Expert Advice: Managing the GPS Constellation for Today’s Needs

John Lavrakas

In a recent editorial in GPS World’s Survey & Construction e-newsletter entitled “No Joy in Surveyville,” Eric Gakstatter lamented the performance of the GPS constellation for surveying. He is not alone. In June, the Australian Broadcasting Company reported that farmers in Australia were experiencing major problems with GPS because two satellites had been removed from service.

For many, GPS is at its best performance ever, with 29 satellites in orbit and user range errors at their lowest levels in years. Yet for others, GPS performance falls short of expectations. What is the real issue here? Is it the number of usable satellites in the constellation — or have the demands of the user community grown?

Today’s Performance

Let’s first take a look at the performance relative to the current constellation. The GPS Standard Positioning Service (SPS) Performance Standard identifies 24 nominal orbital slots for a 24-satellite constellation. In this article, I refer to these as the 24 primary slots. My source material for the slot allocations is the U.S. Coast Guard operational advisories.

Examining GPS performance over the past three years with respect to satellites in the key orbital slots, we see some interesting trends. Figure 1 presents the average number of satellites on orbit as well as the average number of healthy satellites in the 24 primary slots. A healthy satellite is one that has not been removed from service either due a scheduled outage (satellite and clock maintenance) or from an unscheduled anomaly (for example, degraded clock operation or problems with the spacecraft bus).

Figure 1. Average number of healthy satellites on orbit (blue) and average number of healthy satellites in the 24 primary slots (red).

The number of usable (healthy) satellites grew from 26 to about 28 on average, but this has not changed substantially in the past three years. It has varied between 27 and 29 satellites, with no significant upward or downward trend over this period. The number of satellites in primary slots, however, does show a noticeable trend, growing steadily through 2003 until late 2004 when it leveled off, after which it began to decrease. This trend recurs in Figure 2, where we view the same metric in half-year increments.

Figure 2. Average number of healthy satellites in primary slots, shown in half-year increments.

The reduced number of filled primary slots stems from unscheduled outages. Scheduled outages have no significant impact on number of satellites usable since the operators typically remove a satellite from service for only a few hours, and such maintenance is performed on the order of once a month per satellite. Unscheduled outages, however, can last days and may require significant effort on the part of the satellite operators to resolve.

The SPS Performance Standard states that 24 operational satellites must be available on orbit with 0.95 probability (averaged over any day). We see this figure has been met at the 100 percent level over the past three and a half years.

The SPS Performance Standard further states that at least 21 satellites in the 24 nominal plane/slot positions must be set healthy and transmitting a navigation signal with 0.98 probability (yearly average). This figure is met.

Figure 3 presents the monthly availability of the primary 24-satellite constellation (blue plot), that is, the percent of time over a month that there is a full set of 24 usable satellites in their primary slots. Here we see a marked trend, showing a steady growth from the beginning of 2003 up to the end of 2004, followed by a reduction, but still above the 95 percent level. Figure 3 also shows the monthly availability of 21 or more satellites assigned to the 24 primary slots (red plot), which has been at 100 percent over the past three and a half years.

Figure 3. Monthly availability of the primary 24-satellite constellation (bue) and of 21 or more satellites assigned to the 24 primary slots (red).

So What’s the Problem?

If the U.S. government is meeting its commitments, why do users see degraded performance?

Part of the issue is that the government manages the constellation to a set of metrics that is not up with the times, so to speak. The SPS Performance Standard has a legacy dating from prior to May 2000 when the government imposed Selective Availability, the intentional degradation of the positioning and timing accuracy for civilian users.

Surveyors back then were considered eccentrics, as it were, living off the crumbs that fell from the table of the basic service. They took advantage of carrier phase tracking, but were content to post-process the data. Work that took days and weeks prior to GPS could now be done in hours. Well, those days are gone, and the push is now to get work done in minutes.

The familiar adage “Give GPS users a yard and they’ll want an inch” has a corollary: “Give GPS users a process that takes them hours and they’ll want it done in minutes — or seconds!” Users have found they can do their processing much faster, as long as the constellation performance is well above the levels set in the SPS Performance Standard. This has indeed been the case since 2000.

The GPS program has placed into orbit more satellites than originally anticipated. The general thought was that 27 satellites were sufficient to support the 24 satellite constellation. With 28, 29, and even 30 satellites in orbit, GPS has exceeded expectations, yet now the new expectations are that the government will sustain this level of performance.

Improvements Happen

The U.S. Air Force has made significant improvements in GPS operations as well in recent years. The satellite operators have become more user-focused. Prior to taking a satellite offline for maintenance, operators examine the effect of its removal to users worldwide. Also, they have adjusted operational procedures such that anomalies that once took hours to correct are now resolved within minutes. These improvements have directly benefited users, yet despite this, the mindset of the GPS operators is still to provide the service identified in the SPS Performance Standard, which is not the same as day-to-day service that users have come to expect.

The Presidential Policy on National Space-Based Position, Navigation and Timing (PNT), issued in December 2004, provides high-level guidance on what service users can expect. Among its goals for space-based PNT, the policy states:

• provide uninterrupted availability of positioning, navigation, and timing services
• meet growing national, homeland, economic security, and civil requirements, and scientific and commercial demands.

What “availability” is assumed in the first goal? Is it availability for users employing the 5-degree mask angle (as defined in the SPS Performance Standard), or is it the more stringent demand of mask angles at 10 degrees or higher? Is it availability of four satellites in view to support the generic user or that of six satellites in view to support receiver autonomous integrity monitoring with fault detection and exclusion?

What “demands” mentioned in the second goal are to be met? Are they the demands of the precision farmer and surveying community? Or perhaps the tighter requirements of the urban user?

This policy also states that the government will improve the performance of space-based positioning, navigation, and timing services. This implies that the service identified in the SPS Performance Standard will need to be changed to accommodate these improvements.

To some extent, the U.S. government can only go so far in meeting user needs with the current system. The maximum number of satellites today’s operational control system can support is 30. As of the time of this article, GPS had 29 operational satellites on orbit, although at any given time not all are usable, due to necessary maintenance or unscheduled downtime. Whenever satellites are set unusable, the satellite operators look at the resulting performance, comparing it to the SPS Performance Standard. This is where the issue lies. The SPS Performance Standard assumes a generic user with only a 5-degree mask angle, yet this one assumption no longer represents a significant class of GPS users: the precision users.

If there are 29 satellites available, why is a reduction to 27 such a big issue? Today’s GNSS users are more demanding than the users were even five years ago. Accuracy is the thing, and real-time accuracy is the most important thing. Today GNSS is used in precision applications such as agriculture, surface mining, and seismic drilling. To get the needed accuracy, users of GNSS exclude low-elevation angle satellites to mitigate the effects of the atmosphere. They set the mask angle in their receivers to 8, 10, even 12 degrees. This higher mask angle reduces the number of available satellites to the users, and correspondingly the dilution of precision and associated positioning error goes up, as illustrated in Figure 4.

Figure 4. Picture of DOP performance over various mask angles for June 5, 2006.

Is GPS Properly Managed?

So the issue becomes, is the U.S. Air Force managing the constellation in the best interests of all of its users? Is keeping older satellites in orbit the best policy, and asking the satellite operators to do the best they can with the constellation provided to them? Or is it better to expend taxpayer dollars to replace the older, yet still operational, satellites with newer satellites?

From a user’s perspective, the newer satellites are better — far better than the older satellites in range accuracy, health, and resistance to integrity failures. The increased reliability produced by their redundant systems also acts as insurance against longterm failures of GPS.

Today’s users do expect more from GNSS. While today they have but one choice, in the future they will have at least two others, as Galileo and GLONASS come online. So it is important for the U.S. government to continue to adapt GPS operations to support its current user base.

Recommendations

There is no easy solution to the problem of ensuring that GPS continues to meet today’s user’s needs, since the field of users and applications is becoming more diverse and demanding. For many, the preferred answer is to launch more satellites, keeping the level at 30 satellites, but there are significant cost implications with this approach.

On the other hand, relying on an aging constellation to remain operational is also fraught with peril. Many satellites are on their final legs, with key components on a single point of failure. The clocks onboard the satellites are not as stable as they used to be, and require considerable attention from the satellite operators. The cost of losing satellites is significant for certain sectors in our economy, the sectors that employ precision GNSS. In the distant future this problem will be eliminated through the diversity of fully operational Galileo and GLONASS constellations, but for now the issues are immediate and real.

The U.S. government can and should take several steps to better address the increasing demands on GPS:

• The government should update the SPS Performance Standard to accommodate other classes of users and bring the metrics up to date with respect to current performance.
• The satellite operators should refine their assessments of user impact to include a view of how special classes of users are impacted. In particular this should include precision users and aviation applications. This involves using higher mask angles in their assessments and incorporating receiver autonomous integrity monitoring (RAIM) availability.
• The decision-making authorities in GPS should continue to support an aggressive program to replace aging satellites.

Such improvements will continue to benefit GPS users worldwide, and help ensure the U.S. government’s goal of providing the best PNT service available.

John W. Lavrakas is a consultant in satellite navigation. He has spent the past 26 years in GPS, working in satellite command and control, user operations, GPS receiver development, and satellite navigation performance analysis. Contact him at John.Lavrakas@LavrakasConsulting.com.