While most of our customers never see effects from any of the detail below some people enjoy some of the technical attributes of GPS accuracy and geocoding.
How accurate is GPS?
The accuracy rating of most consumer GPS devices is typically expressed as “2 meters Circular Error Probability (CEP)”. This is a fancy way of saying that the device is accurate to within 6 feet about half the time. It also means GPS accuracy to 25 feet 95% of the time.
No measuring device is ever 100% accurate all the time. Taking many measurements over several minutes provides an accuracy rate that gets very close to 100%. This proceeds by an ever-increasing number of decimal places (e.g. 99.99998%).
Like a car in the winter, GPS receivers need to “warm-up”
GPS receivers take several minutes to “warm up” after being powered on so that they can begin receiving satellite timing signals and a kind of almanac of the GPS satellite network in orbit around the Earth. This almanac is called “ephemeris data” and it tells the device which satellites it can expect to see at a certain time of day. Once this information is updated within the device, it can get on with the business of locating itself.
GPS receivers can tell how well they’re doing at their job
The GPS receiver has an internal measurement called “Horizontal Dilution of Precision” (HDOP) which it uses to rate its position accuracy. If HDOP drops below a pre-established threshold for longer than about a minute, it will report its position with 0.0 values for latitude and longitude.
Once the receiver can view 5 or more satellites again, the HDOP rises above the threshold and the device reports highly accurate positions once more. This behavior allows no more than one or two position reports to be sent by the GPS receiver with inaccurate positions before the reported position “goes to zeros”.
Thus, when taken in context with position reports before and after any other single position report, it is very simple to judge whether the accuracy of a single position report is suspect. In all cases, GPS accuracy varies no more than up to one half mile for a minute before going to zeros or snapping back to the “true” position.
While moving down the road, GPS accuracy (horizontal) is very good: generally within 25 feet – but often accurate to within 10 feet. GPS speed determination is also very good while moving: within 1 MPH. Conversely GPS vertical accuracy is poor, sometimes the error is hundreds of feet in elevation/altitude.
Some special cases that are challenging to GPS receivers
GPS speed determination is less accurate while stationary: sometimes showing speeds of 1 to 3 MPH when speed values should be 0. This is due to inaccuracies being summed together in an imaginary straight line, which yields a movement of a mile or two per hour. However, the inaccuracies are scattered around the true point in all directions and cancel each other out over time as shown in the illustration on the left.
There is another phenomenon which typically occurs whenever the “view” to 5 or more satellites is obstructed: GPS Drift. As satellites cross into regions of the sky that may be obstructed from view of the device, the location “fix” by the device may jump around until a solid fix is re-established.
This radical jump may cause the device to believe it has traveled further than it has and generate reported speeds of 10 MPH to 100 MPH and more. We tend to see this mostly with vehicles parked inside of garages, under trees with heavy foliage, or when parked next to a building obstructing the view to the sky.
The device is aware of these events through HDOP and HoloTrak takes steps to not report Geofence or Excessive Speed events that would otherwise be falsely reported. We typically add filtering for such devices on a case by case basis for our customers who have trouble with this for a specific vehicle they’re tracking that is frequently parked overnight in a location with poor GPS reception.
Reverse Geocoding is an inexact science
The “Location” values provided by the HoloTrak software are derived from a process known as “reverse geocoding”. Reverse geocoding converts a Latitude and Longitude pair into a Street Address or other descriptive value that aids the viewer in knowing the approximate location of the device such as “1313 Mockingbird Ln, Dallas, TX”.
HoloTrak does not guarantee Location values are always accurate because even the best map makers aren’t sure exactly where the boundaries of a specific address are.
The method HoloTrak uses for reverse geocoding attempts to find the nearest known place name in a database of place names, but often can’t find an exact street address. When this occurs, the HoloTrak software will revert to a road or highway name, or sometimes just a county and State name. As such, “Location” values should always be interpreted as “in the vicinity of”.