AllJoyn seems to be a kind of everything-to-everyone sort of system, how reasonable is it that it’ll be implemented in the smallest of microcontrollers? Is it modular enough that only certain aspects can be implemented?
Is the underlying protocol simple enough that it could be implemented, for example, on a 16MHz 8-bit AVR?
Supposely the additional processing it requires is pretty small for individual devices. That’s why it can be put into other devices for minimal existing cost. WiFi and the processing power it requires tend to be the main costs.
But the same used to be said about microcontrollers. Now they’re in just about everything.
Normal LED replacement bulbs have quite a bit of circuitry in them as it is - step down transformer/bucks etc. It’s not that far of a stretch to add intelligence or connectivity.
@ mtylerjr - IMO, the downside of Wi-Fi in every device is less about cost and more about security. I’d prefer to limit the number of devices that could potentially compromise my network if the creator didn’t do a good job seeing the device.
For me, credible security is the gating factor here. Ideally, we would use end-to-end security with bidirectional authentication and credible crypto algorithms with good key lengths. Apple’s HomeKit seems a good benchmark here. From our own implementation of the HomeKit protocol, I guess that a 16 MHz Cortex-M0 is about the minimal practical requirement.
Unless the MCU has hardware support for crypto. But I’m not convinced that the common AES128 hardware accelerators with their 128 bit key length will still be considered acceptable say in five years.
Does anyone know what exactly the security mechanism of AllJoyn is?
Interesting and relevant: I asked an electrician buddy of mine what the future holds for lighting, given that it’s expensive and inefficient to put DC lighting (LED) and an AC-DC converter in every fixture.
He suggested that the future is coming fast, and it’s Power over Ethernet. Apparently, some are already doing this, with one central, efficient AC-DC converter, and run CAT6 to the fixtures, carrying both power and data for the types of things that LIFX and Hue do. Eliminates a lot of the security concern, with it not being wireless. Also eliminates a lot of interference, with many fewer devices fighting for spectrum.
I say, bring it on. Might be expensive to retrofit an older home, though.
I like PoE, but like you said, it can be expensive to retrofit.
Plus, the world is (and has been) going wireless. I happen to like wired, but I’m in a minority here. Most hotels don’t bother to have wired connections in rooms anymore. In a keynote I gave on Saturday, I had to go through hoops to get them to outfit the room with a wired connection for my router (my devices are all set up to communicate directly with my travel router).
Even security systems are wireless now, when PoE makes perfect sense for most of the cameras.
PoE is also not high-power. It’s enough for an LED lamp, though. Also, you can’t daisy chain like you can mains power. My kitchen, for example, has seven lights in the ceiling. That would be seven ethernet cables just for that room. If you go wireless, it’s a single romex to the set of lamps, and then wifi takes care of the rest.
So I think the future being PoE is an opinion that is not universally shared.