| View previous topic :: View next topic |
| Author |
Message |
rwbaldwin
Occasional Visitor
Joined: Dec 24, 2012
Posts: 55
|
 Posted: Mon Jan 21, 2013 11:59 am Post subject: |
 |
|
| matthewj wrote: | | M8TJT wrote: | | Why would they make it square? To do that they would have to do the additional computations that you have had to do. However, if it were to be done by using lat/lon, that would make it easier' but it wouldn't be a square as' say, 1degree east/west is longer the nearer the equator you get. |
First, look at the pictures that have been shown - the points are all in a square area, not a circle. As to why, it is far easier and faster to program. The code is simple:
if Abs(Point.X - CurrentLocation.X) GT 50 then
DiscardPoint;
(which is, if distance from current on X axis is over 50, then discard). No complex floating point math involved, which you want to avoid on a big list of items. Since this is a crude thing, you can do crude math too. |
M8TJT is correct - I've worked on GIS software in the past. There is no Point.X or CurrentLocation.X. If you want X coordinates you would have to calculate them from GPS coordinates.
On every system I've worked on, bounds were calculated directly in hms. ie. position +/- hms tolerance. If the point falls in both the lat/lon ranges the actual radial distance is calcuated.
I've never known any company do it differently because it very efficient and accurate. You only have to calculate the bounds once and most of the points are exclude by a simple comparison. |
|
| Back to top |
|
 |
M8TJT
The Other Tired Old Man
Joined: Apr 04, 2006
Posts: 10118
Location: Bexhill, South Sussex, UK
|
| Posted: Mon Jan 21, 2013 1:55 pm Post subject: |
|
|
That would give a trapezoid, with the angle between the E/W sides and the N/S sides dependent on latitude. At latitude 52N, this would give an approximate rectangle with the minor dimension (E/W)about 0.6 of the major dimension (N/S) Similarly the distance between the two N/S sides would vary, again depending on latitude. Unless of course you bring Cos Lat into the sums. Given this, the diagonals would be approximately 1.16 times the major (N/S) dimension.
Having banged on about that, I would chose to do it on lat/lon as well. |
|
| Back to top |
|
|
rwbaldwin
Occasional Visitor
Joined: Dec 24, 2012
Posts: 55
|
| Posted: Mon Jan 21, 2013 2:16 pm Post subject: |
|
|
| M8TJT wrote: | That would give a trapezoid, with the angle between the E/W sides and the N/S sides dependent on latitude. At latitude 52N, this would give an approximate rectangle with the minor dimension (E/W)about 0.6 of the major dimension (N/S) Similarly the distance between the two N/S sides would vary, again depending on latitude. Unless of course you bring Cos Lat into the sums. Given this, the diagonals would be approximately 1.16 times the major (N/S) dimension.
Having banged on about that, I would chose to do it on lat/lon as well. |
All correct - but of as you know, it doesn't have to be accurate - it's just used to reduce the the dataset.
The lat/lon method (creating a bounding 'box') is SQL server efficient - requiring only simple comparisons. Of the data set returned, only about 25% of calculated radial distances will be disgarded (given an even distribution) so all in, it's by far the most efficient method. |
|
| Back to top |
|
|
M8TJT
The Other Tired Old Man
Joined: Apr 04, 2006
Posts: 10118
Location: Bexhill, South Sussex, UK
|
| Posted: Mon Jan 21, 2013 2:56 pm Post subject: |
|
|
| But from the reported behaviour of TT, they first download the data including far off and then discard that which is outside the hoirizon. So it does not seem be a bandwidth thing but more of a 'we know what the user wants' thing, despite loads of posts both here and on the TT forums and elsewhere to the contrary. |
|
| Back to top |
|
|
rwbaldwin
Occasional Visitor
Joined: Dec 24, 2012
Posts: 55
|
| Posted: Mon Jan 21, 2013 3:27 pm Post subject: |
|
|
| M8TJT wrote: | | But from the reported behaviour of TT, they first download the data including far off and then discard that which is outside the hoirizon. So it does not seem be a bandwidth thing but more of a 'we know what the user wants' thing, despite loads of posts both here and on the TT forums and elsewhere to the contrary. |
They are definitely transmitting the same horizon to all devices and I've noticed that horizon does change depending on location. Assuming the full range units such as the 950 are not disgarding data, it is a circle (if you zoom in and browse the map diagonally you see this) but it isn't always centred on your location.
Only the limited range units such as the 1005 seem to disgard data. This is a recent development. This means they are sacrificing bandwidth efficiency. I assume that at some time in the near future the likes of the 1005 will only get data for the 70km range they display.
Trying to see the logic in what TT does is not easy. Maybe there is no logic  |
|
| Back to top |
|
|
matthewj
Frequent Visitor
Joined: Apr 03, 2006
Posts: 751
|
| Posted: Mon Jan 21, 2013 4:21 pm Post subject: |
|
|
| M8TJT wrote: | | matthewj wrote: | if Abs(Point.X - CurrentLocation.X) GT 50 then
DiscardPoint; |
But that would describe a circle, which is exactly the point that I was making about a square. You need to bring root2, and/or Cos into the equation for a square and dont forget we live on a sphereoid not a flat earth  |
The key with my poor example was that I was considering X and Y completely separately, which means you end up with a "square". I accept that this is a naive example though as it is simply how I'd do a quick removal of the data - the actual calculation may use more smarts. As a programmer by trade, that's my first thought on "how it is done" as I'm sure the pictures show "square" results.
Like others I can't work out how the reduction actually reduced bandwidth, but it does. My only thought is that the satnav actually sends a "tell me about incident 12345" request and gets a current status. Repeat for each item. Along with a "tell me what's new" perhaps. That seems less efficient to me than a "tell me all the changes", but that can't help surely. We need more info on what is happening. |
|
| Back to top |
|
|
M8TJT
The Other Tired Old Man
Joined: Apr 04, 2006
Posts: 10118
Location: Bexhill, South Sussex, UK
|
| Posted: Mon Jan 21, 2013 4:26 pm Post subject: |
|
|
| matthewj wrote: | | ....the satnav actually sends a "tell me about incident 12345" request and gets a current status..... |
Sounds reasonable | Quote: | | We need more info on what is happening. |
I think that the expression is: "You'll be lucky". |
|
| Back to top |
|
|
rwbaldwin
Occasional Visitor
Joined: Dec 24, 2012
Posts: 55
|
| Posted: Mon Jan 21, 2013 4:53 pm Post subject: |
|
|
| matthewj wrote: | | ...My only thought is that the satnav actually sends a "tell me about incident 12345" request and gets a current status. Repeat for each item... |
I think there would be too many transactions and of course you are moving.
One way to improve backend efficiency would be to snap your current location to a grid. Doing this means you may have 10s or even 100s of cars getting the same data based on the same grid point. This means that is more efficient to send the whole data set for your horizon.
In SQL terms the overhead of even one individual update request may be greater on the system as a whole than sending a single cached request for all the data in your horizon. |
|
| Back to top |
|
|
M8TJT
The Other Tired Old Man
Joined: Apr 04, 2006
Posts: 10118
Location: Bexhill, South Sussex, UK
|
| Posted: Mon Jan 21, 2013 5:26 pm Post subject: |
|
|
| So one could have a guess at perhaps they first D/L all the data, work out where the 'horizon' is and then only request data refresh within that area? Speculation is great, isn't it? |
|
| Back to top |
|
|
MrT
Frequent Visitor
Joined: Nov 14, 2003
Posts: 2143
Location: Surrounded by A1, M1 & M25
|
| Posted: Wed Jan 23, 2013 11:02 am Post subject: |
|
|
Or maybe the software was badly written and never intended to do this?
_________________
Drivelux |
|
| Back to top |
|
|
M8TJT
The Other Tired Old Man
Joined: Apr 04, 2006
Posts: 10118
Location: Bexhill, South Sussex, UK
|
| Posted: Wed Jan 23, 2013 11:07 am Post subject: |
|
|
| But it never used to do this in days gone by, so TT must have written code to do it. |
|
| Back to top |
|
|
rwbaldwin
Occasional Visitor
Joined: Dec 24, 2012
Posts: 55
|
| Posted: Wed Jan 23, 2013 11:29 am Post subject: |
|
|
| M8TJT wrote: | | But it never used to do this in days gone by, so TT must have written code to do it. |
If I had to guess, I would say that the backend was not capable of creating datasets with a horizon based on the device when the 1005 was released. So they've done it in the system/core of the device.
For TT it is a better move than giving 1005 users 'full range' and then taking it away again when they implement it in the backend. |
|
| Back to top |
|
|
matthewj
Frequent Visitor
Joined: Apr 03, 2006
Posts: 751
|
| Posted: Thu Jan 24, 2013 1:56 pm Post subject: |
|
|
| The key to remember is that the only logical explanation of the need for the restricted range is cellular bandwidth. We know that the devices are quite capable of handling the data onboard. We know that TomTom could buy a few more servers to handle the data if that was the problem. And we also know that TomTom have a contract that was renegotiated to allow the lower price we enjoy today. This is all about keeping the units under that cap, and that tells us the scope of our speculation. |
|
| Back to top |
|
|
Andy_P
Pocket GPS Moderator
Joined: Jun 04, 2005
Posts: 19991
Location: West and Southwest London
|
| Posted: Thu Jan 24, 2013 2:32 pm Post subject: |
|
|
But if it is JUST cellular bandwidth at stake, that doesn't explain the often-reported complete download and then throwing way half the info.
I suspect a lack of processing power and/or available RAM on the unit itself is also a factor here.
_________________
"Settling in nicely" ;-) |
|
| Back to top |
|
|
DennisN
Tired Old Man
Joined: Feb 27, 2006
Posts: 14892
Location: Keynsham
|
| Posted: Thu Jan 24, 2013 4:57 pm Post subject: |
|
|
TomTom have already confirmed that they had to reduce the horizon when the bad weather hit at the end of November 2010 - i.e. too much demand, the system (HD Traffic) couldn't cope. There has never been a suggestion from TomTom that it was the capacity of the devices that was incapable.
_________________
Dennis
If it tastes good - it's fattening.
Two of them are obesiting!! |
|
| Back to top |
|
|
|
![]() Posted: Today Post subject: Pocket GPS Advertising |
|
|
We see you’re using an ad-blocker. We’re fine with that and won’t stop you visiting the site.
Have you considered making a donation towards website running costs?. Or you could disable your ad-blocker for this site. We think you’ll find our adverts are not overbearing!
|
|
| Back to top |
|
|
|
Forum FAQ
Search
Usergroups
Profile
Log in for private messages
Log in
