QUESTION:
In the event of a GPS shut down by the US government, what would be the effects on a communications network that relies on GPS for its Primary Reference?
ANSWER:
When used in a Primary Reference Source, the GPS receiver is always used in conjunction with other circuitry that includes a high-performance oscillator. The GPS receiver timing output (typically a 1-pulse-per-second tick) is considered as a "raw" input to phase/frequency locked loops that utilize the stability of the oscillator to implement long time constants and thus filter out the short-term instability (i.e. noise) of the (raw) GPS receiver output. In this sense the GPS receiver is but a "component" (albeit a very, very, sophisticated component) in a timing device (primary reference source). The cost of the oscillator is usually significantly more than the cost of the GPS receiver (viewed as a component).
The use of a rubidium secondary atomic standard as the local oscillator is especially beneficial; the superior stability of the rubidium oscillator permits deployment of antennas that do not have an unobstructed view of the sky, implying that there are intervals when an insufficient number, maybe none, of satellites are visible. The long filtering time-constant made possible by having a rubidium oscillator allow the primary reference source to operate within the stratum-1 specification even if selective-availability is not turned off (however, it will not be turned on, see below). However, even though the cost of the Primary Reference Source is an insignificant fraction of the cost of all the equipment that rely on it for timing purposes, there is a tendency for some operators to opt for a quartz-based oscillator solution rather than a rubidium-based solution to eke out some meager capital expenditure saving.
Furthermore, the PRS output is usually fed as a reference input to a BITS (Building Integrated Timing Supply) system (also called, variously, TSG, for timing signal generator; SSU for synchronization supply unit; and SASE, for stand alone synchronization equipment). One of the key characteristics of the BITS is the provision of holdover capability. The intention here is to maintain a good output even when the input reference, coming from the GPS PRS in this example, is not available, for any reason. The PRS output may become invalid for various reasons, ranging from the mundane when a craftsperson accidentally disconnects a cable or an act of God such as the GPS antenna being taken out by a lightning strike. The holdover capability is very important to flywheel through these times of trouble. If the holdover oscillator is quartz-based, such as a "Type 1", you can expect a typical drift-rate of about 1 part in 10E10 for the first 24 hours, with some improvement in the drift rate as time goes on. If the holdover oscillator is rubidium-based, the typical drift rate is about 1 part in 10E11 per week.
So the impact of a GPS outage on the wired telecommunication network will be, initially.... nothing! If the holdover oscillators are rubidium-based, the BITS output will be close to stratum-1 accuracy for weeks; with quartz-based holdover, the BITS output will be close to stratum-1 for about a day. The impact on voice services will be an increasing rate of "clicks and pops" related to slips, and human beings what they are, may be willing to tolerate these quality impairments. Modems (such as fax machines) are less forgiving. Data applications may "slow down" since errors resulting from slips may force packet retransmissions. From an end-user perspective, the quality of experience will deteriorate slowly, as opposed to catastrophically, and may not even be noticeable for days or weeks, depending on the network design philosophy on holdover.
NIST, and, no doubt, the DoD, continuously monitors the performance of each of the satellites and publishes the results. However, these results are known after the fact and are of little value for instantaneous corrections in a GPS receiver. The main intent of these performance measurements is to see if the satellite has "gone rogue", or is showing signs of malfunctioning, for any reason. GPS receivers cannot use this information, considering that it is not available in real-time. Most GPS receivers, and definitely those used in telecommunications timing applications, have a RAIM (Receiver Autonomous Integrity Monitoring) feature that allows the receiver to do a sanity check on the signals from a particular satellite and decide whether to incorporate the information from that satellite into the space-time solution. While the frequency errors reported by NIST are not useful for correcting output of a primary reference source, these reports are valuable since they provide a "paper-trail" so that the timing output of the primary reference source can be deemed "traceable" to UTC. The following URL pointing to the NIST website has some useful information:
http://www.boulder.nist.gov/timefreq/service/gpscal.htm).
You can rest assured that the monitoring of satellites by the DoD will ensure that faltering satellites will be either taken out of service or "repaired" and the NIST data will confirm this. In fact, it is the stated policy of the U.S. government to maintain, and continually make improvements to, the GPS system as a whole. The policy also states that the action of the Clinton administration to turn off selective-availability will not be reversed. A policy statement issued on Dec. 15, 2004, is available at the following government website:
http://www.navcen.uscg.gov/cgsic/geninfo/FactSheet.pdf.
One of the highlights of the policy is the recognition of the vital importance of timing services to the telecommunications industry, among other civilian usage. The policy is clear in stating that it is the government's intention to continuously provide civilian services free of cost and to promote the use of satellite systems as well as to interoperate and be compatible with other satellite-based systems that may emerge such as the Galileo system under development in Europe. Whereas the Government reserves the right to disable GPS in the event of a national emergency, this action will clearly be taken only after great deliberation and every effort will be made to keep the outage of as short a duration as possible.
However, it is prudent to consider the possibility of such an occurrence and take reasonable steps to mitigate the impact of GPS outages. What will be learnt is that there are numerous reasons for incurring GPS outages, on a localized if not worldwide base, that happen even when the GPS system is operational. It will be found that the increased capex associated with deploying good holdover is more than offset in reduced opex, and that end-user quality of experience is negligibly impacted by these (hopefully infrequent) GPS outages.
|
