Friday, February 27, 2009

Its All in the Timing. . .

The GPS system, consisting of the constellation of satellites and the receivers that use them, are most often lauded for their ability to accurately locate and record location. However, just as important is the GPS's ability to both accurately receive, display and record the timing of those geographic events. GPS time is so accurate, scientists in indoor labs often mount rooftop antennas to capture the GPS time signals (with GPS receivers) and use them for their experiments-even though the GPS receiver never moves an inch!

Why are GPS time signals so accurate? Because each satellite has an atomic clock, and a central atomic clock continuously co-ordinates the constellation of satellites so these clocks are within a few nanoseconds of each other. A GPS receiver that is on but not "locked" to the satellites (indoors or with very bad signals) is about as accurate as a new quartz digital watch, but a GPS that is locked to the satellites is as accurate as an atomic clock, because its time is being updated every few seconds. Now, although a new digital quartz watch may loose a few seconds or even a minute a month, how much time would it loose in a day? An hour? A minute? Since the GPS is constantly updated by the satellite atomic clock, it is, for all human measurable perceptions, as accurate as the satellite atomic clock.

We can accept that GPS timing is dead accurate. Not just close. Dead accurate. Why? Because of the way GPS works, even nanoseconds of error, which might show up as a large locational error, are tiny compared to our perception of time and events. If a GPS receiver has any lock on any satellite (required to record any position at all) the time it reports is as accurate as any human need can generate. If a GPS cannot get sufficient signal for a lock, it won't record a position.

What are the implications of a totally accurate time-keeping system that 1) can automatically record events and 2) can be accessed by any GPS receiver?
  1. Proactively, it means that diverse members of a unit can act in unison at a given moment, without the necessity to "synchronize their watches." Search warrants can be executed at 20 different locations at exactly 5:30 p.m. local time, thus denying members of a conspiracy a chance to warn other members.
  2. Retroactively, GPS track records can be plotted against each other based on time, and recorded events can be compared to see who was where, when (and what they were doing, according to their own narrative testimony or description.)
  3. GPS track records can be plotted against other time systems to determine actions. For instance, if a GPS track of 3:22 p.m. indicates that driver A (fatally injured) was going 60 m.p.h., and cell phone records indicate at 3:22 p.m. driver A received and read a text message, it can be inferred driver A was watching his phone, and not the road, before his fatal accident with a Semi-Tractor-Trailer, thus confirming the driver B's testimony.
    • Such systems can be compared after the fact to determine their accuracy . For instance, one could call the same type of phone while noting the GPS time, then check the phone records for the time the call was actually received. Most likely, however, the phone company is using GPS time to record its information, for the sake of accuracy. In the unlikely event of a difference, the offset difference can be noted and offset applied to relevant evidence. Accordingly, a one-minute difference would result checking what was going on in the track one-minute later.
  4. GPS track time lines can be used to narrow the investigation based on other time-based recording systems. For instance, using a GPS track timeline and location, investigators could find camera systems (banks, convince stores, surveillance cameras or web cams) that recorded the vehicle or person tracked based on the recorded time of the camera. This would avoid watching hours of tapes, etc. Again, current GPS time can be checked against current camera time to find any offset. While some of these camera systems might be dead accurate, others might have discrepancies that do not degrade the evidencing value-the time coding is only there to find specific frames, not to prove the time. Nor should such a discrepancy in timing reflect on the quality of the GPS track vs. the standalone camera, which may have its "clock" set by the guard's 10 year old watch.
  5. Prove a conspiracy- Accurate time lines can show not only locational co-ordination (that parties met) but that they acted in concert. For instance, phone records show two suspects spoke, GPS locational information might show they both drove away-but the time component might show they drove away simultaneously to do acts to help the conspiracy. GPS time lines further enhance law enforcements ability to demonstrate not only that suspects were just acting illegally, but acting in concert.
  6. For everyday circumstances, such as determining how long a driver worked, whether a person or delivery arrived before a deadline. Having a reliable independently verifiable source to determine "when it happened" may substantially reduce litigation around these areas. For instance, when children change placement, frequently one party claims the other is late to drop off, or that the party charged with picking the late. How to resolve these disputes? The traveling party can carry a GPS, and print the track results for all to see. If the parties use GPS time as their official time, rather than inaccurate car clocks and the like, they may find that both the parties are "on time", they just were using inaccurate clocks.

Wednesday, February 18, 2009

GPS Evidence Repository?

We are now entering the age where GPS tracking information, that is, GPS evidence, is collected on numerous people on a routine basis. We also live in an age where it is easy to keep 1) who (or what vehicle) the GPS track is related to 2) the raw data (which at the least consists of time, longitude and latitude, with some tracks adding altitude).

Whether such information is collected by an employer, a third party working for an employer, or as part of an investigation, clearly there will be evidentiary information related to such a "storehouse" of information. Now is time time to start programs to collect and preserve it.

Imagine an unsolved murder of a young suburban woman at her home. No clues exist to tell who killed her. However, someone using a GPS evidence repository (like a fingerprint or DNA repository) can chart the location of the murder, and look for all tracks that would have passed, for instance, within 200 feet of the murder site for a specific period of time, like three weeks before the murder. If this produces too many tracks, the search can be limited, either by distance or time period. While this does not mean additional tracks will not be followed up on, just filtered to get the most closely related tracks first.

Now imagine the detective in charge of the case has run this check, and found a UPS truck that stopped across the street, and a car stopped near the victim's residence that was tagged as going to a drug house (using the investigative technique I set forth in my Wisconsin Lawyer article). Correlations to know evidence are checked, and the detective discovers the vehicle is owned by an individual who showed up on caller ID at the victim's home, and who admitted a social relationship to the victim, but denied seeing the victim on the day of the murder. Further examination of the UPS truck & target car tracks shows that, although not in exact area of the murder at the same time, the target car passed the UPS truck at excessive speed shortly after the murder. The target car is photographed and the UPS driver remembers both the delivery and that the target car almost sideswiped him at high speed. This identification gives enough evidence to get a search warrant for the vehicle (which the police find easily with real-time tracking as the GPS is still attached.) A few pieces of physical evidence are found in the vehicle, but following the GPS track backward, the police find the target vehicle went to a pawnshop. At the pawnshop, the owner finds items from the victim's home, and picks the suspect from a lineup. The suspect is arrested, and confesses he attempted to enter the woman's home to steal items for drugs, but was surprised by her and killed her to avoid detection.

This little tale (totally fictional) demonstrates how a GPS evidence repository made up of both routine commercial records and law enforcement GPS evidence can be a valuable investigation tool. Today, we are building biometric and electronic repositories, including phone records, international travel records, immigration records, billing records and the like. Most can be retrieved and cross referenced to build cases and serve justice. Why not GPS tracking records?

In a related subject, given a known vehicle but an unidentified driver, is there a way of telling whether the vehicle was driven by a specific person? I think there is. Driving habits, as shown by GPS records. I call it GPS profiling. We all drive differently, and our routes at a given time, our acceleration, our stopping practices and our following practices all differ. Now, our driving may not be so different as to pick one driver from the general population, but driving probably varies enough (given a known baseline, that is, several different incidents of driving compared to one questioned incident) to be able to pick out one individual from say, for example, four or five people who actually had access to a vehicle and opportunity to drive it.

Of course, GPS profiling doesn't discriminate, that is, it looks at the movement of the vehicle as controlled by the driver, in order to tell who that driver is. There are several items necessary to draw any conclusions from a GPS evidence put into a profile:
  1. Exemplar GPS tracks, like the writing samples a handwriting expert needs to judge against a questioned writing. These exemplars would have to be from any person that was suspected of driving during the questioned track.
  2. A high resolution GPS track-Just as a handwriting expert can't make conclusions based on fuzzy copies or blurred signatures, or a photo analyst can't make a conclusion based on blurry photos, a GPS expert would have to have sufficient track points, showing time and location, to actually make a conclusion about driving patterns. As a rule, I would say anything that is beyond 5 or 6 seconds track points would be of doubtful use for GPS profiling, because profiling would require reading subtle differences in driving style. This does NOT mean track points beyond 5-6 seconds aren't accurate and valuable for other types of analysis or evidentiary use, just that they wouldn't be much use for telling who was driving the vehicle based on prior GPS track incidents.