Also known as pseudorandom code. It's the long-string bitstream from a gps satellite that identifies the satellite to the ground receiver and is also used for the algebraic correction of the receivers internal clock. Simple in concept, as it works just like a sextant and chronograph for determining longitude, GPS is in actuality horribly complex. It's the way most things seem to be.
The way it works is this: Inside the receiver unit is an almanac of satellite positions. You power up the unit, and it initializes - i.e. takes a look at the incoming signals, and determines which satellites it's receiving from. It then estimates the distance to the satellite based on the difference between it's internal clock and the time code from the satellite. After that, it's triangulation. Where do the spheres described by the time difference intersect? It's very important that these streams not "match" internally. What I mean by that is that a sequence in the beginning of a satellite's stream shouldn't match a later sequence. This is so that atmospheric lag and signal bounce can't create false time readings by the sequence "auto correlating" with an earlier part of itself. This is where the pseudorandom stream comes in - the binary stream is totally random (low auto correlation), but completely repeatable. It's like recording audio static on a digital tape, and then playing it back. There is one channel of these signals for civilian receivers. Military units get 2, making them more accurate by something like a order of magnitude.
There is the added benefit that it just fucking sounds cool. It's not random, it's pseudorandom. It's like the difference between truly random white noise and the probabilistic distribution of sound energy in pink noise. "Pseudorandom" would make a good E2 handle.