GPS Surveillance

The MPD placed GPS devices in several vehicles owned or rented by Scott Peterson as part of its surveillance.  Rudy Skultety was the officer in charge of collecting the tracking data. 

• January 3, 2003, Land Rover is returned to Scott with a tracking device installed.

• January 8, 2003, GPS device installed on a Chevy S-10 Sonoma rented from Enterpise Rental.  Removed January 10.  Tracked to the Berkeley Marina on January 9.  Tracking data is People's Exhibit 235A.  Also tracked to the San Luis Reservoir Forebay on January 9, Defense Exhibit D5Z.

• January 20, Scott tracked again to the San Luis Reservoir, tracking data Defense Exhibit D6G.

• January 21-23, GPS device installed in the Land Rover because previous one was malfunctioning.  January 26 tracked to the Berkeley Marina.  Tracking data is People's Exhibit 235B.

• January 27, 2003, GPS device installed on a Dodge Dakota rented from Enterprise Rental.  Removed January 29.  Tracking data for January 27 is People's Exhibit 236.

• February 18, 2003, Scott's new Dodge Dakota truck is seized in the search warrant, a tracking device was installed, and it was returned to Scott.  The device was removed April 24.

Pre-Trial Evidentiary Hearings on the Admissibility of the GPS evidence

Peter Loomis, Trimble Navigation   Testimony

• Defense objects to allowing the evidence at trial on the grounds the LE officers plan to invoke 1040, which involves concealing the location of the device.  Delucchi responded:  the location of this  device, wherever it is, may affect -- it may affect its efficacy and sending out the necessary signals, and so forth, and if he's denied access to the location of this particular device, it would essentially deny him his right to cross-examination and to require sanctions striking the officer's testimony.

• Peter Van Wick Loomis,  Fellow Staff Scientist, Trimble Navigation Component Technologies Division (People's expert witness.)  Loomis' area of expertise is grounded in the technology itself, as opposed to how it's translated into a particular device. Admittedly, he is not qualified to make a judgment on whether GPS is generally accepted in the scientific community for forensic purposes. 

• Trimble's sales will be adversely affected if GPS evidence not admitted into this trial.  "It makes all the difference in the world to him, obviously, that this technology gets sanctioned by this court."

• GPS technology is widely used by the military, aviation, shipping, agriculture, and auto navigation systems.

• Orion has a communication board inside the device that takes the information from Trimble's GPS receiver and packages it into data that's sent over the air in real-time.  This is called "over-the-air" download.

• Specific anomalies with GPS datasets produced in this case, which Loomis said he has never seen before:

• a jump of exactly one degree west for a period of about 6 minutes.  One degree represents 60 miles.  Loomis said he had never seen anything like this.  Orion said this happens with over-the-air downloads when the antenna crosses a line of latitude or longitude.  Loomis said this results from Orion's  method of compressing the data, sending it over the air, and decompressing it, which was outside his expertise.

• In an area just north of Fresno, the GPS actually was giving incorrect positions for a period of about 5 minutes.  Loomis said he had never seen this problem before, either. 

• Orion's explanation for the anomalies was the antenna being hidden, not in clear view of the sky.  Due to the covertness of this device, the GPS antenna often cannot be put in an optimal position for the GPS satellites to see 100 percent of the time.  In these particular instances, the antennas were not put in optimal positions for GPS reception.

• Errors can be generated by the geometry or angle of the satellite in the sky.  These errors are multiplicative, up to 10-15 times; i.e., a 1000-meter range of error can generate a spike of 10,000-15,000 meters.

• Satellites too close together will result in calculation errors,  as will satellites too far apart, or those about to disappear over the horizon--all of which also are multiplicative.

• Radio or radar interference can also cause multiplicative errors.

• Loomis disagreed with most of the reasons Orion gave for the anomalies in the datasets. For example, Orion said one anomaly resulted when the GPS device followed a satellite over the horizon, then acquired another one and started to correct itself.  Loomis says the problem resulted from radio interference, as this particular GPS device, the Lassen SK8 II, cannot follow a satellite over the horizon.

• The particular device used on Scott's vehicles had a behavior defect that was corrected in 1998/1999.

• The speeds calculated by the Orion software are inaccurate.  On indications of no motion, the software reported various speeds.  Loomis said Trimble produces a GPS speed that is much more accurate than what Orion delivered.

• The Fresno dataset had speeds that start off at 62 miles an hour, spike up to 2,164 miles per hour, then decrease back down to 654 miles per hour, then to 225 miles per hour--all with no-motion indicators.  Loomis said this illogical data is an Orion function.

• Another dataset recorded a speed of 38,000 mph.  This particular instance is an instance where it is just exactly one degree longitude. If Orion is calculating their speeds by a change in position over those five seconds, it went, oh, 60 miles or so over five seconds. And that very well might be 38,000 miles per hour.

• Loomis had never tested the Orion product himself and was not aware of any control test.  He did not do any tests on the Orion product after the anomalies were called to his attention, and does not know of anyone else who did.  Orion did maintain that they had tested to see if they could reproduce the latitude/longitude line problem.

• Modesto Police had actually done a real life test, and it had failed in a number of these controlled
  situations -- the device had failed. 

• The device used in Scott's vehicles did not include a dead-reckoning device (DR), which is an addition of a couple of sensors to calculate positions when the GPS is not available.

• Distaso's closing point:  "So of that hours of data, we're talking about one six-minute block, a couple second block, and a five-minute block where you did not believe the data was accurate?  For the remainder of the data, you were comfortable, and you testified here today, that that data, in your opinion, was accurate?"

• Geragos' closing point:  "How about, just as an expert who is sitting here testifying about the technology, who is saying that you disagree with a substantial part of the analysis that was done by your end user as to what the problems were, that you have been given only a discrete portion of information that apparently you were not told about the failure of one of these very same units on one of these cars, and that you have never seen one of these blips or unusual situations in ten years, and you didn't know, or apparently just discovered that there was a longitudinal blip that would cause a mile change in direction, when you are talking about getting those kinds of pieces of information, wouldn't you want more information rather than less?"

Hugh Roddis, President Orion Electronics Limited  Testimony

• Background

  • President, Orion Electronics Limited.

  • Orion’s unit consists of the GPS receiver, produced by Trimble, and the Orion part of the unit which takes the information out of the Trimble unit, stores it, processes it, and handles the other operations that go on.

    • The Orion unit used does not have a Wide Area Orientation System (WAOS) or Nationwide Differential GPS (NDGPS).

      • The increased accuracy of WAOS is necessary in aviation, but not in this kind of tracking.

      • NDGPS is no longer necessary because President Clinton discontinued selective availability, which produced a jitter or noise, which then caused substantial errors. NDGPS was used to correct those errors.

    • Vehicle speed is not determined by the speedometer, but is determined by latitude/ longitude – how long it takes to go from A to B.

    • The motion/no motion action is not derived from the GPS. It’s a vibration sensor in the unit itself. If a moving car is not experiencing any vibration (Cadillac-on-a-freeway effect, speed will be registered, but no motion.

  • By looking at the data generated by a unit, Roddis can tell if the antenna was properly placed, because of the consistency of the fixes.  However, Roddis was given only a portion of the tracks to review.

  • A covert placement of the antenna can result in Positional Dilution of Precision (PDOP), which simply means the device cannot compute the precise location to the GPS receiver due to bad geometry.  The unit does not record PDOP, but does it as part of the internal workings.

  • Roddis was not told that the Modesto Police Department did their own test on an Orion unit, using an unmarked police car.  The driver was instructed to go to three separate points.  The unit produced "sporadic communications" and there was some interference in the area which is the subject of one of the tracks of the Berkeley Marina used in this case (Exhibit 1).

• Problems with the unit installed on the Land Rover

  • Problem: the Orion brand GPS data logging system installed on the Land Rover was unable to maintain a track, or retrieve data information due to equipment problems, from January 3rd to January 21st.  There were two reasons:

    • a problem with the cell phone connection, which was resolved

    • one of the components, a microprocessor, had a bad connection between one of its pins on this computer chip and the rest of the circuitry, which caused the malfunction of the unit.

  • During the week of January 21st, 03, Scott returned the Land Rover to the Modesto Police Department to replace of the headliner that was partially removed during the search for evidence.

  • On Jan 24, the same unit in the Rover is producing data, but Roddis could not account for who repaired the unit in those 3 days, from the 21st to the 24th, when the unit was supposedly sent back to headquarters in Canada for repair.  During the break, he was told by Distaso that someone at the MPD decided to fix the problem.

  • Clarified that one of the problems was not that the antenna tracked a satellite over the horizon, but "it held on too long to a satellite, perhaps due to antenna placement, which just went over the horizon" (609:20-22).

• Specific problems with the datasets

  • Problem: one degree of offset occurred on the Jan 8 data for a period of 6 minutes.
    In this particular instance, the Trimble unit had a very -- had a firmware problem in it which only occurred when the unit crossed over from one line of latitude or longitude on the zero -- exactly 61 degrees. Say, for example, when it crossed over from 60.9999 to 61, just momentarily, the output would be wrong. Our unit interpreted that as one degree offset error. (574:14-22)

  • Two more instances of 1-degree offset in the datasets.

  • Spike in Fresno data, took 5 minutes to correct, most likely due to interference with the GPS receiver, as there is no other way to explain it. The spike "is like a flash of interference on a TV set.  You see a good program and a burst of interference, but the program before and afterwards is fine" (622:4-6).

  • For bitmap four, all of Orion's experts agreed that the portion of that track is not valid and "we do rarely see unexpected GPS positions" (635:5)

    • Orion suspects airport interference caused the problem.

•  The July 9, 2003 letter from Orion to Trimble

  • The analysis of the bitmap one track for the Dakota and the bitmap three track for the S-10 was only "partially correct." 

    • At that time, Orion believed "that when crossing the line of latitude or longitude, the GPS receiver occasionally does not update all decimal places of the longitude." (627:13-16)

    • Orion "later discovered through more experiments and more -- the problem with the Trimble receiver, which is actually triggering this issue.  It is a compression problem with the download, but it would have been triggered off by the Trimble issue that we didn't discover until after this letter was written." (627:21-628:1)

  • Orion explained The number two track for the Rover on January 3 as a problem with data from the Census Bureau for Modesto.  When Orion changed to a map with better data, the errors went away.

  • Orion stated that "the error in the data increases the further away you get from your chosen datum; towards the western edge of the county, the error is much less." (633:20-23).

Delucchi heard arguments on GPS and then ruled to admit the evidence.  Arguments & Ruling on GPS

Evidence presented at Trial

Peter Loomis and Hugh Roddis again testified, along with Rudy Skultety, on the integrity of the GPS tracking data that places Scott at the Berkeley Marina on several days during the investigation. Geragos did not challenge the integrity of the data, but rather spent a great deal of time showing tracking data that puts Scott at other reservoirs and bodies of water that were being searched for Laci.  Skultety's testimony also brought out how many times Scott was in San Diego, or in the area, during January to early April, seemingly to undermine the Prosecution's claim that Scott was a flight risk when he was arrested in San Diego.