Mutual clock synchronization in global digital communication networks

Abstract

AbstractThere are numerous synchronization methodologies available for consideration in solving the problem of network synchronization in digital communication networks [1, 2]. They include: independent clocks, pulse stuffing, elastic stores, master‐slave hierarchical, time reference distribution, mutual synchronization, etc. This paper considers the network synchronization performance achievable using the method of mutual clock synchronization observed first by Christiaan Huygens in 1665 [3]. Network synchronization performance metrics studied include: nodal timing accuracy, timing jitter, slip rate, time interval between slips, probability of loss of synchronization. These metrics are shown to depend upon the distance (range) between clocks, ranging error, clock stability, nodal phase error processing bandwidth, data rate, signal‐to‐noise ratio and network connectivity. In this regard, the mutual synchronization performance achievable with long wavelength biological rhythms and electric power system rhythms is compared with the performance achievable using short wavelength rhythms required in wideband and broadband digital communications networks. The results are further applied to the problem of synchronizing a satellite communications network. When intrasatellite communication crosslinks (links between satellites in the same orbital plane) are used in a constellation of communication satellites, it is shown that the maximum data rate, the network connectivity and the constellation altitude drive the achievable network synchronization performance; the latter is set by technological limitations due to clock frequency stability, the maximum range between satellites and the minimum ranging error achievable by the ranging system. In this sense, low Earth orbits (LEO) are preferred over geosynchronous (GEO)Earth orbits. The theory is also applied to the Teledesic and Iridium networks.