Slashdot Mirror
Super-Accurate GPS Chips Coming To Smartphones In 2018 (ieee.org)
schwit1 writes about a new mass-market Broadcom chip designed for the next generation of smartphones: It'll know where you are to within 30 centimeters (11.8 inches), rather than five meters. At least that's the claim chip maker Broadcom is making. It says that some of its next-generation smartphone chips will use new global positioning satellite signals to boost accuracy. In a detailed report on the announcement and how the new signals work, IEEE Spectrum says that the new chips, which are expected to appear in some phones as soon as next year, will also use half the power of today's chips and even work in cities where tower blocks often interfere with existing systems. All told, it sounds like a massive change for those who rely on their phones to find their way.
I had always heard the lower accuracy from gps was a government imposed restriction or limitations of the protocols not a technical one. is that simply an old myth? I know nothing about the tech personally.
What you are talking about was called Selective Availability and it was disabled in 2000 by Clinton's executive order and never used since then. Current GPS precision is limited by ionospheric scattering and reflections of signal from buildings, it is indeed a matter of having enough satellites in sight to filter out outliers, and smart signal processing, to get better precision (while 30cm is probably possible only by also having a decent INS onboard and integrating data for some like, like 10-20 seconds, with sensor fusion with the INS).
So, there are multiple GPS signals. Originally there was a C/A code (on L1), and a P code (on L2). The C/A code is used by civilian receivers, and both codes are used by military receivers for increased precision. There used to be a pseudo random uncertainty added to these codes that required special keys to correct, which only military receivers had. This was called "Selective Availability" and was disabled May 1, 2000. The newer GPS satellites have even more codes which increase accuracy, including some available to civilian receivers.
So, the military does still have more accurate GPS than civilians, but there is no longer an INTENTIONAL error added to the civilian side. Newer GPS satellites and newer receivers include more and more technology to get more accuracy even on the civilian side, though. Most modern GPS chips support multiple systems, including GLONASS, GNSS, and others which can be combined to increase accuracy. I am not sure how this new Broadcom chip works but it is probably a combination of the newer signals and the combination of multiple systems. GPS on cell phones is typically integrated into the cellular modem itself and can make use of location data from the cellular network to help as well.
Check out https://en.wikipedia.org/wiki/... and https://en.wikipedia.org/wiki/...
Technophile
The US government accuracy issues were removed ages ago, partly because the military was trying to use civilian devices and the imposed inaccuracies were hurting them more than the enemy, and partly because it is easy to compensate the imposed error by measuring it with a fixed receiver at a known location.
Current generation mass market GPS receivers use a single frequency band, L1. The current errors are mostly a combination of atmospheric error, the signal is delayed differing amounts in the ionosphere depending on the weather, and errors in the local crystal oscillator of the receiver.
The GPS satellites have long transmitted multiple signals at different frequencies, L2 is encoded to reserve it for military use. Using two frequencies allows the ionospheric error to be mostly removed as the two frequencies are impacted differently. I understand it has been possible for a while to break the military signal with enough data but in practice that isn't necessary, the shape of the encoded signal can be used to provide enough information to correct for the ionosphere.
Satellites launched from 2010 have also transmitted a signal on L5, like L2 it is another frequency with all the same benefits but without the military encoding.
In the past most budget GPS systems have only supported L1 because each frequency requires an independent front end and that is a significant element of the cost. Even survey receivers typically used only L1 and averaged the signal over time to increase the accuracy. The new Broadcom chip has a front end for L1 and L5, in addition to all the other frequencies a mobile phone can receive, allowing the ionospheric errors to be corrected for.
The errors in the crystal oscillator will continue to be a significant problem, 1ns is pushing the envelope for a non-ovenised crystal, and an ovenised crystal isn't really an option for a cost sensitive mobile device.
I do know something about the technology and this kind of accuracy has been in the works for a while. The military grade code was supposed to provide accuracy that only the government was supposed to be able to use. It took me about two seconds to figure out that just knowing some basic properties of the signal could provide similar accuracy if someone was willing to throw enough processing at it. Advancement in computer technology has made the processing needed very cheap, light, and small. There is still an advantage to knowing how to decode the military grade signal, it's just that the advantage is very small unless traveling in ways that are generally unique to the military.
Early NavStar signals had a civilian accessible code which gave a "quick and dirty" navigation and a military grade signal that required the decoding of the first code to be useful. Now the "quick and dirty" code is still there for legacy reasons, then there is a higher accuracy civilian signal, and the hyper accurate (and now independent of the other signals) military coded signal. If one has access to the encryption keys for the military grade NavStar signal then one can get hyper accurate location in real time with minimal processing and few satellites in view.
What it sounds like these new systems are doing to get their location as accurate as they do is stack NavStar, GLONASS, and Galileo on top of each other, and do some intense processing of those signals. This is impressive accuracy that is comparable to what NavStar provides with the military grade signal alone. Unless a shooting war breaks out this hyper accurate GPS will remain just as good as the military NavStar signal.
If there is a shooting war then there is the possibility of the US federal government shutting off the civilian signals on NavStar to deny that to the enemy. The remaining military signal would still give some location information to those willing to throw a lot of processing at it but the accuracy would suck without the civilian cleartext signals to work from. It's this possibility (threat?) of the USA turning NavStar off that prompted the European Union to develop their own system, and Russia their own.
I am armed because I am free. I am free because I am armed.
It used to be the case that errors were intentionally injected into the "coarse" signal. However, the encrypted "precise" signal (reserved for military use) was left untouched.
After some minor SNAFUs during the 90s Gulf war, with allies being unable to source an adequate number of military GPS receivers, and needing to fallback on civilian gear, they navy decided to turn off the error injection temporarily during periods of conflict. As this made a nonsense of the deliberate error injection, which was intended to prevent enemies from obtaining the strategic advantage of GPS, the US gov decided to end the error injection, and switched it off permanently in 2001.
There are technical differences between the "coarse" and "precise" signals, which allow for better accuracy when both can be received and processed together. (The precise signal has a higher "chip rate" which allows its phase to be measured more precisely, and by using 2 frequencies, the signal dispersion in the atmosphere can be directly measured, rather than relying on a general model).
As multiple other countries/multinational goverments have developed their own GNSS technology, there has been migration of some of this "military grade" technology into the civilian space. There was some major political wrangling in the early 2000s, when the EU announced that their satellite navigation system would offer not just the classic civilian signal, but a free to use upgraded 2nd frequency "intermediate precision" civilian signal (giving most of the benefit of the US military signal), and an encrypted (paid subscription) commercial signal equivalent (equivalent or better performing) than the US military signal.
However, the political objections from the US dried up, and the newest US GPS satellites now offer similar upgraded free-to-use signals to the EU systems. China has done the same with their latest satellites.
Although, full roll-out of satellites offering the upgraded signals is not complete, there are now sufficient satellites offering the upgraded free to use signal (known as L5) that receivers with L5 capability can be expected to work out of the box.
You really don't want to completely turn off the gps chip. Getting the L1 and L3 signals back and the ephemeral data they contain takes a lot of time, sometimes quite a few minutes. Most people will not wait that long when they hit their maps app. That said, gps can only tell you where you are, it can't tell anybody else where you are unless there is an outgoing signal somewhere, which is certainly a thing to ponder if you put all the antennas and communication in the same chip.
This is +- 3m in any direction
Good logic, but not quire correct. Yes, horizontal resolution is normally pretty much equal in any direction. Vertical resolution, however, is hampered by a relative dearth of altitude information at low satellite elevation angles and the fact that most satellite observations at any given time will come from satellites at low elevation angles. The latter is a geometry thing. If you're good at (3D) geometry, you can think it through and convince yourself.
If geometry makes your head hurt, you need probably to find some satellite acquisition tables and observe how little, if any, of the pass time (for almost all cases) has elevation angles above 45 degrees. I looked on line for an example, but couldn't find one in a few minutes of looking. Maybe better search terms ...
You can't see ANYTHING from a car, You've got to get out of the goddamned contraption and walk...Edward Abbey