Using WIFI for as Master Slave Instrumentation

[robert.kraml]

Hello,

Keep in mind that I am a novice regarding this technology and I may be chaing blue sky here.  If so, please tell me this is a non starter.  Also, I wasnt sure what category of wirelss this fits into so I posted this here.

I would like to explore the feasibility of a sensor net like function using wifi that will hopefully let us use on the order of hundred (maybe up to 600) or so WIFI devices in an instrumentation application.  I am looking at 802.11 type devices instead of the conventional "sensor networks" mainly due to bandwidth and signal strength considerations and where power consumption is less an issue.

We notionally are thinking of using a multitude of these devices in a master/slave type configuration over either asn acces point or an adhoc configuration (whcievber makes more sense) whereby a single “master” will sequentially and repeatedly query all the slaves on the net for data.  We intend to avoid collision events by enforcing a rule that all slaves are quiet until queried by the persistent designated master.  No dynamic network functions (like Zigbee) are required.  There is only one ever one master and it is persistent indefinitely.  Also, the size of the data sets is notionally "small" (think of a small set of 12 bit A/D samples) and  shouldnt require multiple packets.

A demo ssystem might consist of a laptop hosting the "master" and a handful of laptops hosting cascaded USB hubs with 802.11n WIFI dongles plugged into them.  Is it out of the realm of possibility for the master to be able to query 500 connected clients for 32 bits of data at a rate of say 100 Hz?  Thats a raw bandwidth requirement of 1.6Mbps but can the hardware (and software) handle it?

I need to take into account concerns such as the max number of clients that a device can address, the overhead in volved in querying and waiting for retrun data from each client, the max number of USB devices that can be on a single PC with maxed out expansion devices.  What kind of overhead is involved (size and time) when a device queries another connected device for data.

Again, I look to folks to tell me if I'm dreaming or there might be another better solution.

Thanks in advance.

Robert Kraml, Northrop Grumman Aerospace Systems

310-812-3799

Robert.kraml@ngc.com

[Leo Laohoo]

I would like to explore the feasibility of a sensor net like function using wifi that will hopefully let us use on the order of hundred (maybe up to 600) or so WIFI devices in an instrumentation application.

What kind of "sensor" are we talking about?  Do you mean what Boeing is using whereby assets and parts are "tag" for location tracking?

[robert.kraml]

No this would be an array of stationary data acquisition devices placed within a specified volume on the order of a few hundred foot radius to monitor processes and events such as vibrations, shock, temperature strain, etc.  These devices could have local buffering which would store up sequences of data that would allow less frequent transmissions of a greater amont of data if teh overhead of a round robin connection scheme is too great.  For example, instead of a 500Hz 16-bit sensor transmitting samples 500 times a second, the device would transmit 500 buffered samples once per second.

[Leo Laohoo]

Try to contact AeroScout.  They have some very "funky" tags.  They'll be happy to hear from a company like yours.

[robert.kraml]

I can do funky.  I will check it.