`Selective
`Availability
`
`Reprintedlrom G” WQRLD September/OdobeerO
`
`Yola Georgiadou
`Kenneth D. Doucet
`
`|PR2020-00408
`
`Apple EX1018 Page 1
`
`IPR2020-00408
`Apple EX1018 Page 1
`
`
`
`INNOVATION
`
`
`
`“Innovation” is a regular column in GPS World
`commenting on GPS technology, product
`development, and other issues and needs of the
`GPS community. This time we look at selective
`availability and assess its effect on GPS
`measurements.
`
`This column is coordinated by Richard Langley
`and Alfred Kleusberg of the Department of
`Surveying Engineering at the University of New
`Brunswick. We welcome your comments and
`suggestions for future columns.
`
`Selective availability (SA) is a method of de-
`nying unauthorized GPS users high position
`accuracy. Although it stands for “accuracy de-
`nial," the term was probably introduced be-
`cause it sounds more positive and less
`ominous. The United States Department of
`Defense (DOD) formally implemented SA on
`March 25, 1990 and put an end to “a long pe-
`riod of uncertainty, apprehension and heated
`debate,” as the May/June 1990 issue Of GPS
`World put
`it. Nominal
`three-dimensional
`point position accuracy of the Standard PO—
`sitioning Service (SPS) with SA switched on
`is l20 meters. assuming that the complete sat—
`ellite constellation is in place. A more than
`threefold decrease in accuracy for SP8 (the
`accuracy of SPS without SA is in the 20- to
`40-meter range) is a‘ serious concern.
`In this column we shall review the reasons
`
`that led to withholding the full system accu—
`racy from most civilian users, discuss the im-
`plementation of SA, and examine its effect
`on our measurements and positions. Finally,
`we shall discuss a technique known as differ—
`ential GPS, which appears capable of reduc-
`ing the effect of SA errors for a large variety
`of users.
`
`The Issue of
`Selective
`Availability
`
`Yola Georgiadou
`Kenneth D. Doucet
`
`Geodetic Research Laboratory,
`University of New Brunswick
`
`HISTORY
`
`The issue of selective availability is closely
`tied to the history of GPS development. Af—
`ter the NAVSTAR GPS concept was born in
`1973, a Joint Program Office established by
`the DOD was assigned the task of developing
`and deploying a single system that could
`serve defense positioning and navigation
`needs. Personnel from the Army, Navy, Ma-
`rine Corps, Air Force and Defense Mapping
`Agency were to cooperate towards this goal.
`Extensive testing of user equipment took
`place at this early stage. Ground-based trans-
`mitters and later, Block I prototype satellites,
`provided the necessary navigation signals for
`these initial tests. The idea was that sophis-
`ticated high-precision equipment, utilizing
`the precision or P-eode, would be later re-
`served for military use. Low-cost units oper—
`ating with the less precise coarse/acquisition
`0r C/A-code were intended to be generally
`available to everyone. This is roughly how
`the full system accuracy was to be reserved
`for authorized (military) users.
`The test results were both encouraging and
`surprising. System accuracies at the 10- to
`20-meter level achieved With P—code receiv—
`
`ers lent necessary momentum to the program
`as it entered the full-scale engineering devel-
`opment phase. And there was a big surprise!
`The low cost C/A-code unit proved to be
`much better than expected. Although it was
`predicted to provide position accuracies of
`no better than 100 meters, its actual perform-
`ance was at the 20- to 30—meter level!
`
`This turn Of events marked the beginning
`of a remarkable era. The emergence of the
`C/A—code unit as a precise navigational tool
`was quickly followed by a rethinking Of the
`strategy for high-accuracy availability. The
`DOD invited the Office Of the Joint Chiefs of
`Staff, the Office of the Secretary Of Defense,
`and the National Security Council to estab—
`lish a national policy regarding availability of
`
`GPS to the general public. At this time the
`positive sounding term selective availability
`was coined, Initially, a decision was reached
`to intentionally degrade the accuracy avail-
`able to unauthorized users to 500 meters by
`implementing SA on Block II satellites. That
`figure was later revised to 100 meters. The
`DOD also stated that this policy Of intentional
`degradation would be reviewed annually in
`an effort to increase the accuracy available as
`conditions warranted.
`As soon as Block 11 satellites were avail-
`able for tracking, some research groups
`started performing tests to assess the degree
`of error introduced in measurements and po—
`sition due to SA. Guesses as to the SA ef-
`fect on position became quite fashionable.
`Now that SA has been switched on Offi-
`cially, this period of uncertainty and specu-
`lation is over. All Block II satellites launched
`in the future will have SA enabled as soon
`as each is declared operational. Currently no
`plans exist for implementing SA on Block I
`satellites, but the DOD has reserved the right
`to do so.
`
`According to the current DOD “SA pol-
`icy,” the nominal position accuracy for hori-
`zontal coordinates (latitude and longitude) is
`100 meters at a probability level of 95 per-
`cent. This means that position errors should
`be less than 100 meters at least 95 percent Of
`the time. However, DOD has not indicated
`how large these errors may become during
`the remaining five percent of the time. An in—
`teragency U.S. working group that includes
`DOD representation has proposed a guarantee
`of position accuracy of less than 300 meters
`for 99.9 percent Of
`the remaining five
`percent. That proposal is contained in the cur-
`rent draft of the 1990 Federal Radionaviga-
`tion Plan. (See Washington View column else-
`where in this issue.)
`for
`The corresponding accuracy level
`heights is 150 meters. If we consider horizon—
`tal and vertical coordinates together as a
`three~dimcnsional position, it is 120 meters.
`In velocity determination, errors of the order
`Of 0.3 meter/second are anticipated. And for
`the time—transfer error in GPS time dissemi-
`nation, we may find it is about 300—400
`nanoseconds. considering that a position er-
`ror of one meter translates to a time error of
`about three nanoseconds.
`
`It should be kept in mind that these nomi-
`nal values will be relevant when the full sat—
`ellite constellation is Operational with a dilu-
`tion of precision (DOP) factor Of three to
`five (see the Innovation column in the GPS
`World March/April 1990 issue for a discus—
`sion of DOP). During the buildup to the full
`
`|PR2020-00408
`
`Apple EX1018 Page 2
`
`IPR2020-00408
`Apple EX1018 Page 2
`
`
`
`INNOVATION
`__————_—_-——_——_—_——-———_——-——I
`
`
`
`
`
`constellation. accuracy at times could be
`worse.
`Most GPS users will have to' live with
`these levels of accuracy in position. velocity,
`and time determination. Only authorized us-A
`ers may recover the undegraded data and ex—
`ploit the full system potential. To do so, they
`must possess a key that allows them to de-
`crypt correction data transmitted in the navi~
`gation message.
`in addition to denial of accuracy through
`SA. the DOD can restrict the use of signals
`from the operational Block 11 satellites
`through the encryption of
`the P-ncode.
`Termed anti-spoofing (A~S),
`this technique
`is a method of protecting military operations
`against hostile imitation of the P—code. It is
`widely believed that A—S will only be turned
`on in times of national emergency and for test“
`ing purposes.
`
`IMPLEMENTATION
`
`The way the DoD has implemented selective
`availability reflects the nature of positioning
`with GPS through the fundamental relation:
`
`Psendorange Measurement +
`Satellite Position 2 User Position
`
`The pseudorange measurement is derived
`from a comparison between satellite and re-
`ceiver clocks. SA introduces errors that have
`
`both rapidly and slowly varying components
`into the pseudoranges by manipulating (dith-
`ering) the satellite clock, the so~ca||ed 8—proc—
`ess. Pseudorange errors multiplied by the
`DOP factor are mapped directly into user po-
`sition errors.
`
`The satellite position is extracted from or—
`bit information in the broadcast message.
`Deliberate introduction of errors into the
`broadcast orbital parameters. known as the
`e-~process, leads to a degradation of accuracy
`of the orbital information broadcast by a sat-
`ellite. These orbital errors are expected to ex—
`hibit only slow variations to conform with
`the parametric representation of the orbit in
`the broadcast message. These orbit errors
`transform to user position errors with the
`same slowly varying nature. In most appli-
`cations, the user cannot distinguish between
`thc two SA constituents because both have
`
`slowly varying components.
`
`5A EFFECTS
`
`Basically, a GPS receiver can make two
`kinds of measurements: pseudoranges on one
`of the two codes, and carrier phase measure—
`ments. Both types of measurement reflect
`the distance between satellite and receiver.
`
`without S/A
`
`V1.5:V
`
`0
`
`minutes
`
`40
`
`
`
`Figure 1: C/A-code range measure-
`ment errors with and without SA
`
`Both are affected by SA in the same fashion
`because the same atomic clock on board the
`
`satellite controls the timing and frequency for
`the carriers and the two codes.
`Now let’s take a closer look at measure—
`ments obtained with a C/A~code receiver in
`
`an “SA environment.” We shall compare
`these with simultaneous measurements on a
`satellite free of SA errors. And, most impor~
`tantly, we shall look at the resulting position
`accuracy with and without SA.
`Figure 1 illustrates the potential magnitude
`of SA-induced errors on pseudorange mea—
`surements from GPS satellite PRN 14 (with
`SA enabled). The 40—minute data set was col-
`lected on May 16, 1990, at a known loca—
`tion. The measurement error was computed
`by differencing the observed and theoretical
`ranges. For PRN 14, the differences were of
`the order of 1-: 30 meters. This error level in
`the pseudoranges is consistent with the behav—
`ior expected of SA-affeeted measurements.
`On the other hand, measurements made simul-
`taneously to satellite PRN 9 (SA not en—
`abled) have errors of the order of four
`meters, a level of accuracy conforming to po-
`tential C/Afcode measurement accuracy.
`Figure 2 illustrates the magnitude of GPS
`position errors from C/A--code measurements
`collected at a known location. The top por-
`tion of the plot shows horizontal position er—
`rors for a total of 690 individual position
`fixes. The data was collected on February
`22, 1990, with SA inactive. Position errors
`were at the a: 15 meter level, as anticipated
`from undegradcd C/A—code measurements.
`The lower plot shows horizontal position er-
`rors for a total of 160 individual fixes for the
`same location about two months later (May
`3, 1990) after the official implementation of
`
`100
`
`meters
`
`—100
`
`100
`
`meters
`
`without S/A
`
`' with S/A
`
`
`
`
`
`»-100
`7100
`
`meters
`
`100
`
`Figure 2: C/A-code position errors
`with and without SA
`
`SA. The dramatic decrease in horizontal
`
`positioning accuracy to 1' 100 meters is con-
`sistent with the DoD policy on selective
`availability.
`
`CAN WE LIVE WITH SA?
`
`Several GPS user groups can live quite hap-
`pily with SA-dcgradcd accuracies. However,
`perhaps just as many users need access to the
`full system accuracy for high—precision appli~
`cations. Differential GPS, recognized very
`early in the life of the system as a remedy
`for SA errors, comes to the rescue of these
`high-precision users. This technique can
`counter both clock dithering and orbital data
`manipulation. Moreover, it is not considered
`to pose a security hazard.
`Differential GPS operation involves plac-
`ing a GPS receiver at one or more known
`locations — the monitor stations ~ and com—
`puting corrections for the range errors ob-
`served on the satellite signals. In principle,
`these corrections are calculated by subtract—
`ing the observed pseudoranges from the
`ranges computed with the known station co-
`ordinates, and then are broadcast through an
`appropriate data communications link to the
`user. Because errors generated by SA affect
`measurements at the monitor station and at
`the user’s site in a very similar fashion, the
`
`|PR2020-00408
`
`Apple EX1018 Page 3
`
`IPR2020-00408
`Apple EX1018 Page 3
`
`
`
`
`
`INNOVATION_———————————-—_—_———_
`
`
`
`user can correct erroneous measurements
`with the received differential corrections.
`This procedure eliminates or greatly reduces
`the impact of SA on positioning, as dis—
`cussed in greater detail in the July/August
`1990 Innovation column.
`
`This scheme seems to be an easy way to
`recover the full GPS positioning accuracy.
`But it is not a perfect cure. In general, the
`capacity of the data communications link
`will limit the update rate for the transmitted
`differential corrections. If the errors intro-
`
`duced by SA change faster than the differen-
`tial correction update rate, remote users can
`no longer accurately correct their measure-
`ments. Therefore,
`the degree of error re—
`moval will depend on the rate of change of
`the SA errors in relation to the data commu—
`nications rate. The data in Figure 1 indicates
`a range rate error of about 0.2m/sec. Assum-
`ing this level of error and a data communi~
`cations rate of one range correction message
`per 10 seconds,
`the differential corrections
`would be in error by up to two meters after
`10 seconds.
`Another factor in the effectiveness of the
`differential corrections is the distance be-
`tween monitor station and user. Pseudorange
`errors introduced through changes in the or—
`bital parameters of the satellites are similar
`only if the monitor station and the user are
`in the same geographical region. As distance
`between the stations increases, the correspond-
`ing range errors tend to become different.
`An error of 40 meters in the satellite orbit
`with the user 1,000 kilometers away from
`the monitor station, for example, yields a dif—
`ferential correction that is in error by up to
`two meters.
`
`Let us sum up these arguments. Differen-
`tial GPS may be the only remedy for SA er—
`rors under two conditions: a sufficiently high
`rate of differential corrections, and short dis-
`tances between the monitor station and the
`moving users.
`Surveying can be seen as a special appli—
`cation of differential GPS. Surveyors use
`GPS to measure position differences between
`survey markers. Among others, they were
`quick to realize the potential of measurement
`differencing for error cancellation, and have
`used it through the last decade to extract un—
`precedented and unexpected accuracies for
`static differential positioning 7 centimeters
`
`01
`
`meters
`
`meters
`
`5
`
`
`
`
`
`A technique similar in principle to differ-
`ential GPS is used to reduce considerably the
`SA effects on time transfer. This concept.
`known as common-view, common-mode
`time transfer, is based on the assumption] that
`measurements collected by the two receivers
`involved in time transfer are contaminated by
`similar SA errors. Corrections transmitted
`from the reference receiver to the remote re—
`
`ceiver reportedly can remove most of the SA
`errors at the remote receiver site. More infor—
`
`mation about time and frequency dissemina-
`tion using GPS and SA—related errors can be
`found in the July/August 1990 issue of GPS
`World (Peter H. Dana and Bruce M. Penrod,
`The Role of GPS in Precise Time and Fre—
`
`quency Dissemination).
`
`
`
`Figure 3: Differential GPS C/A-code po-
`sltion errors with SA
`
`CONCLUSION
`
`over tens of kilometers. In surveying, the
`measurements used are carrier phases be-
`cause inherently they have a higher precision
`than pseudoranges. (See Innovation column
`in the January/February 1990 issue of GPS
`World for further details.) No differential cor-
`rections need be computed and transmitted in
`this case. Instead, the measurement data are
`collected and stored at both sites for postpro-
`cessing. This data processing involves direct
`differencing of measurements between moni—
`tor and user station, which eliminates SA er—
`rors caused by the clock dithering and
`greatly reduces the effect of orbital errors on
`differential positioning. In fact, surveyors
`had to use this differencing technique even be-
`fore SA was introduced to achieve the posi-
`tioning accuracies required in surveying.
`Figure 3 illustrates the potential success of
`differential GPS in eliminating the effects of
`SA. The same SA-contaminated data that
`was used for Figure 2 was analyzed to pro-
`duce the differential position errors shown in
`Figure 3. For these calculations, though, the
`measurements were combined with measure—
`ments from a second receiver at another
`known site. The effect of applying differen—
`tial corrections was simulated by a direct dif-
`ferencing of the C/A-eode pseudorangcs ob-
`served at the two sites, as is routinely done
`with carrier phases in GPS surveying. The re—
`sulting position errors are of the order of i 5
`meters, an increase in accuracy by a factor
`of 20 compared to the undifferenced result of
`Figure 2 (lower plot).
`
`Selective availability is now officially in use.
`and apparently it is here to stay. Although
`many low—precision users can live with it.
`SA is a major step backward from what has
`been the biggest breakthrough in navigation
`technology. We have shown how large typi~
`cal SA errors in GPS measurements and
`
`fixes are during 95 percent of the time,
`based on a small data set. The problem is
`that at times they could be much larger! In
`other words, we currently lack a firm and fi-
`nal commitment on the size of errors for five
`percent of the time. The necessity for a more
`precise definition of the impact of SA has al-
`ready been recognized by the DoD. As a re-
`sult, DoD and Department of Transportation
`officials are expected to spell out more pre-
`cisely the accuracy limits of the SPS in the
`final version of the 1990 Federal Radionavi—
`gation Plan.
`the present time is
`One thing is certain:
`too soon to assess the full impact of SA on
`position, velocity, time, and frequency trans-
`fer for the fully operational system. Instead,
`perhaps we should concentrate 011 exploring
`exciting new ideas such as the combination
`of GPS with the INMARSAT and Soviet
`
`GLONASS systems for integrity control and
`differential information. Development and re-
`alization of these and similar concepts, inher-
`ently based on cooperation among nations,
`may render obsolete the reasons that led to
`the introduction of selective availability in
`the first place. I
`
`|PR2020-00408
`
`Apple EX1018 Page 4
`
`IPR2020-00408
`Apple EX1018 Page 4
`
`