Page 2 of 2

Re: Staying alive in the snow

Posted: Fri Jan 25, 2019 11:59 am
by gdurkee
Some good points and questions here. Take home message is there are errors inherent in both dedicated GPS (e.g. garmin etc.) and in cell phones using GPS. Our SAR research group started looking into this a few years back when we got reports of people reporting their coordinates read from a cell phone. Errors would be 1 mile (and 4,000 vertical!) to 6 miles.Obviously significant and a danger to responders. So,what's going on?

For those of you yearning for a long, very long, explanation and introduction to cell phone gps capability, I wrote this a couple of years ago: ... HFFYW1xN1E
A shorter version meant as a workflow to get coordinates from someone reporting an emergency via a cell phone is here: ... kVHWndOWWc
This latter document, though meant for 911 centers and SAR managers, is useful because it provides a few apps you can put on your phone to provide responders with more accurate location coordinates.

I'll try to give a brief overview (uproarious laughter from anyone who's read my assorted eye-drooping explanations of anything...).

First, gps and satellites. The GPS network is pretty old and, when designed, had/has really limited bandwidth to send data from satellite to gps. It only works one way. Your gps, whether dedicated or cell phone, is a passive receiver. It calculates location based on the time signal from several satellites, ideally at least three. To do that, the receiver needs a list of where each satellite is going to be and the distance so it can calculate when the signal will have been sent and the time it gets to the device. When you have at least 3 satellites, it can triangulate your position and show the coordinates. As bobby49 suggests, multi-path can be a serious problem. I've seen individual pings as much as a mile off. This is especially a problem in deep canyons, like the Grand Canyon. However, averaging over time usually shows a semi-accurate fix. A gpx track can be a mess with the occasional ping that's way off but, overall, the route can be seen.

To get a fix, the receiver needs two files. An almanac and an ephemeris. Both contain the information on satellite location. The problem is, the almanac, especially, takes about 20 minutes to download because of the narrow bandwidth available. This is fine when you use your gps often and in the same general area. If, though, you've not turned it on for a couple of weeks and/or moved over 50 miles (and I'm not sure of exact times or distance...), then the receiver has to download a new almanac. Until it does, the results displayed could be very inaccurate. Which is to say that for an accurate reading, especially where you're using it for the first time in awhile or you've traveled a distance, then turned it on, your readings may be way off.

This applies to dedicated gps and, under some conditions, to cell phones.

(and here I take a short break to pound my head on my keyboard. I'd finished the post in the quick editor, went to Preview, where I signed in again, which was blank. I went back and, auuuughhhhh!, it was all gone though I'd previously saved the above elsewhere.)

Still, I'll try to reconstruct my brilliant prose.

Cell phones use several methods to acquire location. The goal of cell manufacturers is to get as fast a Time To First Fix (TTFF) as possible. To do that, they take some shortcuts and don't rely entirely -- or sometimes at all -- on the phone's gps. The location of a cell tower is known, so the phone can calculate the distance from the tower using the round trip time of a ping. Direction is more difficult and, if it's an omni-directional tower (360 deg antenna vs. individual antenna with a cone of coverage), impossible. With a cone you get only a very rough direction but, combined with help from the gps, you can get a location. Finally, best case is triangulating -- or more -- using multiple towers. But, in building canyons, you also get multi-path so there can be error there. I'm not clear on it, but I think known wifi can also be used. There's references indicating that the Googlemobile and other companies suck up all the mac addresses and locations of wifi, allowing the phone to use that data set. I'd read an article that Europe was going to ban that practice but not clear of current status.

Still, phones seem to be accurate most of the time -- except when they're not. The errors we saw in the cases where location was way off appeared to show the phone's last good fix but not it's current location. Also, and not 100% sure, these cases involved reception off only one cell tower. So, in areas with spotty connection, the odds seem to increase of an inaccurate location shown by the phone. Setting the phone to gps only or using a gps-only app would probably reduce the chance of error.

With Android, you can actually set the phone to the level of accuracy you want though perhaps sacrificing speed. I think the setting "high accuracy" is gps only and the other two use different combinations of the above to solve for location. With iPhone, you don't have that choice though I'm not clear on it's effect on accuracy except to say that several of the errors we looked at were from iPhones.

There's workarounds with apps I link to in the Workflow paper. One will send a text to a person who doesn't know their location. In the reply, permission is given to use the phone's gps to send location. Another app by one of our crew will take you to a site which, again with permission, reads your phone's gps and posts it, including a map location. Other apps, like Backcountry Navigator, I believe use only the phone's gps (but not sure...).

While I'm here, I'll also mention that PLB's also can have serious errors as a result of the tech they use. Those are the gizmos used by ships and aircraft (ELT) and individuals (PLB) that work with a dedicated satellite system and the Air Force Rescue Coordination Center to derive the location. Here's another short paper on those: ... XVYb3U1NkE

The good news is, as far as I know, SPOT and InReach are pretty spot-on accurate. They use only GPS though multi-path could effect their results. That said, I've never seen it so they might average the results or use some other post-processing when the route is displayed.

Last! I'm a poor but mostly honest forestry major and semi-retired ranger. I don't have a background in gizmos and took this up because of some serious errors in location on self-reported SARs, and even including 911 transfers. We figured someone somewhere would have a good summary document on how cell phones acquire location. We never found one so I ended up writing this stuff from research our team compiled and contributed. If anyone here has additions or corrections, they'd be much appreciated! Send them to me directly.


Re: Staying alive in the snow, GPS, and cellular phones

Posted: Fri Jan 25, 2019 3:14 pm
by bobby49
First of all, I have to say that this was interesting reading. I do not disagree with your facts and figures, although maybe some of the assumptions are weak. For example, yes, I am a smartphone user, but I power it on for maybe one hour per year for recharging and other checks. I would never consider carrying it on a backpacking trip, since it is mostly dead weight to me. The cellular service coverage is zero or near-zero where I go. Maybe this is my own problem, but I have very poor reliance on the whole cell phone system since I worked around it for years and years, although that was mostly in the GPS business. Even when I power up my phone, I have location services turned off. Call me paranoid if you wish, but I don't particularly want anybody tracking my phone for any purpose, legal or illegal or unethical.

I've seen cellular phones have all manner of GPS problems, and that just starts with dead batteries. Personally, when I go out into the wilderness, I am carrying my inReach device, and I only power it up when I want to send a routine message. It will not send the damned message until it has a fairly decent GPS lock, so when I've been in a difficult area, it has taken a long time to get any message out. Sending an SOS message is rather different since there are different satellites involved, and that involves a different view of the sky. There is one thing that I learned many decades ago when I was a comm chief in an Army infantry unit. If you are relying on terrestrial communications, then you better get up to some high point where your signals will get out. That is only partially true if you are relying on satellite communications. I'm not an honest forestry major or ranger, but I do have a background in gizmos (that is a technical term). I also taught GPS classes occasionally for years.

Oh, one little comment about GPS multipath interference. A GPS downlink signal can bounce around unpredictably if there are _flat_ reflective surfaces, especially if they are vertical. In general, a rough granite rock face does not make so much of a problem that will last for very long. The problem can exist underneath a tree with leaves, although it looks like a different problem when analyzed. Modern GPS receivers are getting better at getting an accurate position fix even with this condition, but they can't always be perfect.

One little GPS anecdote: A team of us was climbing the highest peak in the western hemisphere. When I reached the summit, I lit up my (now ancient) GPS receiver. It was spinning around for minutes, and then locked up a position. The altitude that it reported was one meter away from the official altitude for the peak. Since altitude accuracy is generally much worse than lat/long accuracy, I thought that was close enough for government work. Plus, that was back before the full constellation of Navstar satellites was in service.

Re: Staying alive in the snow

Posted: Sat Jan 26, 2019 5:47 am
by rightstar76
George, thank you for your informative post on GPS and cell phones. I appreciate the time and care you put into this (and all of your other posts - I am still reading the ones you recently made). We are so lucky to have you here on HST! :thumbsup:

Re: Staying alive in the snow

Posted: Sun Jan 27, 2019 12:19 pm
by gdurkee
One more note about any location device. In addition to the errors inherent in the tech, probably the greatest source of error is incorrectly reading or receiving coordinates and, especially!!!!, assuming a coordinate type when it's not. So, a good idea to get in the habit of becoming familiar with various coordinate types and be aware that if someone's reading off some number to you, stop & ask what coordinate type they're using and what the datum is. I immodestly quote from our book Using GIS in Wilderness Search and Rescue:
When reading off coordinates, clearly identify both the coordinate system and the datum. Then read the coordinates saying "degrees," "minutes," and "seconds" where appropriate as well as "decimal." Some have used just a pause between, for example, minutes and seconds and assume the person writing the coordinates down knows to enter the appropriate type. This has resulted in serious errors. In our experience, other than UTM and USNG, decimal degrees are easier to pronounce over the radio and type into mapping software than degrees, minutes, seconds and degrees,
decimal, minutes.
I recommend Chapter 4 on coordinate systems on what I like to think is a pretty clear overview, if you're interested (or, at least read from page 60). ... -ebook.pdf

Re: Staying alive in the snow

Posted: Fri Feb 01, 2019 12:39 pm
by wildhiker
Here's an example of the "multipath" GPS error due to reflections off smooth granite valley walls. This is the raw track my inReach Explorer GPS unit recorded last July hiking up the Tamarack Lake Trail in Sequoia NP. Data points are recorded at 1 minute intervals. (Large circles are when it tried to upload data at 10 minute intervals, I believe). The wild jumping up and down the canyon walls occurred while I was taking a rest break in one spot!

Re: Staying alive in the snow

Posted: Fri Feb 01, 2019 4:49 pm
by gdurkee
Great example! When mapping downloaded tracklogs from a SAR, I edit all that out. I use ArcGIS so can edit the lines and reconnect them for a continuous line track. I'll leave the points as is though don't usually display them. It's often a spaghetti mess, especially when, like you, we've got a short ping interval and when the teams don't zap their previous day's track or turn their GPS on on the flight in.

The idea, of course, is to represent as closely as possible a team's search effort and not distract Plans with gibberish data.

Your rest break cluster reminds me of a SAR many years ago. A group got caught in a snowstorm and bailed back down the hill (Roaring River). You could see from the cluster of points they stopped at the sign for Avalanche Pass and Cedar. Their GPS told them if they went straight down Roaring River they'd be at their car in 3 miles or about 10 over Avalanche. With the benefit of hindsight (a 5 day search for them...) I knew their decision. I wanted to scream at the dot cluster -- "no, don't go that way!" But, of course, they chose the path less traveled, got caught by major runoff and, finally, trapped by a waterfall. Lost a pack with tent and food in the river but, finally, got spotted by helicopter and hauled out. I later asked them since the terrain (river) kept getting worse and worse, why they didn't turn around. "Our gps showed we were getting closer to the car...". Another case of testosterone poisoning.