Yesterday I started testing. As with most tests, I learned things I hadn’t expected to. The test was simple. Evaluate the Smartphone position location stability at rest. Take a short road trip and compare ground truth, the GPS display underway and the Google Latitude record of the trip. Based on this experiment, one can intuit
using Google Latitude to keep track of a boat left unattended is an iffy proposition at best — very location dependent. I also believe l detected some possible software design rules which may render Smartphones unsuitable for marine navigation especially in areas with significant cellular infrastructures.
First, position stability.
I placed my Smartphone in an unoccupied room for two hours oriented North-South — the green place mark shown in the picture below.
In those two hours, Instamapper (to be similarly discussed in a future post) showed the phone in the range of positions marked with “I” placemarks. Google Latitude showed the phone in the range marked with star placemarks. The street to the left is 10 meters wide. Their location sampling was not synchronous, so theoretically, both applications operating synchronously might well have provided virtually identical data (subject to any post-processing the data receives), and they might not have as well.
Some interesting things here. First, the errors are mostly E-W oriented. GPS satellite constellation dynamics over two hours would predict something a bit more random and elliptical.
Second, a couple of significant position jumps leave me thinking antenna power management may have been uneven depending on what other Apps were demanding. I did receive a !@#$%^&* telemarketing call during the test. But the call time does not correlate with a jump. I looked at the App power usage trace for the period involved, and no gun was smoking). I checked local flight schedules and no large aircraft flew near us during the test period either. (Their multipath reflections can lead to jumps.)
Clicking on the thumbnail will give you a few second movie of the error circles associated with the shifting positions reported in Google Latitude . It is interesting to me that the worst accuracies seem to be associated with the western-most positions. Since the western-most positions are closest to the phone’s real position this is troubling.
Second, the road trip.
This graphic is best viewed in a separate browser tab.
The lilac line represents ground truth. The Smartphone GPS display showed us on this path, in the correct travel lane, through the entire trip. No hops, skips, or jumps.
The yellow arrows are the outbound Google Latitude reporting points. The salmon paddles are the return ones. The dark red arrows are the implied outbound path according to GL. The dark blue arrows are the return.
Notice the extraordinary position confusion in the vicinity of the airport to which I took my wife. Notice, also, the following:
Two outbound positions more than two minutes apart are reported in the identical location (with 1857 meter accuracy), and a return position 20 minutes later is reported in the same location with identical accuracy. This has to be a position determined by cellular grid reporting, not GPS.
In fact, it appears that all of the off-path positions are such. So, provided we are getting a good position report (ala the position stability test above), this might be acceptable as warning that the boat had been moved, but it would be potentially useless for tracking the boat thereafter. And it would require routine visits to Google Latitude. We still need ET to “Phone Home.”‡
But if we think further, if Google Latitude is reporting the raw position information supplied by the Smartphone GPS, and the phone’s navigator software is doing a lot of snap to grid and predictive position modelling, then these GPS units are not useful in situations where there is no grid to snap to and the phone’s location cannot be predicted.
I have little enthusiasm for Smartphones as navigation devices just based on screen size (and I’m not keen on handheld GPS for persistent use either). Aside from the navigational risk implied by this little test, there are also the problems associated with the phone providing the proper location-aware data such as tides and services, if the raw positions can be pulled two kilometers off path by using cell tower triangulation when the GPS signal is weak, etc. In this test, fewer than five of 45 positions reported to GL were at accuracies reasonable to expect of a handheld (non-WAAS) marine GPS device.
Given that many of the pad-style devices have no better native GPS than these phones, they appear to me to be a marine navigation accident waiting to happen. If I were inclined to use one (and I’m not) I would certainly port a better GPS to the unit — and of course it then loses its “pad” mystique. I suppose one could use Bluetooth but that introduces another set of interference problems.
Some things to consider with this test. It was on land. The GPS was (mostly) in a car (but in a good visibility bracket). The speeds were as high as 55 mph. When boat projects are done and sailing resumes, we will redo this test in a marine environment and at sailboat speeds.
In all fairness to Google Latitude, it wasn’t intended for this, and props to them for providing the capability. If the issue is purely what the phone is reporting, then no online service is going to work for my purpose — garbage in, garbage out. If it’s about reporting and processing, it will be transport grid dependent to a great degree, and I’m looking for an off that grid solution. So, this is about what the Smartphone — an unrooted Motorola Droid running 2.2 was reporting and whether it was useful for asset tracking. For now it appears the answer is “no.”
‡Getting a Phone Home message isn’t all that hard. Before Smartphones and the lure of their Apps, I did this by having a reed switch below decks and a magnet embedded in a dockline. Instead of phoning on an open switch, it triggered a relay, and a very loud alarm bell in a very hard to get to location rang until it was reset. The only time it ever triggered was when the marina (not my current one) moved the boat without my permission. I’d shown them how to shut it off. They never fessed up, but the relay counter showed a “1” and a lefty had re-cleated the dockline…