Installation of the
Wireless Sound System in Michigan

Dr. Stuart Gage, Michigan State University is the leading Biologist who has, for decades as part of systematic scientific inquiry, recorded, stored, and analyzed natural sounds as a dimension of the Environment. He has been compelled, for lack of scalable technological alternatives, to collect most of the sound readings manually. Either by placing recording systems in remote areas, writing the data from microphones to tape recorders sampling every 30 minutes, or operating very short range analog radios and collecting the data, that way. Then, by a variety of steps he is able to convert these sound recordings into wave files, and either store them on writable CD’s or, as currently, transfer these large files via Hughes Two Way Satellite system to the terabyte-storage facilities at MSU, or move them over a cable modem Internet connection.

Dr. Gage, observing the work we have done in this wireless project communicating visual and numerical data wirelessly, recognized the desirability of connecting up the output of microphones which capture natural sounds placed in field locations, to the Internet in real time.

It was clear this would require devices configured to:

a. Turn the analog signals from field microphones into digital form.
b. Stream the digital sound into digital radios at sufficient bandwidth for transfer.
c. Transmit the digital stream to the closest Internet POP, or do a lab station, wirelessly.
d. Deliver the digital sound stream to either a data capturing device or into audible sound via the net.
e. Contain the sound digitizing system in a waterproof, renewable power source box, that can be mounted close to the microphone(s) and the radios, in the field. And make the system as low cost, and low power consuming with small footprint as practicable.

Mike Willett made a Site Survey trip to Lansing, Michigan to determine the specs for this system. His report accompanies this one, which describes the design and installation of the Digitized Sound System.

The Name of this Project

We have been able to accomplish the tasks outlined above by designing and deploying a prototype system on Dr. Gage’s small field laboratory near Lansing, Michigan, getting it up and running on September 3th, 2002.

Technical Concept

We decided to assemble a small footprint system which could reside in a waterproof box, with solar/battery power outside the box, and standard mono or stereo microphone coming out of it.

At first Mike Willett thought that a one-board system made by Intrinsic, of Canada would work. But close examination of their system showed that their processor and operating system were proprietary. Everything produced would have to be compiled for that processor. Requiring much custom programming.

Instead he found that Win Systems, Dallas Texas, produced standard PC104 boards, with an Intel processor, which could be assembled and use standard software. With that we could run a very small Unix (Linux) system with Icecast public domain software capable of digitizing an audio stream into Mpeg3 files.

PC104 Boards are generally 3 inches on a side, and use 5 volts.

The idea was to construct a small, relatively low cost, custom system whose only function is to digitize microphone sound in a field environment and wirelessly communicate it, interactively, to the Internet.

The System Deployed September 3d and 4th, after two months development work by Mike Willett’s OMC company, consisted of:

1. Sound Digitizer system - stacked 3 inch PC104 Plus boards from Win Systems
a.Pentium 266 Mhz Intel Processor board with pinouts for Ethernet, Serial, and USB
b. Soundblaster compatible Audio board, with 5 standard 1/8inch I/O ports. With Maestro sound chip.
c. 256Mb SD Ram board
d. 1gb Ibm Microdrive (1 inch square) in flash card
e. AC-DC +12,-12,+5 volts converter board
f. Video board (this last is not needed, if the operating system is configured to boot without the usual keyboard and video boards. It consumes too much power.

The Entire System, with PC104 Boards

Configured Field Unit

2. After being configured while being connected up to the Development System (a system with power supply, CD, Floppy Drive, keyboard, mouse, and video connections, the 6 stacked boards were put in a waterproof Nemo box, the various leads coming off the boards to connect to power, radio, external microphone.

3. The operating sound-digitizing software applications were installed on Linux Red Hat 7.3 which was successfully installed within 700mb of the tiny 1gb microdrive, leaving 300mb of space for digitized sound files.

4.The digitizing application software was public domain LiveIce - which digitizes analog sound coming in via microphone into MPEG3 files and IceCast which can stream out the MPEG via the Ethernet port. The software was configured to allow up to 5 streams, and to write, to the microdrive, 24, one minute captures of microphone sound - about 450k each. Each time-stamped so when transferred out of the drive by ftp, they would retain their time-of-capture identity when stored elsewhere.

5. A small one-mac address Aironet 802.11b (called Turtle) radio, with short antenna, connected to the ethernet coming off the board stack, was placed inside the box with the stack, and power connectors.

6. All of these components were mounted in a waterproof case. For this trial, power came from a long extension cord carrying 110v through a transformer to feed the ac to dc board. Later a solar panel, battery system will be used.

Mike Willett and David Hughes doing final configuration

Dr. Gage with one of his field microphones

Base Unit

7. A second, older Access Point 4800 Aironet radio was placed inside Dr.Gage's computer room, line of sight to the sound system in the field.

8. In turn this radio,via Ethernet cable was connected to a NIC card on Gage's Window's 2000 machine used to store, analyze, and convert sound files of various formats, and to upload data via Hughes Satellite Internet service to MSU labs over the Internet.

9. A Windows ftp server software package was then downloaded via the net into Gage's 2000 system,so that the digitized MP3 sound files could automatically be pushed from the Microdrive in the field, into a directory on the Windows machine, via the wireless connection as soon as they were produced, hourly, by the Digitizer.

Mike Willett making the first Installation

Operation

10. A series of script-driven utilities were designed by David Hughes and mounted on the 2000, including the highly flexible ‘WinAmp’ sound software, into which the streaming audio could come via wireless, as MPG3 files, and played over speakers or into headsets.

11. After considerable tweaking of the system, it started collecting the natural sounds coming from a pair of microphones close to the Box, writing the files and then ftping them automatically into Gage’s 2000, while simultaneously the sounds were fed into speakers in his lab to be listened to.

12. The system collected satisfactorily all the first night.

13. Because of routing and interface complexities between Gage’s dual-homes 2000, a cable modem, and the Hughes Direct PC satellite system, we were unable to feed the sound files automatically through the satellite Internet system into his lab at MSU – an eventual goal.

The Hughes Satellite Dish at the Gage Lab

14. There were some odd breaks in the live sound, with WinAmp rebuffering, although the files written to disk had no such problems. The problem seemed to be somewhere in the linkage between the wireless, the 2000, and the sound program. We have yet to track that down.

Future

Upon return from the installation trip, we were informed that the first full 23 hour captures were installed on the MSU Lab servers.

http://www.cevl.msu.edu/envirosonics/locations/gagehome_history_nsf.htm

15. Dr. Gage cannot directly computer-analyze MPEG3 files. They have to be converted to Wav files, which can then be analyzed. He has many tools for doing that in his lab.

16. A better solution to trying the use a single, dual-homed 2000 as a Windows ‘server’ has to be found. A small Linux router may be the answer.

17. A complete field rechargeable power supply system for the field unit needs to be assembled, and tests run to see how many amp-hours are needed.

18. Other, single or two-board alternative PC104 systems, requiring even less power will be explored. The Win Systems board sets worked, but Willett believes there are better, lower power using, and cheaper, solutions. The developmental system, plus one complete board set cost about $3,500.

19. It is clear a network of such field digitizing/communicating sound systems could be deployed, using point to multi-point radios. Since 2.4ghz 802.11b radios cannot penetrate trees well, we may experiment with a new 2.4ghz to 915mhz converter by Teletronics. This system converts 2.4ghz 802.11b signals to 902-928mhz. Which would permit Ethernetted 802.11b radios at each end, but communications via higher gain 915Mhz radio, penetrating trees, in the field. At high data rates. A hybrid solution, but one which takes advantages of the ubiquity and low cost of 802.11b radios (even putting only a PCMIA card radio on the field stack, with pigtail to a 24dbi antenna) the penetrating capability of 915mhz wireless, and taking the digitized sound files or live feed off via the Ethernet port of the base 802.11b radio.

Dave Hughes, PI

PREVIOUS

NEXT