1993-12-28 - RE: GPS and security

Header Data

From: karn@qualcomm.com (Phil Karn)
To: pfarrell@netcom.com
Message Hash: 692cda9794da90e397b275b1230256ca3898125dc7b8f478532e6432b68c7a2b
Message ID: <199312282303.PAA21817@servo.qualcomm.com>
Reply To: <62594.pfarrell@netcom.com>
UTC Datetime: 1993-12-28 23:07:35 UTC
Raw Date: Tue, 28 Dec 93 15:07:35 PST

Raw message

From: karn@qualcomm.com (Phil Karn)
Date: Tue, 28 Dec 93 15:07:35 PST
To: pfarrell@netcom.com
Subject: RE: GPS and security
In-Reply-To: <62594.pfarrell@netcom.com>
Message-ID: <199312282303.PAA21817@servo.qualcomm.com>
MIME-Version: 1.0
Content-Type: text/plain

>Before the Gulf war started, the military needed zillions of GPS units, so
>they bought commercial marine navigrtion units. After the war, there
>was talk of turning on the division, but that was made
>impractical with 'differential' calibration.

You appear to be referring to Selective Availability. It was indeed turned
off during the Gulf War to allow the use of civilian C/A units. Companies
like Trimble couldn't keep up with the military demand during the buildup
to the war.

Selective availability was turned on again some months after the war
ended.  The effects were quite visible. The level of SA has varied
quite a bit since then.

>To get military accuracy, all you have to do is take the commercial GPS to
>known places. Such as the surveying marks at the US Naval Acadamy, or
>"ground zero" in the Pentagon. Get your "imprecise" reading, calculate the
>difference between it at the known value, and use it for subsequent

False. SA-induced errors are time varying, so the differential trick
only works in *real time*. You need receivers at both the reference
location and the location to be measured at the same time. The
experience with differential GPS is that you need updates that are
less than 30 seconds old, otherwise accuracy degrades rapidly.

By the way, the differential updates include both range offset and
range-rate offset (relative velocity). The addition of range-rate
makes the updates last as long as 30 seconds, otherwise you'd need a
very high update rate to get 2-3m accuracy.

>BTW, two satellites is enuff for a general location fix. Three identifies a
>place on the globe, four adds altitude (handy for planes, bombs, and

Also false. You need at least THREE satellites for any kind of GPS
position fix.  And you need to know your altitude ahead of time for
this to work.  Give a unit operating in 3 satellite mode the wrong
altitude, and it will give you the wrong latitude/longitude.  Four
satellites will give you a three-dimensional fix, i.e., the unit will
tell you your altitude.

Most units have an "auto 2D/3D" mode, where the unit remembers the
altitude from the last 3D (4 satellite) fix and uses it in 2D (3
satellite) mode.  Some other units, apparently including my Icom GP-22
unit, seem to assume you're at sea level whenever fewer than 4
satellites are visible. This may be a reasonable assumption for a unit
designed for marine use, but it is a bad assumption in places like
Colorado. When I took it there last summer, I often saw it give one
position as soon as it got 3 satellites, and then jump a half mile to
the correct position when it got the fourth.

About the only useful thing you can do with less than 3 visible
satellites is time transfer to a fixed station. A station in a known
location need only track one satellite in order to keep its clock
updated. At Qualcomm we use special GPS receivers (with built-in
rubidium clocks) in our CDMA base stations as time references to
synchronize our own spreading codes.