QUESTION:
What development work has been done on Time Of Day? What is the current standard and are future standards planned?
ANSWER:
The topic is a vast one. Generally speaking, "Time of Day", or "absolute time", is in the realm of organizations like NIST (National Institute of Science and Technology, located in Boulder Colorado) that maintain the time standard for a particular country. The notion of "UTC" (Universal Coordinated Time) is the worldwide standard, maintained by coordinating the efforts of various organizations like NIST in several countries.
From a communications perspective, "Time of Day" refers to the result of transferring time from a device (or "server") that has an absolute time reference that is traceable to UTC to another device (or "client") for a variety of purposes, primarily time-stamping events of various types.
So far there is one well-accepted and described protocol for transferring time over a packet network (such as the Internet). This is NTP (Network Time Protocol). It is an official Internet protocol and described in RFC-1305. It is a simple protocol that involves an exchange of packets between the server and client. The packets contain time-stamps of origination and reception (of the packets) and thus the client can ascertain the clock offset between itself and the server. The transfer mechanism has some issues, namely the variability of time delay experienced by packets over a routed infrastructure (the Internet) and the non-guarantee of symmetry in delay in the two directions. Nevertheless, accuracy of time transfer of the order of 10 msec (standard deviation) can be achieved.
For packet networks embodied in a controlled environment such as a LAN, another protocol, IEEE-1588, referred to as PTP (Precision Time-transfer Protocol), has emerged. The intent of this protocol is to ensure that all devices on the LAN (or sub-net) have "equivalent" clocks. The protocol allows for one device to become a "master" for the sub-net and all other devices slave to this master. If one device has access to UTC-traceable time, it can serve as the "grandmaster". "Boundary" clocks are defined that allow for time-transfer to circumvent (actually go around) devices such as routers that may impact the quality of the time-transfer. IEEE-1588 can achieve an accuracy of time-transfer of the order of 1 microsecond (standard deviation).
If the medium between "master" and "slave" (i.e. server and client) is point-to-point and not shared, two-way time-transfer protocols operating in a ping-pong mode (the same pair of wires is used for both directions of transmission, minimizing asymmetry in propagation time) can achieve accuracies of time-transfer of the order of a few nanoseconds (standard deviation).
Standards work, particularly in the case of the third method, is in its infancy.
Needless to say, there are other means of establishing time of day. A GPS receiver based system with a high quality oscillator can establish time of day, with respect to the GPS time, with accuracy of the order of 50 nanoseconds. This is the method by which most "Stratum-1 NTP Servers" achieve their time reference ("Stratum-1", in NTP terminology, implies that the server has an external source for time, not necessarily accurate!)
|
