A secodn GPS test, this time with the 405cx on my left wrist and the N95 in my right hand. I had hoped this would expose the N95 antenna and improve its performance. Not so. The red trace (405cx) still looks much better while the yellow one (N95) jumps wiggles and snakes around. The summaries are:
N95: 8.30km in 53 min 13 s, 727 kCal
405cx: 7.92 km in 53 min 31 s, 627 kCal
If you zoom in, the N95 has me walking on water. To be fair the 405cx also looks a bit lost at the beginning of the run (far right), but it does get the street-crossing spot on (middle-right in the pic).
Then for a straight bit along 'Nokia avenue':
That's not pretty. I guess you could argue that on the way out (top) the street has tall houses on both sides, so the N95 (yellow) has a hard time finding satellites, while on the way home (bottom) it's more open and both GPSs perform roughly the same.
Here's part of the route in more open terrain. That seems to confirm that the N95 needs wide open spaces to do well.
For the record, my N95 8GB runs V 15.0.015 firmware (which I've several times unsuccessfully tried to update with Nokia PC-Suite 🙁 ) and I'm using Nokia Sports Tracker v2.06 (S60 3.1). Today I had GPS-filtering turned off in sports tracker.
Somewhat surprising results! I would have thought GPS-positioning is a plain-vanilla technology by now and every company does it with equivalent hardware and software. Apparently not, or the N95 antenna is really inefficient, or I'm doing something wrong...
Interesting stuff! Does N95 save hdop/vdop/pdop information to see the quality/precision of the satellite signal received?
I confirm your findings. I've even thought of doing some auto-correctioning for this N95 undershoot in curves, it is so deterministic every time. You can see how badly it behaves after curves if you take a 90-deg turn and then go back and forth along that path, the path only slowly converges to the right track while one would expect the back-and-forth paths to match rather well. Another "feature" of N95 is that it starts to lag behind so that the lag accumulates when you go to forest area etc.
However, it seems that in bicycle speed the quality of N95 trace improves significantly (at least for me), for walking you'll need really clear skys.
I suspect all these are artifacts created by some too "intelligent" thing happening in between GPS raw data and actual output given to the applications. It seems that other explinations fail to explain why the device couldn't recovery nearly right away after a curve when under clear skies. Whether it's os or chip doing, I don't know.
Did you remember that the Nokia N95 assumes that you keep the lid open all the time when you are using the internal GPS feature? This is influence of poorly antenna design and positioning. I got tired of this and did this mod: http://www.noeman.org/gsm/hardware-firmware/64357-hardware-hack-improve-n95-gps-reception.html
The GPS reception is now far more better and the tracking works significantly better (even the lid closed) from my pocket.
awesome mod Lauri! I will have to try that with my own phone also at some point. And compare again to the Garmin...