Large Scale (250+ module) Wireless network

Hello. My name is Nic and I am an engineering and physics student at Rensselaer Polytechnic Institute. I am working on a tiled smart flooring system which requires a wireless module per tile, so as to eliminate the need for physical connections between tiles. I am unfamiliar with wireless networking and am wondering what the best module would be for my application.

Essentially, each tile will have a microcrontroller and wireless transmitter (along with various other electronic components) and will upload activation data to either the internet where the data will be displayed on a webpage, or the data will be forwarded from many secondary tiles to one primary tile which will then upload all of the data to the internet. My near-future installation is planned to be approximately 50 tiles with an expanded installation of 250+ tiles within the next year. I am wondering what wireless module would be ideal for this application: XBee Series 2, WiFly, etc…

From what I understand, the ZigBee protocol would be inefficient for my installation…

Thank you for your help. It is very much appreciated.

-Nic

From what I understand, the ZigBee protocol would be inefficient for my installation…

What is it about the XBee protocol that you feel would be inefficient?

along with various other electronic components

What other components might these be?

and will upload activation data

How much data? How often? How far?

to either the internet where the data will be displayed on a webpage, or the data will be forwarded from many secondary tiles to one primary tile which will then upload all of the data to the internet.

XBees are radios, not internet devices. A XBee can be connected to an USB Explorer board, connected to the PC, via serial port. That PC can be running an application to take data from the serial port and store it in places/ways that a web server could access. Expecting one device to receive data from 250 other devices may be unrealistic, depending on how much data is involved.

You are aware, I hope that an XBee + XBee shield + Arduino per tile will run you in the neighborhood of $80 per tile. 250 of them will put a big dent in your wallet.

How critical is the collection of the data? If some got lost, would that be catastrophic? If not, cheaper radios exist.

Could some of the tiles be wired together, so that one Arduino, for instance, collected data from 4 tiles? That would save you money.

What is the Arduino actually doing? Many sensors can be connected to the XBee directly, with no need to have an Arduino involved at all.

I was reading the book "Building Wireless Sensor Networks" and it mentioned that AODV mesh routing would be inefficient for networks consisting of more than 40+ nodes due to repeated routing requests.

The other components include piezoelectric plates, an accelerometer, ultra-bright LEDs, and several capacitors. The floor harvests energy from pedestrian and vehicular traffic (it's been tested and produces a max of 45VDC at 7 watt-seconds per footstep) and is intended to serve as an energy harvesting system which will charge throughout the day and be used to power lighting and displays around campus at night.

The data being transmitted will include voltage and current output of each tile, along with footstep detection and the amount of energy currently stored. The data would have to be transmitted a maximum of 250ft every time a tile is activated (so the "how often" will vary depending on how busy the installation is).

The data will be used to monitor pedestrian traffic on campus by indicating which tiles are currently active. Also, eventually the tiles are planned to be connected to the lighting system on campus so as to provide proximity lighting only when certain tiles are active, thus saving power when no one is around.

The only thing the Arduino is doing (currently a Pro Mini) is reading the input from the piezo elements to determine voltage and current, then directing that data to the wireless transmitter. I was unaware that sensors could be connected directly to the XBee without the need for an Arduino.

Yes, connecting 5 tiles together is definitely an option (one in the center with 4 surrounding). I'd still like to gather data from the other tiles and that would reduce cost dramatically.

I guess what I'm looking for is an effective way to relay the activation data from the tiles to either the internet or another computer system which can be used to monitor tile activations and use it to control lighting, heating, cooling, and ventilation systems (for an expanded installation).

So, I guess I'm more confused now than before. If the tiles generate electricity, that electricity is not local to the tile, is it?

If not, then I don't see why the data has to be wirelessly transmitted anywhere. From the collection of tiles, I could understand, but not from each tile individually. If the tiles are rectangular, a tile has 8 adjacent tiles - 4 share an edge, and 4 share a point. It would be awkward to try to lay out a grid of t-shaped tiles if only those connected along the edges are tied together.

Perhaps you could separate the data into two packet types - one being a very small packet that says, in effect "Tile n was stepped on" that was sent whenever a tile was stepped on, with the other being a larger packet that contained the voltage and current and energy stored data that was sent at much less frequent intervals.

The book you are referring to barely acknowledges that Series 1 XBees exist. For them, the routing requests are not an issue, since they do not also act as routers.

ZigBee will work for 250 modules. Digi routinely tests networks of up to 1023 modules in special racks of equipment in Utah and Minnesota designed expressly for those simulations. In fact, look for a cool new video about that coming out in the next week or two...

In the meantime, check out the Routing section in your Building Wireless Sensor Networks (http://www.sparkfun.com/products/10324) on page 240. Depending upon your network you would probably want Many-to-one Routing or Source Routing rather than AODV. Add a commercial gateway like the ConnectPort X2 for Ethernet or the X4 for mobile data if your installation is going to be permanent. It will be way more reliable than using a PC, though for prototyping a PC should be fine.

Good luck! The project sounds like it would be an amazing and fun interaction.

Woo hoo! Lets Go Red!

Another option might be low cost transceiver with a low cost arduino (see the mini-uino in my siganuture link), with 1 master polling each one for status in a big loop to keep some semblance of order. http://arduino-direct.com/sunshop/index.php?l=product_detail&p=188

RPI class of '85 ...