviated versions to identify subsequent portions.
The need for addressing makes asynchronous TDM inefficient for bit or byte interleaving. Imagine bit interleaving with each bit carrying an address: one bit of data plus, say, three bits of address. All of a sudden it takes four bits to transport one bit of data. Even if the link is kept full, only a quarter of the capacity is used to transport data; the rest is overhead. For this reason, asynchronous TDM is efficient only when the size of the time slots is kept relatively large.
Variable-Length Time Slots Asynchronous TDM can accommodate traffic of varying data rates by varying the length of the time slots. Stations transmitting at a faster data rate can be given a longer slot. Managing variable-length fields requires that control bits be appended to the beginning of each time slot to indicate the length of the coming data portion. These extra bits also increase the overhead of the system and, again, are efficient only with larger time slots.
As its name implies, inverse multiplexing is the opposite of multiplexing. Inverse multiplexing takes the data stream from one high-speed line and breaks it into portions that can be sent across several lower speed lines simultaneously, with no loss in the collective data rate (see Figure 7.3-9).
b. Inverse multiplexing
Fig. 7.3-9 Multiplexing and inverse multiplexing b. Inverse multiplexing
Why do we need inverse multiplexing? Think of an organization that wants to send data, voice, and video, each of which requires a different data rate. To send voice, it may need a 64 kbit/s link. To send data, it may need a l28 kbit/s link. And to send video, it may need a l.544 Mbit/s link. To accommodate all of these needs, the organization has two options. It can lease a l.544 Mbit/s channel from a common carrier (the telephone company) and use the full capacity only sometimes, which is not an efficient use of the facility. Or it can lease several separate channels of lower data rates. Using an agreement called bandwidth on demand, the organization can use any of these channels whenever and however it needs them. Voice transmissions can be sent intact over any of the channels. Data or video signals can be broken up and sent over two or more lines. In other words, the data and video signals can be inversely multiplexed over multiple lines.
Figure 7.3-10 illustrates the application of inverse multiplexers, which offer the capability to interface to multiple 56 kbit/s interfaces.
DCE: data circuit-terminating equipment DTE: data terminal equipment Fig. 7.3-10 Illustration of inverse multiplexer operation
Was this article helpful?
Read how to maintain and repair any desktop and laptop computer. This Ebook has articles with photos and videos that show detailed step by step pc repair and maintenance procedures. There are many links to online videos that explain how you can build, maintain, speed up, clean, and repair your computer yourself. Put the money that you were going to pay the PC Tech in your own pocket.