(include notes from hw meeting on 20020806 + brief summary of design descisions/options/etc+ –nik gaffney)

Work is continuing on a new wireless sensor aquisition system, to hopefully overcome the problems found with all the previous systems. The current version is still built on a series of comprimises between price,availability,elegance,wirelessness and flexibility, however it should be edging towards something better.

Hardware setup

• CerfBoard (strongARM) see: Cerf Notes
• 3d acceleromter units
• WCF12 802.11b compact flash card
• sony info lithium battery + 5V regulator

Software setup

• i-Linux (instrinsyc's branch of familiar from http://www.handhelds.org )
• adxl drivers

• calibration + sensing code (gpio → OSC )

details can be found in the techspec (linked from Project Txoom System ), code is in cvs, and notes can be found in the Cerf Notes node.

datarates.

• sampling periodicity: ~10ms/vector or as close to 4ms/point as possible (100-250Hz)
• latency: should be as low as possible. if a vector takes too long to store, send points.
• thruput: accelerometer data is single byte (8bit) per point.

calulations for 2x3d acceleromters (for reference)

• 8bits x 6axis = 48bits/0.004s = 12kbps (at 4ms sampling rate) per player.
• 8bits x 6axis = 48bits/0.01s = 4800bps (at 10ms sampling rate) per player.
• 8bits x 6axis = 48bits/0.02s = 2400bps (at 20ms sampling rate) per player.

(note: raw data, no correction, or stop bits)

                                                                                                                                                                                                                                                             
                                                                                                                                                                                                                                                             {{sensor_monitor.png}}

see also CerfNotes for more details

wired connection

• prerequisites
  • ceed
  • sensors + cables
  • network with DHCP server
  • power suppply or batteries
• attach ethernet cable to the ceed box
• powerup ceed
• wait for a while,
• find its ip
  • run nmap or similar to find which ip# it has …
  • eg. # nmap -sP 192.168.254.1-15
  • check which ips are in use before + after turning the ceed on
• check there is something listening to port 5333 on the host machine (eg, 'oz', 'nc -lup 5333' or an OSC capable program)
• connect to the ceed # ssh -lroot 10.0.0.x
• load the adxl module # insmod adxl.o
• run the sensor program in test mode # ./operate -c debug.conf -h <host.ip>
• check which chanels are active
• edit the operate.conf file if required.
• run the sensor program # ./operate -h <host.ip>
• sit back + relax

wireless connection

• make sure the base station is active
• check SSID set to f0am (using ap-config under GNU/Linux)
• contine as above from the 'powerup ceed' step

see also IPaq Notes for more details

these notes are meant as a quick guide to setting up the ipaq + sensor sets for the tgarden/tx0om system(s). insprired by my lack of memory, and the fact that teh wearables can be fickle creatures, .. .

so,

• connect access point to network
• turn on ipaq
• press the “Q” button to boot linux
• open 'Utilities → Input Methods → Xkbd' by clicking on the screen
• open 'Root' window by clicking on the screen
• (optional) type 'bash' for smoother shell interaction
• type 'ifconfig' and see if there is an entry for 'wlan0'
• type 'pump' and check ifconfig again.
• if there is no entry for 'wlan0' device, type…

ifconfig wlan0 192.168.254.3 netmask 255.255.255.0 up 
(substituting the suitable ip# and mask for the network)

• check ifconfig again. try to ping a host (using ip#)
• if there is no route to host, try typing…

 
route add default gw 192.168.254.1 
(or whatever the ip# of the access point is)

• type 'route' and check the routing table is correct
• dont bother with /etc/resolv.conf unless you must
• it should now be possible to connect to the ipaq using ssh

now, the sensors…

• make sure the stamp is charged + sensors connected
• on the ipaq, enter 'cat /dev/ttySA0' to display data comming in the serial port
• connect the stamp to the ipaq serial port
• sit back + watch the numbers,. ..
• if there are no numbers, there is a problem (find it + continue)

calibrating…

• cd to 'sensors' directory of $HOME
• calibrate -c &<sensor number>
• check the min + max values for each sensor channel

operating…

• operate -h <host running oz>
• sit back + relax

see also: Mobile Computing, Wearable Computing, Sensor Technology, Human Interface System