Hacker News new | comments | show | ask | jobs | submit login
The future of GPS (opensignal.com)
89 points by sinak 1782 days ago | hide | past | web | 45 comments | favorite

So I've worked in this space before (realtime GPS tracking for a Fortune 100) - reducing GPS battery drain is great...but getting the signal back to 'The Cloud' (ie cellular) to be processed is what sucks.

And if you do barf out the data via cellular and are in cahoots with your cellular provider - you get tower information that helps estimate/verify location. But again, all of this eats away at whatever savings from being impatient with the GPS signal.

With cellular and a modern LiPo, the battery life is, at best, measured in weeks.

The 'Future' of GPS will come ~2015 when the last of the 'GPS Block-IIIA' satellites are in orbit and we move from 50bps to 500,000bps. [1]

[1] http://en.wikipedia.org/wiki/GPS_Block_IIIA

Is this going to require new GPS receivers? I don't really see it addressed on the wikipedia page.

Using the new L2C signal will require a new receiver, but the current L1 (C/A) signal will continue to be broadcast. So you will not need a new receiver to continue the same level of service.

Yup, it operates on a new range of frequencies.

If we're talking animal tracking, as the author mentioned as a possible use case, could the data be stored on-chip and processed upon recovery? I imagine that would handily beat out any cloud-based processing in terms of battery life, but of course assumes that we only eventually want the data.

From what I understood, that's exactly what the paper proposes. You store the raw data, and only correlate it with the ephemeris later. I don't know why the author involved the cloud into it, it sounds like it can be done very easily on any Internet-connected device.

Sure, but even in the domain of animal tracking I imagine very few cases involve capturing the animals afterwards. That seems labor intensive (and prone to biting, scratching, etc).

If 'afterwards' is when it hits a certain geofence or known network - you still are still stuck with waking up the device and querying the network interface...or doing some non-C friendly code to interpret a geofence coordinate.

How is it compared to the Russian 3rd generation GLONASS-K satellite?

What benefits do the new satellites bring?

I don't think the future is GPS at all. I remember reading a while back about a researcher who was working on developing tiny optical gyroscopes that were ridiculously precise, the idea being that you can calibrate their position once and after that all future locations are determined simply by integrating the acceleration of the gyroscope. It would work in places that GPS can't (like caves).

To me, that's really awesome. Personally, I think the coolest use would be in running watches to finally get a perfectly accurate pace instead of a slowly-updating estimate.

I'll see if I can find the link to the researcher's page.

Dead reckoning is subject to cumulative errors, so that would have to be REALLY accurate.

To put it into perspective. Imagine a dead-reckoning system with 1ppm error (ridiculously low) and no drift (again ridiculous). Say you use this for in-car navigation. After driving around for 10k miles, the error is now 53 feet, which is too great for navigation.

There's more to navigation than city streets. That accuracy would be awesome for shipboard use.

Grandparent was talking specifically about replacing GPS; I pointed out one example where even an absurdly accurate DR system wouldn't be sufficient to completely replace GPS.

TIL "Dead Reckoning" is more than just a badass movie and band name. That's really neat!

This term comes from "ded" (short for "deduced") -- deduced because you are advancing your last fix based on course and speed.

That can synergize with GPS. Many OEM Nav systems in cars these days do something similar in tunnels, where GPS signals are unavailable.

But if gyroscopes became precise enough, I can't think of a reason for needing GPS anymore, can you?

To keep track of your position using inertial measurement you need two types of sensor - gyroscopes, to work out which way you're pointing, and accelerometers, to work out how fast you're going.

An IMU in a smartphone would be subject to 10 m/s/s acceleration at all times under gravity, 50 m/s/s on a rollercoaster, and 100,000 m/s/s if you drop your phone on a hard surface.

Let's say we don't mind losing your location if you drop your phone, so we pick an accelerometer with a maximum range of 50 m/s/s.

Now, how accurately can we measure acceleration? The raw image setting on a fancy digital camera is 14 bits, which gives 16384 levels. CD audio is 16 bit, so it has 65,536 levels. Assume you can come up with a design that offers a 32 bit range, for a full 4 billion levels.

That means your measurements will be precise to 0.0000000116 m/s/s - pretty accurate, right?

The thing is, after 24 hours your phone will have an inaccurate estimate of its speed (0.0000000116 m/s/s * 24 hours = 0.001 m/s) and after 24 hours with that inaccurate speed estimate you'll have an inaccurate position estimate (0.001 m/s * 24 hours = 86.4 m) and the longer you leave it going, the bigger the error can get.

TLDR: You need super-precise sensors to do dead reckoning that stays accurate for long periods.

One of the secrets of useful dead reckoning systems is damping which is basically reducing your velocity toward zero (or some expected value) by a measurement slightly greater than your error. The result is if your cellphone sit's on a table you don't accumulate errors. Coupled with the occasional calibration and you end up with some surprisingly accurate systems.

Yes indeed! Of course, that doesn't reduce your position error, it only stops your position error growing while you're stationary. And you can't be stationary all the time (or if you can be, you don't need a navigation system at all!).

Inertial Navigation Systems like OxTS make [1] use GPS to get rid of this integration error that accumulates over time. For applications in vehicles you can also use the vehicle's speedometer, so you're measuring speed directly, which means fewer integration errors.

[1] http://www.oxts.com/default.asp?pageRef=69

So you can put the phone to sleep. Depending on the power usage of the gyroscope.

I am somewhat doubtful about gyroscope replacing GPS. If it can actually happen, it will require, as noted by others, re-tuned/tweaked/calibrated/adjusted every day/week to void accumulated error/noise.

Keep in mind that as precise as gyroscope can get, they could be susceptible to shock (if you drop it on the ground).

But I am interested in seeing how it could develop into

IMUs [1] have been used in planes for a while now, I am not sure whether it is the future or the ancestor of GPS.

[1] http://en.wikipedia.org/wiki/Inertial_measurement_unit

Ancestor. GPS has displaced IMUs in aircraft as the primary navigation device.

And what happens once the phone's battery dies?

The key line in this article: "The disadvantage with this is that you wouldn’t get your location in real-time – it’s processed later."

The viability of this seems tied to how long 'later' is. Most consumer uses of GPS require quick response times or you'll see a huge dropoff in usage.

At the very least, I'd love to be able to put my handheld GPS receiver into a low-power "track-only" mode for the times when I'm not actually checking my coordinates. This would also be a nice thing to have integrated into fitness gadgets given that it wouldn't require massively increasing battery size.

If I'm running, or plotting a path in general, I don't need the GPS immediately. I think instantaneous GPS is mostly used for navigational and augmented reality purposes.

So, I could see there being some consumer uses for it, if you could demonstrate that the power saving was worth the wait.

Post processing requires the GPS to store much more data than just a lat/lon pair. It may need more power not less to store the extra data.

Yes, for products we usually think of us consumer GPS products such as phones and navigation units this won't change much--unless you're willing to listen to your navigation unit's voice yelling at you to make a U-turn onto that road you drove by four hours ago.

On the other hand, this will open up new uses for GPS and potentially let you utilize it in a range of applications that just need to record time/position instead of immediately report it to an app/user. In exchange for lack of immediacy, users acquire higher fidelity data. I'm expecting to see this be more relevant in scientific fields than consumer electronics.

But for a lot of purposes – e.g. tracking the locations of animal migration, this could be a huge development.

This is where it would really come into its own. From what I've worked out in the past, best-case for battery life today for trackers is to record a coordinate roughly every 4 hours and sleep. Longer than 4 hours, and you have to do a cold lock, which as discussed takes significantly longer. With my GPS unit, this translated to around 3-6 months of battery life with 1000mAh capacity, depending on how good you are at optimizing power.

From the article, it seems to me that "later" is "when you get an Internet connection".

The paper: "Energy Efficient GPS Sensing with Cloud Offloading"


I suspect that eventually we're going to start integrating chip-scale atomic clocks[1] into our mobile devices, which will allow a much longer synchronisation period. Right now they're still ~35g and ~130mW power, but that's only going down. There's going to be a whole bunch of interesting new distributed processing stuff you can do with highly accurate synchronised nodes; something like the Google Spanner system on a more local scale, perhaps.

[1] http://www.symmetricom.com/products/frequency-references/chi...

It looks like this is actually the strategy that Memoto is doing with its life logging camera, which makes perfect sense in that use case: save battery by offloading the geotagging math to the server.

1.5 years of continuous GPS tracking with just 2AA batteries - opens up some scary possibilities.

I thought this was going to suggest cloud route processing, like Apple Maps/Google Maps/Nokia Here where the GPS is done on the phone, but the routing is done in the cloud.

This is why your iPhone or Android is a hundred or even several thousand times faster to "recalculate" a new route when you take a wrong turn than your trusty, old Garmin. That's also why if you get lost in the middle of nowhere w/out cellular reception, you'll be hoping your old TomTom is still in the glove box - the maps and routing algorithms are all stored on the device itself (though this means the data on it gets outdated and you will need to update it, often for a price).

Doesn't google maps allow downloading of data to the phone for use without internet?

Yes, you can download a fairly large region and keep it on your device. It also downloads a route when you begin navigating so you could just turn off your wireless once you start and it'll still get you where you are headed.

OSM does too (OSMand client for android). I like it. It was very useful when in Ecuador and I only had a voice-only cell signal.

I'm guessing that's Android only?

Since it would be post processing anyway, one could possibly also make use of corrections from IGS ( http://igscb.jpl.nasa.gov/components/prods.html ) (or similar) to get extremely precise tracks. Quite nifty.

Off topic, but the "flying brick" would actually let you measure longitude; you get a lattitude fix via stars, and then the time from the position of the flying brick (which has a known period).

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact