Just a quick note on hardware trials thus far:

* Broadcom BCM43362 (http://community.broadcom.com/docs/DOC-1422). Despite putting a Cortex-M3 in there, Broadcom really do not want you to hack on this chip. A shame since I'm trying to do it in a single component. That being said it does appear to be more than competent, if a tad expensive and with a slightly bizarre development environment. It can also - and this is really cool - be an AP and an STA simultaneously. I may yet get involved with proxying between two wireless networks :)

* TI CC3200 (http://www.ti.com/product/cc3200) is kinda wonderful. Unlike the Broadcom chip, this one is very much a single chip solution. Out of the box it runs an AP with a web application on it. Everything just works. If it wasn't for a moment of clarity in which I realised Bluetooth LE *would* be good enough, I'd be working with it today.

* Nordic nRF51822 (http://www.nordicsemi.com/eng/Products/Bluetooth-Smart-Bluetooth-low-energy/nRF51822) is another *begging* to be hacked single chip solution. Significantly down on processing power from the WiFi chips, as you can imagine, but seems at least likely to be *enough*. I've only just started working on it and while the dreaded Keil is supported there are also guides for using gcc+eclipse. Getting this going is today's work...