| |
| ProfessionGPS |
|
|
| POIenFrance |
|
|
| Visiteurs |
Actuellement en ligne: 195
Dont membres: 0
+ d'infos...
|
|
|
| GPS Receivers for PDAs and PCs in 2005 |
3. GPS Performance and TTFF
Introduction
This part has been a long time coming while I was logging dozens of simultaneaous raw tracks but also because the current generation of GPS receivers for PDAs and PCs largely takes after the performance of the GPS Chipset that "powers" them and these were convered in great detail in the Chipsets and Receivers Comparison Article earlier this year. It includes performance analyses for the SiRFstarII, SiRFXtrac, SiRFstarIII, uNav, Sony v2, Garmin, Nemerix v2 chipsets in various environments, on the Road, behind a coated windhsield, on foot and even indoors. As a result, the current page will focus on comparing SiRFstarIII based receivers currently the most popular family of receivers as well as less well known receivers using different chipsets like the Solarius (Trimble), the Rikaline 6033/Apos-Tec GPS26 (RFMD) or the Apos-Tec GPS36 (brand new Sony v3 chipset)
About TTFF : "Cold, Warm & Hot" Fixes
TTFF, or "Time to First Fix" is the time it takes for a GPS to determine its current position, the interesting ones being a "cold fix" when it starts from scratch with no valid ephemeris information stored in its memory and a "hot fix" when getting back on its feet after losing the fix in a tunnel for instance. Both vary quite a lot depending on the immediate environment but fairly little from one chipset to the other, here's why.
There are three steps for the GPS to get a fix : acquiring the signal, tracking the signal and downloading the navigation information (the ephemeris data that remains valid for 4 hours). The downloading can only occur if the GPS receives a minimal level of signal from the satellite of about 28dB-Hz (so no "cold fix" in most indoor environments), and at this level the signal will be clear enough so as to not require a lot of processing power for the "correlation" (acquisition and tracking) to occur and the fix to be established. This means that for a "Warm" or "Cold" Fix, TTFF will be relatively close from one GPS to another with 35/40 seconds being the minimum time required out in the open, but with large variations when the environment is less favorable. A "hot fix" will be anywhere from 2 to 10 seconds with a good view of the sky in a moving car. GPS receivers based on a powerful chipset like the SiRFstarIII or RFMD will be less subject to these variations I've found, especially when trying to establish a "cold fix" in a moving car. On the other hand these chipsets can "track" at much lower levels than the previous generation as seen in the Chipsets and Receivers Comparison Article.
As a side-note, it's clear that the limiting factor here to getting a "warm" or "cold" fix in obstructed environments (indoors for instance) is having to download the ephemeris data from the satellites and it can be overcome with A-GPS, as this consists in downloading this same data via a GSM/CDMA network ("control plane") or directly via GPRS/UMTS ("user plane"). So far, the only available "user plane" solution is on the HP 6515 (reviewed here) that cannot leverage it totally as it has a tiny antenna and uses the Globallocate chipset that has less correlation power than the SiRFstarIII or RFMD chipsets. Hopefully there will be a similar solution based on these chipsets in the near future. UPDATED 02/2006 - SiRF have just announced their own "predictive ephemeris" solution called SiRFInstantFix, read on for more details.
SiRFstarIII based Receivers
This is a recurring question in the forums, "what"s the best SiRFstarIII GPS receiver"? It's a fairly easily question actually, there isn't any "best receiver" as far as performance goes! There are a few variations in behaviour as can be seen below, but the level of performance is stellar and miles ahead of any previous consumer receivers with an excellent overall level of comfort.
Below are side-by-side comparisons of raw tracks made with the following receivers, the Globalsat BT338 (Semsons M3), Holux 236 (Guidetek 238), Fortuna GPSlim, Leadtek 9533, Royaltek RBT2001, TomTom MkII, SysOn SmartBlue and Haicom 305iii.
   BT338, 236, GPSlim and 9533 in the small streets of Nogent, behaviour is nearly identical for the first two, the 9533 and the GPSlim are a little less accurate - In Paris on the right, the GPSlim produces a better result though than the 338 and 236   On the left, you can see the TomTom MKII tracking off compared to the 338, 236 and GPSlim, on the right the 338 and SysOn perform on par in the streets of Paris with the Xtrac v2 based Navigon TriCeiver sometimes struggling to keep up   "Tunnel endurance" testing in the Nogent tunnel, on the left the 338 goes further than the RBT2001, in the middle the 236 and the TomTom perform similarly while the 338 and GPSlim power away, on the right the Haicom HI305iii comes out on top !
Other GPS receivers:
RFMD/Nemerix/Ublox/Xemics/Sony/Xtrac v2/HP6515 AGPS
In light of the "power" of the SiRFstarIII brand, well justified by the performance in the field, one might ask if there is any room left for non-SS3 receivers ? When the Chipsets and Receivers Comparison Article was published earlier this year, the answer would probably have been a "no", but the arrival of receivers like the Rikaline 6033 (aka as Apos-Tec GPS26)based on the new RFMD chipset and the Apos-Tec GPS36 based on the new Sony v3 chipset put up a decent fight, let's take a look.
   No less then 8 receivers logged simultaneously above with a zoom in the middle on the Solarius Xemics and the Nemerix v3 that don't track as accurately as the SiRFIII receivers, on the right the Delorme LT20 based on the ST Micro Vespucci chipset does well   On the left you can see the progress Nemerix made between the v2 and v3 but it still has the occasional "lapse", the Socket v2 based on the uBlox Antaris is also hampered by the occasional "jump" and lost fix   The Rikaline 6033 loses the fix fairly quickly in the Nogent tunnel and you can see in the middle that it filters the position when used on foot (similar to SN on SiRF based receivers but it can't be disabled), but on the right it does just as well in the narrow streets of Strasbourg behind the infamous coated windshield of the Citroen C5   Now for some testing across the pond in Manhattan where the ClipOn Xtrac v2 just can't keep up with the BT338 while the Rikaline 6033 is a bit "wobbly" in the middle compared to the BT338 and BR355 mouse GPS, but still does ok and on the right it even does better approaching the Lincoln Center, seemingly dealing better with multipath thant the SiRFIII units on that occasion   The good surprise of the fall is the AposTec GPS36 based on the new Sony v3 chipset that does almost as well as the BT-338 on this urban trip while the Polstar based on the Sony v2 had performed very poorly (see the chipset comparison article). The HP6515 does show its limits here, likely due to its tiny antenna
If you have questions or comments, you can use this discussion thread in the forums
|
|
|
|
|
|
