Caltrans report says Skymeter "substantially better" than straight-GPS


A Caltrans draft report says tests of Skymeter show it performs "substantially better" than regular GPS (satellite location finding) units in providing accurate locations in a difficult urban environment. Skymeter is being offered by the small Toronto start-up company of the same name as a possible wide area tolling device that could assess tolls in a variety of city streets with minimal roadside infrastructure.

Most electronic tolling is presently performed with very short range wireless sensing and messaging and an infrastructure of gantries just meters or feet over the top of vehicles at each tolling point. 5.9GHz in development offers a range of up to 1200m (3/4 mile) vehicle to roadside, and also vehicle-to-vehicle communications. But 5.9GHz needs considerable roadside infrastructure too, and its deployment is uncertain.

Two main technology needs

Asfand Siddiqui, an engineer in the California Department of Transportation's (Caltrans) Division of Research and Innovation says Caltrans won't endorse or promote any particular product, but it has identified technology needs and must be in a position to tell a city or a region which is interested in area road pricing or congestion pricing: "Here's a technology, here's what could work for you."

He told us in an interview today that they foresee two uses for satellite location finding (GPS) in California:

- snowplowing, principally in the Sierra Nevada mountains for which regular GPS works well

- urban road pricing/congestion pricing where the 'urban canyon' effect of tall buildings generally disrupts GPS causing it to perform poorly due to what is known as 'multipath' or dispersal of signals

Last year, Siddiqui says, during discussion of area pricing they got curious about Skymeter.

"We wrote a matrix of requirements for testing. We decided to hire a prominent GPS expert, Jay Farrell, and get him to test Skymeter alongside the best commercial off-the-shelf unit in the urban canyon environment where we have these big problems - in San Francisco."

Farrell, an electrical engineering professor at University of California Riverside and a leading author on GPS organized the test of Skymeter against a NovAtel unit, considered by Caltrans to be the best commercial off-the-shelf GPS unit made.  

The tests were conducted January 29.

NovAtel is based in Calgary Alberta, and is a leading GPS supplier. Skymeter by contrast is a Toronto Ontario startup with 5 or 6 employees that has only prototypes and is best known by the tireless proposal writing and advocacy at conferences of its chief scientist Bern Grush.

In the small world of toll technologies to be "grushed" has become a verb for oversold.

Paper is draft text, out for peer review

Skymeter has recently been distributing a 33 page Caltrans paper on the tests (see link at the bottom of the page) that has no title, no date, no author and no attribution. Siddiqui at Caltrans says it is a draft text he wrote with Farrell and others of a forthcoming report that is currently out for peer review and will be subject to revision. It is a work in progress.

We're quoting from that 33 page draft text in what follows and the pictures also come from it.

"Amazed" at difference

Unlikely to change is the major conclusion that Skymeter starred where the competition failed - in the urban canyon of San Francisco.

"We were quite amazed at the difference.  It was hands down, no contest. The tests were very positive (for Skymeter)," Siddiqui says.

The draft report executive summary says: "The data obtained during these experiments support the Skymeter claim that their product performs very well in urban environments. Although Skymeter performance was not perfect, it performed substantially better than the COTS (commercial off-the-shelf) receiver with which it was compared."

Skymeter and COTS were mounted in an SUV, wired to record their data in a laptop computer and driven on a  course through the 'canyons' of downtown San Francisco. GoogleEarth was used for reference.

Skymeter was usually able to continue to get locations through dense urban canyons, whereas the COTS shut down for long stretches and was well off course in many of its reads.

Ability to process poor signals

The report attributes Skymeter's success to its superior ability to process low signal-to-noise ratios. Also the Skymeter software incorporates a vehicle kinetics model that allows it to fill in when the satellite signal disappears.

Skymeter people have been shy in telling us quite what they do that others don't. Siddiqui thinks it is partly a more sensitive antenna, but mainly better signal processing.

"Their magic shows up when the (satellite) signal is very poor," he told us.

We'd written once that Skymeter makes use of some supplementary positioning, such as cell towers. That's wrong, Siiddiqui says. Skymeter relies entirely on the satellite signal. If there is no signal at all it fails like the competition, but that is rare even in among high buildings. Its strength is in making sense out of broken satellite reads.

The draft concludes: "The data sets acquired on January 29 2009 support the Skymeter claim that their product performs very well in urban environments."

The draft paper:

http://www.tollroadsnews.com/sites/default/files/DraftEval.pdf

Skymeter:

http://www.skymetercorp.com/cms/index.php

TOLLROADSnews 2009-08-31

AttachmentSize
DraftEval.pdf3.84 MB