How does the speedometer in a GPS work?

[ThomD]

My understanding is that the cell towers act as differential GPS? No?

Wikipedia tells me that many phones use "Assisted GPS", where the phone uses information delivered through the cell network to help it figure its position and speed up the initial position fix. It also says "Some A-GPS solutions do not have the option of falling back to standalone or autonomous GPS." That might explain why some phones don't have functioning GPS when there is no cell signal.

Differential GPS is something else completely. No idea if phones are using it.

 

[DaveInDenver]

Velocity is calculated using D-band (so the band from 1 to 2 GHz) Doppler shift from the GPS birds. Typically the accuracy limit is about 0.2 m/sec and most consumer chipsets are limited to a max of 999 MPH. You can actually measure your velocity in 3D, but non-military receivers usually don't and just give you a 2D speed.

 

[Pskhaat]

The expert speaks and puts us all up as he should! With your accuracy my in-my-head and likely very incorrect calculation at 1GHz puts like doppler delta at like...what...like 0.2nm?

 

[DaveInDenver]

The actual frequencies for L1 and L2 are 1.57542 GHz and 1.2276 GHz , so you can extrapolate carrier phase Doppler (L1 is typically used). A range-rate algorithm would sum the velocities for all visible vehicles (~7 km/s), the Earth's rotation (~400 m/s equatorial) and your relative motion (say around 25 m/s) to calculate Doppler shift. Typical PLL delta of about 15~20 degrees on a 19 cm wavelength translates into roughly a cm or so of delta, which ends up being for short Doppler intervals a very small shift in single satellite velocity. Yeah, for short samples you'll see in the cm/s of velocity error on each vehicle, but with longer measurements the accuracy gets better. Overall you will see in the tenths of m/s accuracy when measuring with >3 vehicles and allowing for noise.

 

[soonenough]

The actual frequencies for L1 and L2 are 1.57542 GHz and 1.2276 GHz , so you can extrapolate carrier phase Doppler (L1 is typically used). A range-rate algorithm would sum the velocities for all visible vehicles (~7 km/s), the Earth's rotation (~400 m/s equatorial) and your relative motion (say around 25 m/s) to calculate Doppler shift. Typical PLL delta of about 15~20 degrees on a 19 cm wavelength translates into roughly a cm or so of delta, which ends up being for short Doppler intervals a very small shift in single satellite velocity.

Holy crap, and I thought I was pretty technically saavy...

 

[soonenough]

(Sorry, double post)

 

[DaveInDenver]

Holy crap, and I thought I was pretty technically saavy...

Meh, there's a lot of info out there in the public domain about NAVSTAR. If you are interested in knowing more about the system, I think one of the best general coverage books on it is the compilation of papers Understanding GPS Principles and Applications. It doesn't pull punches but also doesn't spend 50 pages deriving extremely esoteric stuff. Couple that with a book on digital communications and filters (I recommend Prokais' book Digital Communications if you're an EE or math major, otherwise maybe Electronic Communications System by Tomasi as a start) and you'll understand enough about how it all works. Although, yeah, it sorta is rocket science. ;-)

 

[esh]

Now I am curious if my G1 (HTC, T-Mobile, Android stuff) actually has a hardware GPS device. Thought it would be easy enough to find a technical review of the hardware, but the best I've come up with other than "it has a GPS" has been wikipedia (taken with a grain of salt)- "The Dream includes a GPS receiver for fine-grained positioning, and can use cellular or wifi networks for coarse-grained positioning."

bleh