Chapter 2. History of Timing in the Switched Network

Telephone companies and long distance carriers started using digital technology in the 1960's to improve service and lower the costs associated with the transmission of analog services. The first digital system installed was known as the T1 transmission system, and it was used primarily for pair gain, or the reduction in the number of pairs used to carry voice traffic. The T1 trunk operating at 1.544 Mb/S carried 24 64 kb/S channels, each 64 kb/S channel carrying 8000 8-bit bytes per second. Each byte represented one sample of analog information. The remaining 8000 bits per second were framing bits (24 x 64 = 1536 kb/S plus 8 kb/S = 1.544 Mb/S). The introduction of digital toll switches, with stored program control, allowed for even lower cost and more features to be provided in the analog voice network. Later, with the advent of local digital switches, digital transmission was extended to the local office.

Private companies usually have constructed telephone systems to obtain more efficient communications between locations. Originally, these networks were small copies of the analog transmission and switching systems used by telephone companies. The need for fast and effective data communications arose with the advent of the use of computers as a corporate tool. The typical data network was constructed using analog modems in a separate point-to-point network. As depicted in Figure 1, this scheme often led to two separate networks, one for switched voice, and one for data transmission.

Unfortunately, the voice quality of these analog private networks proved to be less than desirable due to the low bandwidth and losses of the analog channels forming the transmission path and the switching systems. Data transmission over the analog network suffered from error bursts, which resulted in lower speeds. The new digital system provided better voice quality, fewer wires between offices, higher reliability, and improved data transmission.

Initially, T1 was an asynchronous system. Each pair of end terminals ran at their own clock rate, and each terminal used its receive timing to demultiplex the incoming signal. The transmit and receive sides were independent of one another. Later, when digital channel units were used, one end terminal was designated as the master and had its own timing reference (Figure 2). The other end terminal was a slave, and derived timing for its transmit side from the data being received (looped timing). This arrangement worked as long as the end terminals were no more complex than a channel bank.

The growth of T1 (DS1) networks to include higher multiplexing rates and long distance DS1 connectivity introduced various synchronization problems. Digital switches with DS1 port interfaces exemplified the shortcomings of an asynchronous system. If the two switch clocks were not at the same frequency, the data would slip at a rate dependent on the difference in clock frequencies. A slip is defined as a one frame (193 bits) shift in time difference between the two signals in question. This time difference is equal to 125 microseconds.

Slips were not considered a major impairment to trunks carrying voice circuits. The lost frames and temporary loss of frame synchronization resulted in occasional pops and clicks being heard during the call in progress. However, with advances in DS1 connectivity, these impairments tended to spread throughout the network. To minimize these, a hierarchical clock scheme was developed, whose function was to produce a primary reference for distribution to switching centers in order to time the toll switches [Reference 1].

Local switching in that era was primarily analog, so that synchronization was not required at the end offices. Later, digital switches and direct digital services or networks (DDS or DDN) became common at the end offices, providing digital services to customers. This meant timing had to be distributed to local levels.

As shown in Figure 3, the resulting hierarchy evolved into four levels. Level 1, known as Stratum 1, is the primary reference. lt was known originally as the Bell System Reference Frequency or BSRF. The second level, Stratum 2, is used at toll switches. Stratum 3 is used at local switches. Channel banks and end terminals that use simple crystal oscillators are known as Stratum 4 devices. Recently, SONET (Synchronous Optical Network) networks have created the need for a clock stratum level better than Stratum 3, which is called Stratum 3E.

When the Bell System broke up into the local service providers and the long distance carriers, the timing hierarchy became less well defined. Now each local company could no longer take its timing from the long distance carrier, but had to engineer a system, either a hierarchy or otherwise, to distribute timing to their offices. This made everything more difficult because failures in the transmission systems may cause "islands" or areas without a reference to Stratum 1. In such a case, the "island" is in holdover at whatever stratum level has been provided. Even if a particular network is still traceable to Stratum 1, the traffic of concern may be coming from such an "island" and will therefore slip at some rate, as shown in Figure 4. Having a Stratum 1 source is, in itself, no guarantee of a slip free network.