B.S.T.J. Briefs: A Self-Reorganizing Synchronization Network

01 September 1963

New Image

This paper describes a method of increasing the reliability of synchronization in a network of remote clocks. A common synchronization technique embodies the master-slave relation where one particular clock is a reference for the others. The geometry of a typical network can be likened to a "tree" structure. The master clock transmits its signal, simultaneously, over several transmission links to synchronize slave oscillators at the ends. The slaves retransmit the reference signal to other slaves one link away. This process is iterated as the system expands. Each slave has only one input; hence, a network of this type can be disabled, in varying degree, by the failure of any clock or transmission link. The probability of failure can be reduced by creating redundant paths between nodes in the network. These paths can consist of more than one link and include the intervening clocks. This allows bypassing of a disabled clock or reorganization of synchronization authority. To implement such a system we need interrogation equipment at each clock station to decide which incoming path has the highest priority for a given situation. If the reorganization takes place automatically, we can refer to it as a "self-reorganizing synchronization network." Application of this scheme in a large "tree" network can become very complex. A technique which provides orderly organization of complex networks was patented by CI. P. Darwin and R. C. Prim (Patent No. 2,986,723). Their system, in its most general form, requires a three-part "signature" to be transmitted, along with the synchronizing signal, from each local clock.