Survey and Mapping

January 4, 2008

GLONASS – The Way Ahead

Designers and manufacturers of GNSS products for consumer mass markets may find their next big boost coming from a surprising source — Russia’s GLONASS system.

That was an unmistakable message — and aspiration — expressed by a series of high-ranking Russian governmental officials and representatives of home-grown commercial enterprises speaking at a major GNSS conference in Moscow on April 9 and 10 2007 — the 25th anniversary of the GLONASS program.

Designers and manufacturers of GNSS products for consumer mass markets may find their next big boost coming from a surprising source — Russia’s GLONASS system.

That was an unmistakable message — and aspiration — expressed by a series of high-ranking Russian governmental officials and representatives of home-grown commercial enterprises speaking at a major GNSS conference in Moscow on April 9 and 10 2007 — the 25th anniversary of the GLONASS program.

With 15 operational GLONASS satellites expected to be broadcasting by the end of April and 18 by the end of the year, Russia is looking to bolster its domestic market for GNSS commercial applications and project its presence into international markets over the next few years. Russian officials are fostering a GLONASS industry association and at least 120 Russian companies were reported to be active in the GNSS sector.

More than 600 delegates registered for the International Satellite Navigation Forum, which featured 87 speakers and three tracks of technical sessions. The event was organized by Profi-T-Centre, a Moscow-based conferencing company, and endorsed by the Russian Federal Space Agency (Roscosmos), the Russian Ministry of Communications and Information, and the Moscow City Government on whose premises the forum took place.

An announcement of a decision to add a CDMA signal to GLONASS that would more closely align the Russian system with GPS and Galileo was not forthcoming at the conference, as many had hoped. Nonetheless, a remarkable number of private companies and public institutes joined the proceedings and discussed their efforts to build and use combined GLONASS and GPS receivers.

Russian President Vladimir Putin has put the restoration of and modernization of GLONASS high on his political agenda and is following its progress closely, a fact underlined by the stature of the officials taking part in the forum: Anatoly Perminov, head of Roscosmos; Yuri Nosenko, Roscosmos deputy chief, head of the GLONASS coordination board, and chairman of the forum’s plenary session; and Lt. Gen. Alexander Kvasnikov, deputy commander of the Russian Space Forces.

They were joined at the opening session by Yuri Urlichich, director general of the Russian Institute of Space Device Engineering (RISDE), which designs the GLONASS space and ground equipment; Nikolai Testoyedov, director general of NPO PM “Reshetnev,” which builds the GLONASS satellites; M. G. Lebedev, a senior advisor to the Russian minister of communications and information; and Sergei Burov, vice-governor of the Yaroslavskaya region near Moscow that has served as a kind of GNSS showcase.

Roscosmos’ Perminov noted, “Development of positioning, navigation, and timing capabilities is one of the top priorities of the Russian Federation, particularly through use of GLONASS as a dual-use system. We have a primary objective of [achieving] compatibility and interoperability with, first, GPS, and, second, Galileo.”

Russia has increased its federal budget allocation for GLONASS to 9.88 billion rubles ($379.7 million) in 2007, more than double the 4.72 billion ruble ($181.4 million) federal expenditure in 2006. Launches of six modernized GLONASS-M spacecraft are scheduled this year — a triple launch in September and another in December.

Despite its international title, the event drew a largely Russian audience, with only a few dozen attendees from outside the country. Nosenko underlined this aspect of the forum, saying, “The primary purpose is to inform a broad Russian audience of satellite navigation and its applications.”

Unlike GNSS conferences in most other venues, the focus was pointedly on Russia’s own GLONASS system. Indeed, although the English translation of the event’s title was “satellite navigation,” the Russian name was “International GLONASS Forum.” (In fact, GLONASS itself is the Russian acronym for Global’naya Navigatsionnaya Sputnikovaya Sistema — or global satellite navigation system.)

Official Imprimatur
Formally enshrined in an April 19, 2006, government directive, Russia’s initiative to develop mass market equipment and applications faces many of the same obstacles to commercialization that GPS has had to overcome during the past 15–20 years and some new challenges as well.

The participation of several high-ranking U.S. officials involved in GPS affairs reflected the growing cooperation in GNSS programs between the two countries: Ken Hodgkins, deputy director of the State Department’s Office of Space & Advanced Technology; Mike Shaw, director of the National Coordination Office for Space-Based Positioning, Navigation, and Timing; and U.S. Air Force Col. Mark Crews, chief engineer at the GPS Wing in the Space & Missile Center, Los Angeles Air Force Base.

“Multiple [GNSS] systems create a winning situation for consumers,” Urlichich said, announcing an initiative to create a GLONASS Forum or what he called “an association of lovers of GLONASS.” By working to make the various systems more compatible and interoperable, Urlichich said, Russia will help lay the foundation for global mass markets. “All together, it will make it possible for mass consumers to have GNSS.”

Russia’s struggle to transform a command economy shaped by more than 70 years of top-down, Communist Party–led governmental planning and direction remains a work in progress. Many of the institutions, terminology, and practices of market-based economies remain unfamiliar to both public officials and nascent businesspeople.
Lebedev, the Communications and Information Ministry advisor, undertook a sort of tutorial on entrepreneurialism and private business in his presentation. “It is necessary,” he told his plenary audience, “to understand the value chain in order to successfully develop markets.”

Lebedev showed several slides from presentations by the Galileo Joint Undertaking and the U.S. Office of Space Commerce to illustrate the GNSS value-added chain and GNSS market projections. Later, he noted that in order to “extract profits in this sector, we need to develop business models.”

Although the level of such discussions might seem basic — even primitive — to Western ears, it does reflect a clear desire to learn how to do business in a completely new way.

Home-Grown GNSS
Perhaps the most notable aspect of the conference emerged in the numerous presentations by domestic Russian companies designing multi-system GNSS receivers and offering GNSS-based services.
Since the dissolution of the Soviet Union, when almost all major manufacturing and business activities were based on government ownership and management, commercial activities have appeared in a variety of forms: privatization of former public enterprises, public/private joint ventures, commercial spin-offs from public institutes, and, increasingly, true private startup companies.

ZAO Navis, for instance, exhibited a variety of GLONASS/GPS products at the forum — mostly larger form factors such ad racks and boards for aviation, commercial vehicle tracking, and synchronizing communications systems.

Adding GLONASS to GPS increases costs by 10–20 percent, according to O. A. Borsuk, chief designer for the 11-year-old Moscow-based design bureau. The company has announced a new GPS/GLONASS module, CH-4706, and plans to bring out a 0.13 micron GPS/Galileo L1 chip in 2008.

Another 13-year-old company, Granit, began developing GPS navigation units in 2001 without government support, E. V. Vikharev, Granit’s deputy director of research, told his forum audience. Characterizing the company’s self-financed progress in post-Soviet Russia as “a difficult experience,” he described four generations of product development, including the current Granit Navigator 04 based on SiRF Technology’s SiRFstarIII.

Vikharev said the company has sold 25,000 units in 15 different models to more than 250 different Russian companies and organizations, including 200 Navigator 02 units installed on city buses in one of Yaroslavl’s projects. Granit has developed a prototype GPS/GLONASS/Galileo and should complete the unit by next year.

RISDE and the St. Petersburg–based Russian Institute for Radionavigation and Time (RIRT), two institutes that have relied on government support for much of their existence, have launched commercial development activities. RISDE’s Urlichich described an agreement signed last month for a “public-private partnership that will develop and produce user equipment.”

S. V. Filantchenkov, deputy chief of RIRT’s research division, traced several generations of GPS/GLONASS receivers developed by the institute since 2004. Known primarily for developing the atomic clocks for GLONASS satellites and ground facilities, RIRT is currently designing a receiver that can process GLONASS M and GPS IIR signals. By next year, Filantchenkov said, the institute’s engineers expect to complete an OEM GPS/GLONASS/Galileo RFIC module that would sell in the $12 range for large volumes.

Telematics services, particularly vehicle tracking/fleet management applications, appear to be the most common GNSS businesses to have developed in Russia so far. The Granit and Navis presentation touched frequently on their telematics products and customers. Other telematics-oriented Russian companies taking part in the forum included M2M, ITS Soft, Geizer, and SecTrack.

A typical business development path for the new enterprises is to secure contracts from public agencies at the federal, state, or local levels to design and install systems. These contracts then establish a foundation for further government contracts and product line extensions.

Lebedev cited “expert opinion” in estimating the current Russian market for GNSS products at up to five million units, primarily in government-regulated commercial and professional markets, including safety and security.

Looking ahead to a true consumer market in Russia, he pointed to two platforms that have incorporated GNSS technology in many other countries: private cars and portable electronic devices. Russia’s automobile market over the next five years is expected to produce sales of two to three million, while 380,000 portable PCs and mobile phones were sold in the country last year.

Although the central government is accumulating large financial reserves from taxes on extraction and exports of natural resources, however, Russia still lacks channels, workplans, and, the experience needed to recycle part of these through the nation’s emerging small and medium enterprises. One promising indication, however, could be found among several representatives of private Russian banks who attended the event to meet with the entrepreneurs and evaluate the prospects for investing in the GNSS businesses.

CDMA: Decision Still to Come
All this GNSS development activity is particularly remarkable considering that GLONASS is a frequency division multiple access (FDMA) system that differs markedly from GPS and Galileo. FDMA is, in fact, the inverse of GPS’s code division multiple access (CDMA) design in which each satellite broadcasts a distinctly coded signal on a common frequency.
In contrast, GLONASS transmits a single code on different frequencies allocated to antipodal sets of GLONASS satellites using two swaths of spectrum — currently from 1598.0625 to 1609.3125 MHz (above GPS L1 centered at 1575.42 MHz) and from 1242.9375 – 1251.6875 MHz for L2 (compared with 1227.6 MHz for GPS L2).

The $64 million question — or, closer to the mark, the $68-billion question (to pick up on Shaw’s projection for the global GNSS market in 2010) — is how compatible the Russians will decide that GLONASS will be. Russia has committed to reaching a decision on the question of adding a CDMA signal by the end of 2007.

Different perspectives and philosophies are competing among the country’s various institutional groups. A new generation of engineers appears inclined to forge greater interoperability with other GNSSes by adding a CDMA signal on a common frequency.

The main arguments raised against CDMA seem to be: single point of failure if all GNSS signals at L1/E1, national security issues, the matter of paying for new civil signal design, and an element of Russian uniqueness.

Numerous GNSS manufacturers — among them JNS, NovAtel, Trimble, Leica, Magellan, and Topcon — already offer combined GPS/GLONASS receivers, typically for professional and commercial activities such as surveying, geodesy, and construction. But such equipment is substantially more complicated in design and expensive — a long way from becoming consumer-friendly.

By having a L1 civil signal apart from the band in which consumers will find most benefit from GPS and Galileo (and, for that matter, China’s Compass GNSS), GLONASS requires manufacturers to widen the reach of their receivers’ antennas and “front-end” components.

As the GPS Wing’s Crews pointed out in his presentation, the key technical issue may be that CDMA-based systems can more easily filter out a common delay in the GNSS time signals on a common frequency. With FDMA systems, he continued, “We can calibrate our filtering for multiple frequencies [and time delays], but it increases costs. That means it’s an issue for making affordable, mass market equipment.”

Nonetheless, the American delegates went out of their way to emphasize that GLONASS signal design questions are completely up to the Russians to sort out.

By

January 2, 2008

Geospatial Fusion on the Fly

(The following online version is text only. To see graphs, charts, and images, download the article pdf above.)

The development of GNSS worldwide has fundamentally changed the way many professions conduct their business.

Arguably, the profession of surveying has been most affected because surveyors, at their core, are experts at measurement. For millennia they have been the first to take advantage of any new technology that improves their ability to locate objects accurately.

(The following online version is text only. To see graphs, charts, and images, download the article pdf above.)

The development of GNSS worldwide has fundamentally changed the way many professions conduct their business.

Arguably, the profession of surveying has been most affected because surveyors, at their core, are experts at measurement. For millennia they have been the first to take advantage of any new technology that improves their ability to locate objects accurately.

Until recently, the landmass of Alaska has had little in the way of either control networks or boundary surveys. This is why GNSS has been a godsend for our company, Tanana Chiefs Conference (TCC).

TCC is a nonprofit corporation that primarily consolidates medical and social services for 42 small Alaska Native villages located in remote, mostly roadless regions of the interior. However, we also employ a small group of professional surveyors whose ongoing task is to lay out boundaries for the Alaska Native Claims Settlement Act (ANCSA) village and regional corporations.

These surveys, covering thousands of square miles each summer, are part of a much larger 35-year effort by the United States Bureau of Land Management to delineate government and tribal land claims throughout the state.

In the days before GNSS, an ANCSA project required a major expedition each year to hire surveyors, assemble the equipment and supplies, and mobilize for a survey based in some distant village. It took up to six crews of surveyors and helpers, an office staff of five or six, and a DC-3 full of tripods, total stations, and chainsaws.

Today, 15 years later and with sophisticated GNSS equipment, we get by with a lot less. The results are more accurate and trustworthy, and only a single person is needed to run the surveying office, which consists of a laptop computer.

Villages without Boundaries

Although GNSS has solved many difficulties of large-scale remote surveys, it hasn’t been nearly as helpful at the local level. The villages where our crews are based each field season are scattered over 235,000 square miles — a region slightly smaller than Texas.

These villages generally have a few boundary problems of their own and always a subdivision or two that needs to be surveyed. In remote Alaska, flying in a survey crew is very expensive, and few villages can afford it, so little has gotten done over the years in the way of addressing village boundaries. To be helpful, our company generally donates a week of what we call VTS (village triage surveying) to the various places we visit.

Unfortunately, once we get there, the local work is time consuming because traditional field methods are needed for much of the control and design work. For example, Athabascan villages are communal in nature and rarely contain fences that divide housing and possessions. A good deal of time is needed to locate everything in sight and figure out who owns what. Moreover, original boundary markers are scarce, and hours are spent digging up old axles and snowmobile parts in an effort to uncover the few remaining survey monuments.

It occurred to us that aerial photography might be a worthwhile tool to make our efforts more useful to the locals in the short time we had. For example, if a subdivision could be designed not from a weeklong topographic survey, but from a table-sized, high-resolution orthophotograph, it would save a lot of time and trouble.

Most villages in interior Alaska have been aerially photographed at one time or another, but timely orthophotography is rare, and the resolution of even the best photos — about one pixel per foot — is less accurate than needed. To distinguish the incredible variety of objects scattered throughout a village, something in the range of two to three centimeters per pixel (about half the width of a soda can) would be more useful.

Although new photogrammetric techniques make this high resolution achievable, commissioning new low-altitude photography and the associated expedition – a very expensive undertaking — is not an option for these distant villages.

Off the Shelf Solutions?

We were naïve enough to think that, with a little experimentation, we could achieve these results with off-the-shelf consumer equipment. After all, we had an available helicopter that was used for U.S. Bureau of Land Management (BLM) work, and high-resolution, 10- to 20-megapixel consumer cameras were just now appearing on the marketplace.

It sounded simple enough, why not rectify a series of hi-res, low-altitude digital photos taken from our helicopter?

However, spending a little time investigating this idea only demonstrated how little we knew about photogrammetry. The process wasn’t nearly as easy as we thought. We almost abandoned the idea, but, once again, GNSS saved the day and provided the key to a solution that made everything work.

Digital aerial photography cameras are precise and complex instruments and cost upwards of $500,000. Their large 23×23-centimeter charge-coupled device (CCD) array must be tightly calibrated in conjunction with a fixed camera lens to compute distortion values unique to each camera.

Based on this calibration, software algorithms can then warp each pixel exactly the right amount to remove the lens distortion, which, in turn, allows for pixel matching and the creation of accurate digital terrain models (DTMs) from stereo pairs of georeferenced photos. The calibration repeatability in these cameras is so high that accurate orthophoto mosaics can be assembled using relatively few photo control points on the ground.

A consumer-grade camera, however, even a good one, is not designed for this tight a tolerance. Although such cameras’ lens characteristics can be calibrated, the repeatability is diminished as even a slight change in alignment— say, due to a tiny machining error in a lens bayonet mount — can change the calibration values each time the camera is used.

As important as the camera is the software. Dedicated, full-featured photogrammetric suites are used to rectify digital aerial photos—but these start at a major-league price of $50,000.

Then we came up with a possible alternative.

In recent years relatively inexpensive photo-modeling software has appeared in the marketplace. This software is capable of making accurate 3D models of anything that can be photographed — something as small as a Neolithic human footprint preserved in shale or as large as the ornate façade of a medieval church. It is also commonly used to reconstruct automobile accident scenes, creating 3D computer models for forensic evidence.

In spite of the smaller scale of such subjects, the photo-modeling software shares the same mathematical principles used by dedicated photogrammetry suites. So we explored this idea. Some searching on the World Wide Web led to the discovery of a 3D photo-modeling software used primarily by architects and archeologists.

Although created as companion software to be bundled with an imaging total station, the software can also serve as a stand-alone product that can manipulate any set of controlled stereo pairs — a pair of images containing a minimum number of corresponding photo control points with accurate x, y, and z coordinates. The program is designed to work with tiny, circular photo targets, which can be automatically registered with an order of magnitude greater precision than the human eye.

The technique can produce remarkably accurate results, but, as always, there is no free lunch. To compensate for the looser reliability of lens calibration on small format digital cameras, the software requires a denser network of photo control targets. The total station with which the photo modeling software is usually paired, for example, can populate its digital photos with scores of accurate data measurements for use by the software.

This photo control requirement has relegated photo-modeling software to working in small confined areas. Theoretically, however, it should also work on a larger scale if sufficient photo control is available.

So, it was tempting to think that, with the eight dual-frequency GPS/GLONASS receivers we normally employ in BLM surveys and a few rented four-wheelers, the requisite photo control could be readily established on a village scale. (I imagined survey crews scooting around on ATVs, scattering small aerial targets in their wake like Frisbees, each measured to sub-centimeter accuracy using on-the-fly GNSS!)

Upriver for a Real Test

The Alaska summer is short. We barely had time to fly a test mission with the helicopter and work out altitude, camera settings, and target sizes before we needed to get under way.

Our first real trial took place at Huslia, a village on the Koyukuk River about 10 days by barge from Fairbanks. This river flows from the south flank of the Arctic Divide through broad, glacially carved valleys in the rugged Endicott Mountains of Alaska’s Central Brooks Range.

The Huslia village council had requested a new subdivision survey because about half the residents lived in a still-unsurveyed portion of town. In this congested central village space, subdivision lots must be custom designed using polygonal shapes to conform to each tenant’s use and occupancy. The polygon lots are separated by a chaotic layout of existing roads and trails.

This was exactly the type of situation we had in mind for aerial surveying, but the timing was rather tight. Only a week earlier we had ordered the software from Nick Russill, managing director of TerraDat UK Ltd., a geophysical consulting and contracting company based in Cardiff, Wales. Nick had generously volunteered to help us with the Huslia project because the photo-modeling software has a learning curve, and the giga-pixel, square-kilometer aerial survey would be pushing this modeling software into uncharted territory.

The software package was delayed in transit, however; so, at the last minute Nick changed his travel plans, jumped on a flight to Alaska, and hand-delivered the software.

He arrived by helicopter, intercepting our survey barge, Seloohge, on the Koyukuk River about a day’s voyage below the village. Talk about customer support!

The following morning, as the barge neared Huslia, the crews crowded into the Seloohge’s pilot boat and sped away with a stack of homemade targets that consisted of several dozen 18 inch diameter white vinyl disks packed with beach sand. At the village it didn’t take long to rent a few ATVs from which to scatter the targets, and, by the time we arrived with the big boat, about two hours later, all the requisite photo control was in place.

The targets were roughly distributed in open areas at 80–100-meter spacing throughout the site. As soon as the targets were placed, they were measured to sub-centimeter accuracy using dual baseline, stop-and-go GPS techniques, consuming another one to two hours.

Soon thereafter, Nick and I found ourselves hovering 1,200 feet above the village in a helicopter with the rear door removed. Compared to traditional aerial photography, the technique was definitely low-tech. The camera and stabilizing gyro were suspended from a bungee cord looped around the neck of the photographer, who then leaned out the door, pointed the camera straight down with an outstretched hand, and took photos every second or so as the aircraft slowly flew parallel strips across the village.

Although we soon learned that an onboard guidance system utilizing preprogrammed routes would be more efficient and provide for consistent coverage, this first effort relied entirely on the pilot’s ability to fly parallel routes based on observed ground features, a task more difficult than it sounds. The resolution of the imaging at Huslia topped out at six centimeters per pixel but subsequent improvements in our camera handling techniques improved this to three centimeters per pixel.

The digital camera was then calibrated using a companion program of the photo-modeling software. The program automatically computed the lens distortion parameters by analyzing a series of photographs taken from various angles of a target grid, an E-sized plot of a .dxf image file (included with the software) that we had carefully taped to the galley window of the barge.

Next, choosing 14 photos from our overflight of the village that provided the best overlapping coverage, Nick guided me through the stereo pair registration, measurement of control and tie points, and the creation of a DTM. The process is fairly straightforward once a host of various keyboard shortcuts are mastered.

The software maintains a point data file which can be quickly populated with the adjusted x, y, and z coordinates of the aerial targets. Stereo pairs, selected from a set of two photos that contain roughly 60 percent common overlap, are oriented by the identification of a minimum of four aerial targets visible in both images, plus an additional four to six tie points. Tie points are distinct, uncoordinated points, such as a white food bowl in a dog yard, which can be positively identified on each photograph.

The software automatically matches pixels at a selected tie point and will either accept it or reject it based on the certainty of the match in the corresponding photo. As each pair of tie points are identified, the accuracy of all the points can be examined with a network bundle adjustment routine.

The creation of a digital terrain model, a three-dimensional surface model of the overlapping area contained within the stereo pair, is a little more problematic as it relies on user input to identify breakline positions that are required to assist the software in making accurate pixel matching and elevation determinations.

Breaklines, which are drawn as polylines, are placed where sudden breaks in terrain exist, such as a ditch at the edge of a road or where the ground meets the wall of a house. Photogrammetrists rely on stereo imaging displays for this time-consuming process which manages to be both tedious and frustrating. Fortunately, for those of us using a laptop without a stereo display, the modeling software contains a useful workaround by supplying an auto-correlater that assists the operator with the exact placement of each corresponding polyline vertex.

By late afternoon we had a product: a mosaic of orthophotos held together with cellophane tape that was then proudly displayed on a big table in front of a lively crowd at the village council office. The resolution of the mosaic was such that an observer could easily pick out the smallest objects, and the villagers had no trouble identifying each other’s possessions as we sketched in new lot lines. Visible power poles and overhead wires helped with creation of utility easements.

After dinner, the marked-up photo mosaic was imaged in computer-aided drafting software, and vectors were created to match the layout. Point coordinates were identified for each lot corner position in the subdivision, exported into GPS receivers and, the following day, using real-time kinematic (RTK) techniques our survey crews set the monuments that defined the new subdivision.

The approximate accuracy of the resulting boundary monuments produced 1:50,000 closures, basically the sub-centimeter accuracy one would expect from dual-frequency, differential GNSS measurements. Note that the accuracy of the surveyed monumentation is independent of the accuracy of the aerial photo. In only two days we had accomplished what used to take a week of hard work, and at the same time we created a
very useful product for future land planning in Huslia village.

Low-Cost Accuracy

Of all the things we learned from this experiment, what surprised me the most was the accuracy of the orthophoto. A bundle adjustment of the control and tie points generated error ellipses well within the subpixel range, a fact verified by quality control checks comparing GNSS measured features with corresponding photo locations.

Without doubt it is the dense network of precisely measured control points that allows for this exactness, by constraining the photography like tacks on a board. A field-generated, high-resolution orthophoto of this accuracy could be a powerful new tool for surveyors.

The speed, precision, and reliability of GNSS-measured target networks, combined with the development of high-resolution small-format cameras and well-designed photo modeling software, now makes this possible.

What began as an idea to make pro bono work in the villages more efficient is now opening doors for revenue-generating enterprises such as accurate terrain modeling of mining, development and environmental sites, and low-cost, high-resolution, confined area photo-mapping for projects such as road intersections, construction sites, and siting and layout of resorts.

This is just one example of how the development of GNSS, surveying, and the technology of measurement – and the village of Huslia — have benefited in more ways than had been anticipated.

By

December 1, 2007

Active GNSS Networks and the Benefits of Combined GPS + Galileo Positioning

Without question GPS has revolutionized precise positioning since its advent about 20 years ago. Real-time methods to quickly fix carrier phase integer ambiguities — the key to precision — have been developed and are often referred to as RTK (“real-time kinematic”) techniques.

Without question GPS has revolutionized precise positioning since its advent about 20 years ago. Real-time methods to quickly fix carrier phase integer ambiguities — the key to precision — have been developed and are often referred to as RTK (“real-time kinematic”) techniques.

RTK is an advanced manifestation of the principle of differential positioning, a method that requires at least one reference station with known coordinates to simultaneously track GNSS satellite signals. Carrier phase measurements are used in addition to pseudoranges due to their superior accuracy.

Nevertheless, ambiguity resolution is only possible as long as the user (the “roving receiver”) is located in the vicinity of this reference station — let us say, within a radius of approximately 10 kilometers. Within this short range the benefits of the often-employed “double differences” technique can be effectively exploited: Differences of observations between a primary and a secondary satellite are formed on both the rover and the reference site and these two quantities are then subtracted, yielding a derived measurement between both sites that is free of satellite and receiver clock offsets or errors.

Fortunately, the atmospheric errors are spatially correlated and can be reduced in the double difference measurements to a reasonable extent. Thus, it is relatively easy to fix ambiguities of short baselines, whereas it becomes increasingly difficult to do so over longer baselines due to decorrelation of the atmospheric delays.

As a result of this decorrelation, the service area of conventional RTK systems allowing for quick ambiguity fixing covers about 300 square kilometers. To provide service in an area the size of the contiguous United States (9,800,000 square kilometers) would require more than 30,000 reference stations. Even for a country as small as Germany (357,000 square kilometers) more than 1,100 reference sites would still be needed to provide complete coverage — an enormous challenge in terms of infrastructure installation, operations, and maintenance costs.

The solution for this problem: Use multiple reference stations to derive atmospheric corrections. Because the coordinates of these fixed stations can be determined precisely — or can be treated as tight constraints — the atmospheric (ionospheric and tropospheric) effects on GNSS signal propagation can be derived from the correlated data.

These station-, baseline-, or satellite-specific corrections can be interpolated at the rover site. Hence, atmospheric errors can be significantly reduced and GNSS reference networks can substantially increase the distance between stations while still providing the accuracy level on conventional RTK systems.

The reference networks that provide such correction data are often called “active GNSS networks,” referring to their continuous operation. Most of them offer both real-time and post-processing services.

By adding to the number of satellite signals available to these networks, users on the road/in the field can improve their performance by allowing optimization of satellite geometry (the selection of a subset of available signals that reduces the dilution of precision (DOP) factor), use of multiple frequencies for carrier phase integer ambiguity resolution, and for achieving so-called “overdetermined solutions.” With multiple GNSS systems under development in addition to GPS that are increasingly compatible or even interoperable, this prospective approach is becoming ever more attractive.

This article outlines the added value from combined GPS+Galileo data processing — rather than GPS-only data processing — in the framework of active GNSS network positioning. In particular, we will look at how such an approach can improve performance in the presence of traveling ionospheric disturbances that produce marked increases or decreases of signal propagation delays.

(For the rest of this story, please download the complete article using the link above.)

By

October 14, 2007

Precise Point Positioning and Its Challenges, Aided-GNSS and Signal Tracking

Q: What is precise point positioning (PPP), and what are its requirements, advantages and challenges?

A: Precise point positioning (PPP) is a method that performs precise position determination using a single GPS receiver.

This positioning approach arose from the advent of widely available precise GPS orbit and clock data products with centimeter accuracy. These data can be applied to substantially reduce the errors in GPS satellite orbits and clocks, two of the most significant error sources in GPS positioning.

Q: What is precise point positioning (PPP), and what are its requirements, advantages and challenges?

A: Precise point positioning (PPP) is a method that performs precise position determination using a single GPS receiver.

This positioning approach arose from the advent of widely available precise GPS orbit and clock data products with centimeter accuracy. These data can be applied to substantially reduce the errors in GPS satellite orbits and clocks, two of the most significant error sources in GPS positioning.

Combining precise satellite positions and clocks with a dual-frequency GPS receiver (to remove the first order effect of the ionosphere), PPP is able to provide position solutions at centimeter to decimeter level, which is appealing to many applications such as airborne mapping. PPP is different from double-difference RTK (real-time kinematic) positioning that requires access to observations from one or more base stations with known coordinates. The word “precise” is also used to distinguish it from the conventional point positioning techniques that use only code or phase-smoothed code as the principal observable for position determination.

(For the rest of Yang Gao’s answer to this question, please download the complete article using the PDF link above.)

Q: Does Aided-GNSS improve signal acquisition, tracking, or both?

A: A-GPS (Aided/Assisted-GPS) and more recently its extensions, A-GNSS, have been introduced to substitute for missing satellite broadcast data when access is intermittent, difficult, or impossible due to signal obstruction. It has expanded the capabilities of the traditional receiver in reducing the time to first fix (TTFF), enabling “high sensitivity” modes, improving the performance in urban canyons and indoors, and incidentally, boosting the receiver’s efficient use of power.

Multiple ways have been developed to deploy an A-GPS server, and to distribute the process flow between the server and the mobile. The mobile station–based method places the position determination in the receiver, while the network-based method relegates it back to the server.

Other notable factors influence the architecture. The assistance can be one-way, where information accessible at the server flows down to the receiver, or in closed loop, where the information is uploaded to the server, processed remotely applying far larger computational resources and extra knowledge not available to the receiver, and then pushed back to the receiver in its final form.

We will now introduce two simple rules that will illustrate the rest of the explanations:

Rule 1: For A-GPS to be practical, the assistance information should not be stale when ready to be used at the mobile. In more technical terms, the assistance information persistence needs to be longer than the sum of network latency plus server and mobile processing times,

Rule 2: In a closed loop architecture, the information collected at the mobile, and processed at the server should be returned fast enough before the internal state of the mobile changes too much. We can reformulate it as the sum of the round trip network delay plus server and mobile processing times have to be shorter than the process time constant to control the mobile.

(For the rest of Lionel J. Garin’s answer to this question, please download the complete article using the PDF link above.)

By

A New Version of the RTCM SC-106 Standard, the Probability of Solving Integer Ambiguities

Q: The RTCM has announced a new version of its widely used differential GPS (DGPS) standard. Why did the group decide a new standard — Version 3 — was needed, and what are the benefits compared to Version 2?

Q: The RTCM has announced a new version of its widely used differential GPS (DGPS) standard. Why did the group decide a new standard — Version 3 — was needed, and what are the benefits compared to Version 2?

A: Initially, the RTCM (Radio Technical Commission for Maritime Services) Subcommittee (SC104) Version 2 standard was developed with marine DGPS as the target application. One of the design goals for the RTCM SC 104 standard was to have correctional information readily available for user equipment. The outlines of the messages were specifically tailored for low bit rate data links.

Not until the beginning of the 1990s did RTK (real-time kinematic) surveying applications come into focus for RTCM. The committee drafted new RTK messages based on the proven DGPS messages. Although the DGPS messages only contain corrections, the RTK messages also allow transmission of raw observables from the satellite signals. RTCM tentatively published redundant means of disseminating precise RTK information with the aim of gaining practical experience during their implementation.

The first implementations by different manufacturers had diverse interoperability issues. For instance, various manufacturers have different sign conventions for representing the carrier phase observations, which resulted in incompatibilities when mixing receivers of different manufacturers.

(For the rest of Dr. Hans-Jürgen Euler’s answer to this question, please download the complete article using the PDF link above.)

Q: What is the probability of correctly resolving integer ambiguities and how can it be evaluated?

A: Resolving, or “fixing”, carrier phase ambiguities to integer values is ultimately based on statistical assumptions and testing. As such, a probability is associated with resolving any particular ambiguity correctly. Evaluating the probability of correct fix (PCF), that is, the probability that the ambiguities are fixed to the correct integer values, is particularly important for safety-critical applications where an incorrect ambiguity fix would produce hazardously misleading information (HMI).

In fixed-ambiguity carrier phase processing, the usual procedure is to begin by estimating the carrier phase ambiguities as real-valued (“float”) parameters and then to determine their integer values. A difficulty with this method is that, although the least-squares adjustment or Kalman filter used to estimate the real-valued ambiguities provides an estimate of their quality (a covariance matrix), it is not obvious how to obtain an estimate of the quality of the integer ambiguities.

In most carrier phase ambiguity estimation methods, integer quality is validated using some sort of statistical test. These generally involve testing the least-squares sum-squared residuals of the best fitting integer solution against the second best fitting solution. The test statistic is then compared against a threshold value, the idea being that if the best solution is sufficiently better than the second-best solution, then it must be correct.

(For the rest of Kyle O’Keefe and Mark Petovello’s answer to this question, please download the complete article using the PDF link above.)

By