2.2. Time Division Multiplexing (TDM)
TDM is a digital technology under which individual signals (i.e., a stream of binary digits or bits) are interleaved in time to produce the multiplexed, composite signal. Under TDM, recurring time-slots or channels are created such that each channel carries the bits from an individual signal. The total transmission bandwidth is split among the time-slots, with each component channel getting some fixed percentage of the total signal bandwidth. This total signal includes not only the payload bits for various component channels but also bits for performing overhead functions. TDM signals have a frame structure that repeats at regular intervals.
TDM has a long history in telecommunication networks. Prior to the advent of TDM, telephone calls were handled in the analog domain. Long distance calls were routed over twisted pair, coaxial cable, or analog microwave between switching offices. In early 1960s AT&T began installing DS-1 T-carrier services between long distance switching centers. Under this service, there were channel banks, which took 24 analog telephone lines, converted them to digital signals (called DS-0), time-division multiplexed them into a DS-1 signal, and then transmitted them over copper wire. At the other end, the DS-1 signal was de-multiplexed into the constituent DS-0 signals, which were then converted back into analog. T-carrier reduced the number of copper circuits required between switching centers and improved the quality of the telephone calls by reducing noise and cross talk. As the traffic volume grew, the number of T-carrier circuits between switching centers increased. In order to cope with the demand, higher speed T-carriers were introduced. In the late 1970s, optical communications became feasible, enabling high-speed and high-fidelity digital communication. One of the first commercial fiber circuits was installed in Chicago in 1977 and operated at 45 Mbps (DS-3 rate).
The frame structures of the DS-1 [ANSI95b] and the European E1 [ITU-T98a] signals are shown in Figure 2-1. The DS-1 signal consists of 24 payload channels plus overhead. The basic frame of each of these signals repeats every 125 µs, that is, 8000 times per second. With 8 bits carried in each channel, this gives rise to a basic data rate of 64 Kbps for each channel. The requirement for this data rate stems from the need to sample the analog telephony signal 8000 times per second and encoding each sample in 8 bits.
Figure 2-1. Basic Frame Structures for the DS-1 and E1 TDM Signals
A DS-1 frame contains 24 channels, each consisting of 8 bits, plus 1 framing/overhead bit, leading to a total of 193 bits. Since the frame repeats every 125 µs (or 8000 times a second), the total bit rate of the DS-1 signal is 1.544 Mbps. Similarly, the total bit rate of the E1 signal is 2.048 Mbps (32 channels of 8 bits, repeating every 125 µs).
In a TDM frame, each channel within the frame gets a fixed amount of bandwidth that is not shared with any other channel. It is, however, necessary to determine the start of a TDM frame to de-multiplex the channels within.
With DS-1 signals, the start of frame can be identified by a particular pattern of framing/overhead bits occurring in a consecutive series of frames. A DS-1 superframe consists of 12 regular frames as shown in Figure 2-2 [ANSI95b]. The twelve-bit pattern, 100011011100, occurring in the superframe is used to delimit DS-1 frames. In addition to enabling this function, the superframe structure allows the subdivision of the bits in a channel into lower rate channels.
Figure 2-2. The DS-1 Superframe Structure
In DS-1 networking, a technique called “robbed bit” signaling is used to convey signaling for each of the 24 channels within a DS-1 signal. Under this technique, the least significant bit of each channel (i.e., bit 8) in frames 6 and 12 of the superframe are used to convey signaling for that channel. The bit rate of this signaling channel can be computed as follows. A single bit out of a channel corresponds to a bit rate of 8000 bps (since the frame containing the channel occurs 8000 times per second). Because the signaling bit is present in only two out of twelve frames, this rate is reduced by 1/6, that is, 8000/6 = 1333.3 bps. Note that robbed bit signaling prevents the use of this bit for carrying user data. Hence, only 7 of the 8 bits are usually used for data services, leading to a 56 Kbps service over each channel.
While the DS-1 superframe allows signaling for the payload channels, there is no provision for conveying information between the end points to aid in managing the signal. This was addressed with the DS-1 extended superframe (ESF) format [ANSI95b]. The ESF is 24 regular frames long. Of the 24 overhead bits, 6 are used for frame alignment, another 6 are used for error monitoring, and the remaining 12 are used to form a 4 Kbps data link between the sender and receiver.
With the E1 signal, the first octet (byte) in the frame contains overhead bits. A multiframe consisting of 16 frames is used with the base E1 signal.