Using several Raspberry Pi computers to create a household Internet of Things

Question

I am interested in the concept and deployment of the Internet of Things. One of the primary requirements is for cheap computers and pervasive networking. I want to know if the Raspberry Pi in its present form or a forthcoming model would be suitable for this kind of task (I know next to nothing about the device). Suitable equipment would need a general purpose processor capable of running TCP/IPv6 and probably also a HTTP stack or an implementation of the ZigBee API. Some kind of dirt cheap, low capacity non-volatile storage (e.g. flash). It would need 802.11 support compatible with common household networks (a/b/g/n/ac/ad) and the ability to self-organise into a mesh network and perform message propogation.

Does the Pi come anywhere close to this? If not, please detail what is missing and what is coming down the pipeline from the manufacturers. If it does meet all these requirements, please provide an example if you can of an existing deployment demonstrating its use.

Answer 1

The Raspberry-PI misses several features that you want (802.11 support, Zigbee API implementation).

However, it is possible to add them to the PI.

First there are some boards that extend the PI capabilities (see PiggyBackboard and its demo). Cost of the board is around 23€.

Second, there is a Debian distribution that is actively developped on the PI: Raspbian. It is Debian, so it works like a Debian and you can manage whatever you want, almost like with a "normal" PC. As there are projects like this one (linKnx) that is already available for the PI (see this forum).

About Zigbee, I also found this, but I don't know if it works for the PI: http://www.open-zb.net/

Now, there are also a few things you should be aware:

1. A Raspberry-PI board needs at least 700mA power supply. There are several issues due to lack of power. Also, this means that each board consumes at least 3,5W. This might limit you depending on what you want to do. Minimal current power is recommended to be at least 1000mA.
2. If you want to plug power consuming USB devices, you will need to buy an USB hub.

As noticed by TomG, cost of equipment might be important. But you will have a lot of work to do to integrate all of this together. And don't forget that RPi is not an industrial solution (like ALIX system boards for instance) and reliability of the board may not be adapted to your needs.

Answer 2

You can try HomeGenie on Raspberry Pi:

http://www.homegenie.it

supports X10, Z-Wave, Philips Hue and much more. It also allow control components directly over GPIO/SPI/I2C and it has a powerful automation program editor supporting Arduino Sketch, C#, Javascript, Python and Ruby... see the above link for further informations and examples.

Answer 3

I'm currently building an application core for my smart home project - there will be 6 to 10 devices on the network each controlling its own area and performing different tasks and operations - security, notifications, sensors reporting, controlling lights and devices, et cetera.

RPI can be treated as a regular computer able to run regular and odd operating systems. It does meet your requirements although you seem to mix hardware and software requirements into a single stack which is bit odd.

It is not clear from your question what the primary focus of Internet of Things you are attempting to build. I see two key fields Internet of Things is valid for - knowing/sensing and controlling/operating. Both are possible to a good extent with RPI in its current hardware and software state.

Sensing and computing for anything complicated may not be always possible with RPI - things like face/voice identity and recognition without assistance of other hardware will be either slow or impossible, reading data off sensors and controlling servos and lights is easily done - most of stand-along solutions for this types of tasks being sold are done using hardware that's way behind RPI specs.

I'd suggest that you identify your key priorities and look for example implementation with RPI.