« 23,700,000 | Main | FW »

The Wind

(aka my Bluetooth project)

The Wind, as I have stated before, creates a complex interaction between its viewers and a field of devices which simulates grass blowing in the wind.

On a technical level, there are three components, if you will, to the installation:

1. the field itself
2. the bluetooth detection, collecting bluetooth wireless data
3. triangulation and "comments" sent to the field

I am happy about the prior achievements made in area #1, and am working to further the other areas, with a focus on #2 this semester, with some complications along the way.

Not to focus on the lack of (my expected) progress, I will say these positives about it:

  • purchase, and delivery, of 3 small computer systems (norhtec jrsx) on which to run bluetooth data-reception software. these systems are small (4.5" square, 1.3" thickness, see photo) but are full-fledged PCs by many measures, except the CPU does not have hardware floating-point. But they run quietly and on very low power consumption. Very cool! I purchased these from Norhtec, which is a company based in Thailand--the units themselves appear to have shipped directly from Thailand to my house! They cost only $85 a piece, and since I thought I may need upwards of 4 to 8 of these for what I'm trying to do, paying $200-$300 a piece for some of the alternatives seemed to far exceed my self-funded budget for this project.
    http://www.norhtec.com/products/mcjrsx/index.html http://www.linuxdevices.com/news/NS5563564014.html
  • Getting an operating system to boot on the norhtec jrsx from the compact flash device. There is no spinning disk on these systems. I decided to go the DIY route and did not purchase from Norhtec their "Puppy Linux" pre-installed on a compact flash. It turned out that was a mistake, at least in terms of time and effort. But I had in mind to run a BSD operating system over a Linux distribution. (Old loyalties never die!) I tried FreeBSD, OpenBSD, NetBSD, DragonflyBSD, more than one version FreeBSD, in fact; these all had trouble because of the lack of floating point hardware. Then I started various Linux distributions: Knoppix, Ubuntu, DSL, Puppy Linux (the last two in fact are the ones the company runs), and had other issues. And finally Debian Linux. I don't know why I didn't at first go to Debian, since that is the basis of many Linux systems, and is the core operating system promulgated by the GNU set of projects. My impression was that it was "too big" whatever that might mean. I had to fit an entire system on a 512MB flash drive. HOWEVER, Debian is the one that worked right out of the box. Now, given *more* time, I am convinced many of these other distributions, even some of the BSD ones, would work, but I was already running low in time. I spent a considerable amount of effort just getting this to work.
  • I have a reference Debian Linux system (4.0 Etch) setup on which I can generate the Compact Flash images for booting, and the process runs fairly smoothly. The process is scripted somewhat, but I'm improving on it yet. At this very moment, due to other reasons, my reference Debian Linux system has become unavailable to me (other emergencies have taken over). This means progress in this direction has been momentarily suspended.
  • I have purchased three different bluetooth devices (dongles, if you will): Trendnet TBW-105UB, Zoom 4311, and iDOTconnect 9017. The idea behind this was to research a little on which brands worked the best for the usage I intend. (Since I'm not doing the mainstream thing with this but instead measuring signal strength, that is something that seems just outside the normal or standard. At least, the RSSI command does not seem to be widely implemented, nor documented among the various devices available.) This, like discovering which OS would run on my jrsx system, is like another research project. However, I don't have to perform these tasks on the jrsx system itself; instead I can run these tests on my reference Debian Linux system, and only do some final testing on the jrsx boxes. I have started this process on FreeBSD, only because my Debian reference is unavailable, and so I am not sure the drivers and results will be useful knowledge toward my end goal, but learning about the bluetooth dongles themselves and their capabilities, I hope, will. I have started with the iDOTconnect and Zoom products first, and at least FreeBSD recognizes them at a low-level driver point of view!
  • I have created a context diagram which I think depicts more clearly what I am trying to do with this project. See below.

ContextDiagram.jpg