Hacker News Reader

Comment by pithon | Hacker News Reader

Discarding assisted GPS where GPS data messages are obtained through about channel such as cellular or Wifi...

Almanac gives coarse satellite position information (and some other stuff), good enough to know which ones are probably visible and therefore prioritize signal acquisition attempts which used to be very very costly in terms of signal processing. That's the message that takes up 12.5 minutes to piece together. Nowadays you can just brute force all possible satellite signals and there's no need to wait around for the almanac information. Each satellite signal broadcasts precise satellite location information (ephemeris), which takes maybe 30 seconds to get a few frames I believe. So that's basically the bottleneck for a modern chipset which starts with zero information and relies solely on the GPS signals to navigate.

View on HN

I believe this might be the correct answer. The signal from a GPS satellite is incredibly weak (way below the thermal noise of a typical amplifier) and can only be detected by correlating it with the satellite's unique gold code. In addition, the satellites move pretty fast which leads to sizeable Doppler shifts of their carrier frequencies (tens of kHz). This has to be taken into account in the signal demodulation.

Classical GPS receivers use the almanac (and a reasonably accurate local clock) to determine which satellites are probably in view, and with which Doppler shifts. I would not be surprised if modern GPS chips had enough compute power to simply correlate the received signal with all gold codes and at all reasonable Doppler shifts. The almanac is then no longer necessary.

Once the initial code is had, you can just store the almanac and reuse it most of the time with reasonable accuracy to make the computation faster.

Usually that approach cuts the sync time by factor of ten, even if the real ephemeris diverges from the general almanac.

The almanac is still needed, it has the orbital positions and those change over days. I think current GPS receivers can receive from lots if not all satellites at once. Different satellites are transmitting different parts of almanac. The receiver downloads the almanac in parallel.

Modern GNSS receivers with modern multi-band antennas can track hundreds of satellites simultaneously across all the various GNSS systems including GPS, GLONASS, GALILEO, etc....

This is how they build RTK surveying equipment, and they can have their own RTK reference point that uses statistical averaging over time to get accuracy down to the centimeter range or below, and then rebroadcast that to the local RTK rover nodes. There are multiple correction systems available that you can subscribe to, some of which can transmit their corrections via 4G/LTE/5G or other radio band communications, so that you don't need your own RTK reference for your rover units.

For good RTK equipment, you can easily pay $40k or more, but if you look around you can find some equipment that costs less than $1000. My current favorite is the SparkFun GPS RTK Express Kit, although you need a pair of them if you want to do your own RTK reference point and you don't subscribe to one of the various other correction systems.

The value of RTK is partially that they are stationary and can average out noise over time to get an accurate position estimate, but mostly that they're sending you their live signal measurements so you can cancel out biases common to receivers which are geographically co-located.

Your reference point is stationary, yes. But not your Rover units.

For RTK to be really useful for surveying a site, you need both the stationary reference unit and the Rovers.

TBH having live real time position error transmitted to live in motion "local area" moving bodies (drones | aircraft | vehicles | missiles) is useful for live dynamic tasks (shooting the enemy) .. but not especially useful for surveying per se.

I spent decades in high precision geophysical surveying using ground stations (fixed GPS recievers) and craft (aircraft | helicopters | etc) and it was common practice to collect data and post process ... merge airframe GPS data and ground station corrections in conjunction with other numerical corrections .. at the end of day.

For various forms of survey you really want to not transmit to|from your airframer "live" in any case - for EM | magnetic survey that's just another source of noise best dealt with by having everything not required switch off.

I concede the value for regular ground site survey tasks .. there's no downside to transmitting and applying corrections in real time - but it's not strictly required and for some forms of surveying it's preferable to record raw unaltered data and apply fixes later with the option of rolling back | examining the raw | etc.

I can see the value of recording the raw data at the time and then applying the corrections after the fact. I'm sure there are applications for that mode of operations.

But for the ground surveying and ground control RTK equipment I've been looking at, you do want to apply those corrections in real time. For example, if you're controlling a tractor that is pulling a lot of heavy equipment through a field of crops, and you want to make sure you line up correctly on the product being farmed (e.g., corn), you want to make sure that your tractor goes exactly where you want it to go, and in real time.

In my mind, that's what Real Time Kinetics is all about.

Heh - I came back to correct myself, I completely forgot about ground surveyors hammering in pins | monuments - they want the real time corrections as they go for that task.

But yeah, real time control of moving vehicles is exactly where you want transmitted corrections - it originated with military requirements to correct in real time as interceptors close in on targets and today (like yourself?) I use it in the agricultural domain for tractors and drones.

It occurs to me that if you're just passively looking at the data that is collected, then collecting that data and correcting it afterwards may well be the best approach.

But if you're taking actions that might result in changing the thing or the space you're surveying (like putting in markers, or controlling heavy equipment, or directing the application of military force), then that's where you will most likely want to have those real time corrections.

The almanac is not needed. The ephemeris is basically just almanac orbital data plus extra terms to make it accurate enough for meter-level positioning or better. I'm not sure what you mean by saying the orbital parameters change over days. Ephemerides are updated every ~2 hours to maintain precise positioning.

That makes sense!

Reply on news.ycombinator.com