Writing a chapter about geolocation history, I am digging the issue of GPS accuracy as it is often a "pain points" in the user (driver) experience. The Road Measurement Data Acquisition System has an interesting paper about it, by Chuck Gilbert.

Gilbert shows how complex the problem of GPS accuracy is and how misleading the advertisements are as they do not convey an intelligible vision of that topic. In general, the admitted accuracy (if there was such thing as admitted accuracy) is between 15 and 100 meters). But what does that range corresponds to? Is it achieve under optimal conditions? under difficult or extreme circumstances? The accuracy values is therefore represented statistically with different means but there is never enough room in an ad to depict this complexity; Gilbert finally recommends not not to use advertisements as an evaluation of GPS accuracy.

The factors that should be listed are the following:

"Required occupation time Type of data recorded (phase or pseudorange) Type of processing (phase or pseudorange) Environmental conditions Maximum allowable PDOP Minimum allowable signal strength Maximum allowable distance between base and rover receivers Horizontal accuracy versus vertical accuracy"

Why do I blog this? although I often focus on the environmental limitations (e.g. narrow streets), the situation is far more complex and it's interesting to pinpoint the different factors that can make a GPS device be inaccurate. How can design take this into account?