Logging Bluetooth Accelerometer Data on a PDA?

sessha asks: "I want to log data from a Bluetooth Accelerometer (actually 5) attached to different parts of the body. How one would go about logging the data onto a bluetooth compatible PDA. I know quite little about PDA OS's, but the main concern is functionality - it doesn't matter if it is PalmOS or Windows CE or whatever. The Linux and Windows directions for data logging are shown, but how would one best go about this on a PDA? Also, it would require at least 256 MB storage, preferably more, since this will be recording 300 of these ASCII strings a second for a number of days. Suggestions as to the best equipment and software to use for such an endeavor would be greatly appreciated."

Comments

Re:cheap solution: sharp zaurus 5500

[Bronster]

the 5500 is a pocket linux workstation. it will run most linux software, particularly the non-gui/server oriented stuff, just fine.

Pity it only has a battery life of an hour or so when playing MP3s according to the Unofficial Zaurus FAQ. I guess you might get another hour if logging strings, especially if your app is designed to do it without too much load on the system.

Whoa!! Where can I get some of these!?

[slithytove]

I've poked around the website linked from the story, but I see no mention of how/whether others can aquire these units. In particular, the FAQ is, well, lacking ;)
I've been looking for something like this for a loooong time! The closest I've seen are serial devices in the $500+ range. These would be perfect for a number of applications I've envisioned.

Compress that data

[AlecC]

The data has 100% overhead from being in ascii, not hex. You can halve the data storage by converting it to binary immediately. You don't need to store \n - that is implicit in the fixed structure. The first field will generally be simply countin, so you can omit it completely if you can inser an extra code for skipped samples. What coherence is there between successive samples? If sampling rate is high, most samples will be quite similar to their predecessors, and can probably be mor compactly coded with a delta code not an absolute.

I would have thought that you could cut the data to be stored by a factor of 10 by some thoughtdul, but not very complex, coding. Normally, I would advise againse excessive attempts at compression. But this is eactly the kind of bulk data application where compression works.