Friday, July 18, 2008

The last first impression

Actually, I wanted to do this last week, Friday, June 11th. Unfortunately, it took PosiMotion a whole week to publish their G-Spot. Anyway, it is available now and I had a closer look to it. To be precise, I had a close look about what the location provider provides, not what G-Spot does.

I startet the application inside the house. This didn't just give me the usual cell tower based location with an imprecision of 1799 meters, but also another funny bearing of somewhere around 77° N, 12something° W with an imprecision of 5000 meters.
This is an interesting issue, as Google Maps on my other cell phone sometimes does the same when I try to determine my position by not using GPS (the only difference was that with Google Maps I was supposed to be somewhere off the russian coast, just nordwest of Japan).
I wonder how this can happen anyway. The cell phone can "see" several BTS at a time, all identifying themselves with several parameters, including MCC and MNC. How could anyone ever think that there might be a BTS with MCC 262 right in the middle of the arctic ocean?

Being outside, a 3D GPS fix could be acquired really fast, so the download of the latest almanach seems to work (and also to speed up acquiring satellites, as it is supposed to be).
While travelling, accuracy readings kept jumping between 47 and 17 meters, when going real slow (like in a jam) even 9 meters of accuracy were reached. Those readings are a little disturbing, because usually you should get readings in between.
Unfortunately, I don't know how G-Spot works internally, so I don't know if they just took "raw" accuracy readings or not.

If you want me to investigate further, just donate $99 to me so that I can get an appropriate SDK license to.... just kidding ;-)

47 meters is bad, real bad. 17 meters might be okay for road navigation, but you have to do some tricks, so I'm still curious how (good) TeleNav will solve this problem. Nine meters would be okay, but only if you had that accuracy all the time.

Which leads me to another assumption: I don't think that WAAS (or EGNOS resp.) is enabled because of power consumption. There might be other power saving technologies used that degrade accuracy as well.
Unfortunately, I don't have an answer to my inqiry to Infineon yet, so that I don't know what the chipset is capable of. Perhaps, it can't even use WAAS?

No comments:

Post a Comment