Intuition The one core idea
A GPS receiver listens to radio "time stamps" from satellites in the sky and works out how far each one is by measuring how long the signal took to arrive. Because its own clock is cheap and wrong by an unknown amount, it must solve for where it is and how wrong its clock is at the same time — and that single extra unknown is why the whole subject exists.
This page is the toolbox. Before you read the parent topic , every letter and squiggle it uses should already feel like an old friend. We build each one from nothing, anchor it to a picture, and say exactly why GPS needs it.
Definition A point in 3D space
( x , y , z )
Three numbers that pin down one exact spot . Imagine a giant invisible box drawn around the Earth with three rulers meeting at a corner: how far East (x ), how far North (y ), how far Up (z ). Any location — your phone, a satellite — is just "how far along each ruler."
Intuition Why GPS needs it
The whole job of GPS is to output one such triple: your ( x , y , z ) . Everything else in the topic is machinery to compute these three numbers. GPS actually uses a specific ruler set fixed to the spinning Earth — see Reference Frames — ECEF and WGS84 — but for now just picture "three perpendicular rulers."
Distance-from-a-corner A point ( x , y , z ) is that many units along each of three perpendicular axes.
We know the satellite's spot ( x i , y i , z i ) (it broadcasts it). We want to know its distance from our unknown spot ( x , y , z ) . How do you turn two triples into one distance?
Intuition WHY a square root of squares — this is just Pythagoras twice
On flat paper, the straight-line distance between two dots is ( across ) 2 + ( up ) 2 — the hypotenuse of a right triangle. In 3D you do it once in the floor plane, then once more going up. The square root is the tool that turns "side lengths" back into a "straight-line length." Squaring first makes every gap positive (a gap of − 3 and + 3 are the same distance), and the root undoes the squaring to give real metres.
Definition The symbols here
r i = the true geometric distance ("range") from satellite i to us. The little i is just a label : satellite 1, 2, 3, 4...
x i , y i , z i = the satellite's known position.
x , y , z = our unknown position.
Why the subscript i It is an index labelling which satellite; r 3 means "distance to satellite 3."
See Time of Flight and Ranging for how a distance is physically measured rather than computed.
c , the speed of light
c ≈ 3 × 1 0 8 metres per second — how fast radio waves (and light) travel. It is a fixed constant of nature.
Intuition The picture: light as a tape measure made of time
A radio pulse leaves the satellite and races toward you at c . If you know it travelled for a time t , then it covered a distance c × t — exactly like "60 km/h for 2 hours = 120 km." So c is the exchange rate between seconds and metres. GPS never measures distance with a ruler; it measures time and multiplies by c .
Worked example How big is a tiny time error?
One microsecond is 1 0 − 6 s. Multiply by c : ( 3 × 1 0 8 ) ( 1 0 − 6 ) = 300 m. So a clock wrong by a millionth of a second misplaces you by the length of three football fields. This one line is the reason the clock bias matters so much.
Definition Transmit and receive times
t t x = the moment the satellite sent the signal, stamped by its atomic clock (extremely accurate — see Clock Bias and Atomic Clocks ).
t r x = the moment we received it, read off our cheap receiver clock.
The subscripts t x and r x are radio shorthand for t ransm it and r ec eive.
Think of a runner leaving a start line at a stopwatch reading t t x and crossing your finish line at your stopwatch reading t r x . The race lasted t r x − t t x , and the track length is c ( t r x − t t x ) . If your finish-line stopwatch is set wrong, every race you time comes out wrong by the same amount.
b
The clock bias b is how far off the receiver's clock is from true GPS time, measured in seconds . If b is positive, the clock reads ahead ; if negative, behind .
Intuition Why one number corrupts everything — and why that is a gift
You have one receiver clock. Whatever it is fast or slow by, it shifts every t r x by the same b . So every distance you compute is off by the same c b . Because it is the same unknown appearing in every equation, one extra satellite gives you enough information to pin it down. A shared error is a solvable error.
Why is the clock error identical for all satellites There is only one receiver clock, so its offset b contaminates every measurement equally.
ρ i = r i + c b
The pseudorange is the true distance r i plus the clock-caused error c b . The Greek letter ρ ("rho") is the standard symbol; the "pseudo-" prefix means "looks like a range but isn't the true one yet."
Imagine drawing a sphere around each satellite with radius equal to the pseudorange. Because of c b , each sphere is slightly too big (or too small) by the same amount. The true spheres would all meet at your position; the pseudo-spheres miss — until you find the b that shrinks them all back into agreement.
Common mistake "Pseudorange = distance to the satellite"
It feels right because it is computed as c × (time). But it still carries the clock error c b . It only becomes a true range r i after b is solved and removed.
Intuition One range = one sphere
If you only know you are 20,000 km from satellite 1, you could be anywhere on a huge sphere of that radius around it. Add satellite 2: you must be on both spheres → their overlap is a circle . Add satellite 3: three spheres meet at (generally) two points , and the Earth throws one away. That is classic trilateration .
Intuition Why GPS needs a
fourth sphere
Three spheres pin a point if the radii are exact. But our radii are pseudoranges — all wrong by c b . The fourth satellite provides the fourth equation whose only job is to compute b , un-inflate all four spheres, and let them meet cleanly. Geometry says 3; the physics of clocks says 4. More at Trilateration and Multilateration .
Two spheres intersect in a... circle. Three (with exact radii) in two points.
Definition Unknowns vs equations
Unknowns: x , y , z , b — four numbers we must find.
Equations: one pseudorange per satellite. To solve for 4 unknowns you need at least 4 equations, hence at least 4 satellites.
Mnemonic Count the question marks
Every "?" needs its own equation. Three space coordinates plus one clock = four question marks = four satellites minimum.
Definition A vector and a unit vector
A vector is an arrow: it has a direction and a length. A unit vector is an arrow squashed to length exactly 1 — it keeps only the direction , throwing away the length.
Intuition The line-of-sight unit vector
( e i x , e i y , e i z )
Point an arrow from satellite i straight at the receiver and shrink it to length 1. Its three components e i x , e i y , e i z answer "which way is the satellite, as seen from me?" The parent topic shows these are exactly r i x − x i etc. — they appear because the rate of change of distance as you nudge your position is just "how much of that nudge points along the line to the satellite." Direction, not distance, is what controls the geometry.
Why the line-of-sight vector has length 1 We only care about direction ; dividing by r i normalises the length to one.
Definition A matrix as a grid
A matrix is a rectangular grid of numbers, written in a big bracket. Each row here is one satellite's line-of-sight direction plus a 1 for the clock:
G = e 1 x e 2 x ⋮ e n x e 1 y e 2 y e n y e 1 z e 2 z e n z 1 1 ⋮ 1
Definition The symbols you'll meet with
G
G ⊤ ("G transpose") = flip the grid over its diagonal, turning rows into columns.
G ⊤ G = a small square grid summarising how the satellite directions relate.
( G ⊤ G ) − 1 = the inverse — the matrix "undo" operation, like dividing.
Δ (Greek "delta") = a small change in : Δ x = tiny nudge in x , Δ ρ = tiny mismatch in pseudorange.
Intuition Why a matrix at all?
With four satellites you have four intertwined equations. A matrix is just the bookkeeping that solves all four at once . When there are more than four satellites, the system is over-determined and you fit a best compromise using Least Squares Estimation ; the ( G ⊤ G ) − 1 G ⊤ combination is precisely that best-fit machine.
What does the column of 1's in G represent The clock-bias direction — every satellite feels b equally.
Definition Randomness of measurements
Noise = the small random wobble in every real measurement.
σ (Greek "sigma") = standard deviation , the typical size of that wobble.
σ UERE = the typical error in a single pseudorange (User-Equivalent Range Error).
Covariance = a table describing how the errors in several numbers spread and lean together. Propagating it through the solution is done with Covariance Propagation .
Intuition The picture: a fuzzy dot, not a sharp point
Because of noise, your computed position isn't a crisp dot but a small fuzzy cloud. σ is the radius of that fuzz. DOP (Dilution of Precision) is the number that says how much the satellite geometry stretches that fuzz — good geometry keeps it a tight round blob, bad geometry smears it into a long streak.
Which factor is pure geometry, DOP or σ UERE DOP — it depends only on satellite directions, not on measurement quality.
Transmit and receive times
Four unknowns need four satellites
Spheres and intersections
Unit line of sight vectors
Noise sigma and covariance
Everything downstream — the linearised solve, the DOP formula, the Kalman Filter in GNC that later fuses these fixes over time — rests on the twelve pieces above.
Can you state, from memory, what each symbol means and why GPS needs it? Reveal to check.
( x , y , z ) The receiver's unknown position — three perpendicular ruler-distances; the final answer GPS outputs.
r i True geometric distance to satellite i , via 3D Pythagoras (the square root of summed squared axis-gaps).
c Speed of light, 3 × 1 0 8 m/s — the exchange rate converting measured time into distance .
t t x , t r x Transmit time (satellite's atomic clock) and receive time (our cheap clock); their difference times c is a range.
b Receiver clock bias in seconds — one shared unknown that corrupts every measurement by the same c b .
ρ i Pseudorange = true range r i plus c b ; a "looks-like-a-range" that still contains the clock error.
Why 4 satellites, not 3 Four unknowns x , y , z , b need four equations; the 4th satellite's job is to solve the clock bias.
Sphere picture Each range is a sphere around a satellite; their intersection is your location.
( e i x , e i y , e i z ) Unit line-of-sight vector — pure direction from satellite to receiver, length 1.
G and G ⊤ G , ( G ⊤ G ) − 1 Geometry matrix of directions (+clock column); its transpose-product and inverse form the least-squares solver.
Δ "A small change in" — used for nudges Δ x and pseudorange mismatches Δ ρ .
σ UERE and DOPSingle-measurement error size, and the geometry-only amplification factor; multiplied they give position error.