
ContinuousWave Whaler Moderated Discussion Areas ContinuousWave: Small Boat Electrical Accuracy of GPS Receiver Speed Measurements

Author  Topic: Accuracy of GPS Receiver Speed Measurements 
jimh 
posted 02082009 10:49 PM ET (US)
What is the accuracy of speed measurements made by a consumergrade GPS receiver? This topic was discussed previously here but yielded no real scientific conclusions.. I thought it was worth a second look. I searched for other published studies and found several. A study in 2004 using a nonprecision enhanced GPS receiver by T. H. Witte and A. M. Wilson concluded as follows: "The GPS is accurate for the determination of speed over ground (about 10 times more accurate than a car odometer) when moving at relatively constant speed in straight lines and is competent at determining speed on curved paths, although some overshoot does occur during transitions. Absolute error increases slightly at higher speeds but in percentage terms is less. In addition, when the system is tested under conditions of sudden changes in speed some inadequacies become evident. The system smooths the peaks and troughs of rapid accelerations and decelerations, which is attributed to inherent smoothing within the mathematical algorithm and the one sample per second output of the system." http://www.rvc.ac.uk/sml/people/documents/WitteJBio2004GPS.pdf Another study, perhaps not as rigorous, concluded: "Higher speed sampling rate eliminates aliasing errors, increases the accuracy of the average speed and produces data that better represents acceleration. Speed measurement accuracy approaching 0.01 knot becomes possible." http://mtbest.net/speed_sailing_helmet.html Witte and Wilson revisited the topic in 2005, this time with precisionfix GPS receivers with WAAS enhancement. They then concluded: "In conclusion, GPS units employing WAAS/EGNOS technology provide much more accurate position data and display enhanced accuracy of speed determination compared to nonWAAS GPS units." http://www.rvc.ac.uk/sml/people/documents/WitteJB2005WAASGPSfinal.pdf The speed error in the GPS WAAS test mentioned above was found to be less than 0.2meters/second for roughly twothirds of the measurements. An error of 0.2meters/second is 0.2meters/second x 2.2369MPH/meter/second = 0.45MPH I think you can say that the speed shown by a GPS receiver with WAAS enhancement when a boat is traveling in a straight line is accurate within about 0.5MPH most of the time. There are some anecdotal reports that manufacturers of GPS receivers are claiming that their devices are accurate to 0.1MPH, but I could not find a direct citation of that claim being made in the OEM literature. I also found some speculation that consumergrade GPS receivers may compute speed only based on a twodimensional fix, so if there is any vertical component to the motion, the speed may be in error. Vertical position calculation seems in general to be less accurate than horizontal. I often watch the altitude reading on my GPS receiver jump around when in fact the receiver is quite stationary. In marine applications, one would hope there was not much vertical motion involved, so speed errors from vertical velocity should not be a problem. The notion that speed is calculated with an accuracy of about 0.45MPH seems reasonable. I have often observed that a stationary GPS receiver will report its speed on occasion as being as much as 0.4MPH. 
jimh 
posted 02092009 08:52 PM ET (US)
Following up on the citation above which mentions speed measurement by GPS receiver with an accuracy of 0.01knot, it should be mentioned that this level of accuracy is not available in consumergrade GPS receivers which use a trackpoint speed measurement, that is, which measure the speed by finding the distance traveled between calculated positions versus time. Measuring speed based on the calculated positions will contain errors induced by the errors in the deduced positions. The claimed potential for high accuracy speed measurement is based on using Doppler frequency shift measurements with sophisticated and specialized receivers. For more on the method and the results, see 
Tom W Clark 
posted 02102009 12:07 AM ET (US)
That's OK Jim, I'll take your word for it. So there you have it: 1/2 MPH accuracy. That is very good but not as precise as I had imagined GPS calculated speed to be. Thanks you for the clarification. 
jimh 
posted 02102009 09:22 AM ET (US)
I do not believe that the accuracy of a GPS receiver position finding is particularly dependent on the speed at which the receiver is moving, that is, the position accuracy is probably the same if the GPS receiver is at rest or if it is moving 30MPH. If the receiver were moving very fast, perhaps there might be some greater influence. (I'll leave this for others to explore.) In calculating the speed, it seem clear that the influence of any error in position will be greater at lower speeds, at least in terms of a percentage error. If the speed calculation is based on track points that are only a short distance apart, any error in those positions will have more influence on the calculated speed than if the two track points were, let's say, ten times farther apart. Speed is likely to be more accurate as a percentage error at 30MPH than at 3MPH. Errors in position would have the most impact if they resulted in the incorrect position being exactly on the course line track, where they would either add to subtract from the distance travelled. The errors should tend to be random, and therefore they will likely not be on the track line course, but displaced to one side or the other. Over time the errors probably tend to be distributed in a random distribution about the true position. This means that if a boat holds a straight course line for a reasonable period, the position errors will probably tend to be random and their influence on the speed calculations will tend to cancel out, as long as some sort of averaging of the speed calculation is being performed. This detail is probably unique to each brand and design of GPS receiver. I suspect that the speed averaging, if any, is probably only done over a short period, perhaps over a second or two. Some newer GPS receivers are able to produce calculations of position and speed at much faster rates than older receivers. Some new GPS receivers are operating at 5Hz, that is, they produce five position calculations each second. Older receivers are not as fast, perhaps more like 1Hz. If the speed averaging algorithm uses a fixed period, say five seconds, then a receiver with higher data rates will be averaging more measurements. This may lead to higher accuracy in the speed calculations. 
bluewaterpirate 
posted 02102009 09:29 AM ET (US)
Here's my 5 year old Garmin 376c in the Automobile Mode in action. Note the data windows to the right on the display. http://albums.phanfare.com/4892322/2159339_2459167#imageID=31003240 Tom 
pglein 
posted 02102009 02:26 PM ET (US)
.41mph may not seem like a lot to someone who cruises along at 2030 knots in a planinghull boat, but it is a huge number for racing sailboats and displacement hull cruisers. On an eight hour cruise at 6mph, that's an extra 45 minutes. 
jimh 
posted 02142009 10:14 AM ET (US)
In their new 2009 product brochure, Standard Horizon give a specification for the accuracy of the speed calculation of their new GPS receiver, which is a quite modern receiver rated as being capable of "50 channel" (i.e., 50 simultaneous tracked satellites). The stated speed accuracy specification is: ACCURACY: Speed 0.3 knots RMS Converting to statute miles, this is 0.345MPH. The notation "RMS" generally is interpreted as "root mean square." RMS is a statistical measure of the magnitude of a varying quantity. You might consider it to be an averaging technique. This is similar to the notation of statistical averaging of Witte and Wilson when they specify the measured speed error being within a particular range for 66percent of the time. This figure for speed accuracy (0.345) from a marine GPS receiver manufacturer is in close agreement (0.105) with the measured accuracy reported in the tests of Witte and Wilson (0.45). This seems to reaffirm the notion that the accuracy of the speed indication on a consumergrade GPS receiver is in the order of about plus or minus 0.5MPH most of the time If others encounter speed specifications from the equipment manufacturer of a GPS receiver, please pass along the information. 
Chuck Tribolet 
posted 02142009 02:48 PM ET (US)
Which is a whole better than most car speedometers. I've calibrated several, with a stop watch and measured miles, or WAAS GPS, or both. They all read 24 MPH high at freeway speeds. Stopwatch and GPS agreed more closely. Why do all read high? Because the automakers don't want a

number9 
posted 02152009 04:49 AM ET (US)
Read the post and references a few days ago which made some reference to updating speed and decreased accuracy while accelerating in what most would consider small increments. Seems there is no question of the GPS steady speed accuracy. Several months ago posted a topic on the Velocitek SC1 and my perceived advantages of being able to quickly pick up the relatively minor speed changes. Seems the technology used is similar to the referenced Navi GT11 GPS units. Do we know from all this referenced material if the modern technology consumer marine GPS units provide the same speed sampling rates as the Navi GT11 and Velocitek? Only asking because I can't really figure it out. Thanks. 
jimh 
posted 02152009 09:09 AM ET (US)
The Velocitek SC1 uses the UBlox Antaris LEA4A GPS receiver. From what I can tell, the UBlox Antaris LEA4A chipset is nothing out of the ordinary. There is no mention in the manufacturer's data sheet http://www.ublox.com/products/Data_Sheets/LEA4x_Data_Sheet(GPS. G4MS406143).pdf of any sort of advanced feature like a Doppler frequency shift measurement. The sales literature of the manufacturer says: "The LEA4A is a lowcost GPS module featuring the new ublox 16channel ANTARIS 4 receiver technology...." I don't see any basis to infer that this GPS receiver is going to be more accurate than the ones tested or rated above, that is, about 0.345 to 0.45MPH speed accuracy most of the time. The principal claim to fame of the SC1 is its data logging ability, and through the use of specialized software, to replay the tracks of multiple data loggers and plot the positions for analysis of sailboat racing performance and tactics. However, it could be that the Velocitek SC1 uses the same Doppler speed technique discussed above. Perhaps the best way to discover if that is the case would be to contact the developers of the SC1. 
jimh 
posted 02152009 10:26 AM ET (US)
After several cups of coffee this morning, and quite a bit of reading, I have learned a lot about speed measurement using GPS technology. There are two techniques being used: measurement of track point distance and time measurement of GPS carrier Doppler frequency shift The track point method seems to have been the method generally in use with most marine GPS receivers. This method is an indirect measurement of speed. It deduces speed by calculating the distance between position fixes, then dividing that distance by the difference in time at those positions. This method seems to resolve to an accuracy of about 0.45MPH when the positions are deduced with precision enhancement such as from WAAS or differential GPS (DGPS) ground stations. The time information is highly accurate, but errors in the position will introduce errors in the speed. As we all know, the position data is only accurate to about three meters, so any speed deduced from these positions will have error. As far as I can tell, this is the method used in most marine GPS receivers to compute the boat speed. The Doppler frequency shift speed calculation is a direct measurement of speed. The L1 carrier frequency of the GPS satellite will appear to be shifted in frequency by the relative motion between the satellite and the receiver. Measurement of this shift is a direct measurement of speed. From what I can tell, speed measurement using this technique is a relatively recent development and may not be directly implemented in the GPS receiver chip itself. It seems that some newer GPS receiver chips can provide data output which gives a measurement of Doppler shift, and from this data, by using additional algorithms and software, a very accurate speed measurement can be obtained. There is a quite readable discussion of speed measurement using Doppler in http://www.gpsresults.com/GPS_Speed.pdf which concludes: "The error margins of handheldGPS measured speeds can be drastically improved by the use of Dopplerspeedbased NMEAdata as compared to the traditionally used positionbased evaluations. About one order of magnitude in accuracy can be gained...." Curiously, much of this interest in more accurate speed measurement seems to be driven by another marine use: wind surfing. There is software available which can be used to produce enhanced speed data from a GPS receiver which outputs the necessary Doppler information in its NMEA data stream. One such software package is RealSpeed: http://intellimass.com/RealSpeed/Index.htm which is available at modest cost (20euro or €20). I am not certain if this software works on real time data or only on recorded data. 
jimh 
posted 02152009 10:56 AM ET (US)
The GPS receiver from LOCOSYS which has replaced the GT11 (often mentioned in the articles and papers referenced above) is the Genie GT31. http://www.locosystech.com/product.php?zln=en&id=30 This receiver mentions specifically that it provides the Doppler data output necessary for enhanced speed calculation: "Major features include...support Doppler speed." Bundled with the GT31 is software for enhanced speed measurement: Valueadded Software bundled with GT31 RealSpeed (from www.intellimass.com): excellent software to get your real speed for your GPS tracks GPSResults (from www.gpsspeed.com) : powerful software to analyze your windsurfing data and report results to the GPS speeding community The GT31 has become popular with wind surfing enthusiasts who are apparently using it to measure their speed with greater precision than otherwise can be obtained with traditional GPS methods. 
jimh 
posted 02152009 11:57 AM ET (US)
According to some very recent discussion on a forum dedicated to highspeed wind surfing, which, again seems to the the group most uptodate on these GPS Doppler speed measurement techniques, the GT31 GPS receiver may be somewhat unique in its capabilities to provide Doppler information for speed measurement: "The GT31 has been specifically developed for speed accuracy. It is light years ahead of every other GPS in this area as well," says Andrew Daff, one of the participants from that forum. He also says, "It is possible that there is another GPS manufacturer that is inspired by the capabilities of the GT31 to try to build something to match it but I have not heard of it." http://www.gpsspeedsurfing.com/default.aspx?mnu=forum&forum=6&val=10928 
jimh 
posted 02152009 01:48 PM ET (US)
ASIDEUse of Other Doppler Techniques in Boat Speed Measurement Use of Doppler techniques in boat speed measurement has been applied to SONAR devices. Furuno manufactures a Doppler speed log which is able to provide very precise measurement of boat speed. This device is often used aboard very large vessels to help judge their speed when making up to a dock or other mooring. http://www.wmjmarine.com/ds30.html Furuno is quite proud of their work with their Doppler speed log, if judged by the price: $79,995. 
number9 
posted 02172009 07:13 PM ET (US)
Really appreciate the additional reading materials added. Will check them out when time permits. Jim, what brand and blend of coffee you drinking? Think I need to make some changes here. Since my Pavoni went down down a few years ago have resorted to the brewed stuff. 
20dauntless 
posted 02172009 10:58 PM ET (US)
I normally have a Raymarine C80 and Garmin Nuvi 500 with Bluechart's running at the same time in my CDory. The speeds given by each unit are always within .1 knots of each other. I'm not sure if this means anything but I thought I'd throw it into the discussion. 
Powered by: Ultimate Bulletin Board, Freeware Version 2000
Purchase our Licensed Version which adds many more features!
© Infopop Corporation (formerly Madrona Park, Inc.), 1998  2000.