[Article] Finally a Wi-Fi enabled light switch for the DIY crowd (UPDATED)

The protocol is going to be so easy, the switch runs Android, so there are so many integration options, it's a no-brainer.  One way or another, someone will come up with a driver/interface for this platform.  I already backed them, wish I had the finances to buy 100 switches ;)  Seriously, ask anyone in the chat room, I have been saying for YEARS that this is what we need.
 
I noticed they lowered their goal, but still not sure if they will make it.  That said, they did win a $1,000,000 grant, and the CEO used to be the CEO of CEDIA, so there is some hope.
 
I agree on the potential.  They got back to me on my communication already...  here are a few relavant snippets some of you may be interested in...
 
 
We are excited that you are considering ube for your project. I have known the Founder of HAI for years and will reach out to him regarding their Connectivity Partner Program to ensure the ube product works well with their platform. What i do know about the HAI control panels is that they all include ethernet connections and my understanding is that you are able to send IP commands to & from network devices.
"We opted to publish our API for exactly this purpose. Open standards and APIs broaden the consumer's choices and sparks innovation like our WiFi Smart Dimmer."
...
Kind Regards,
Utz Baldwin 
Co-founder | CEO
 
I actually just checked my list, I need 25, not 15.  I think I am going to jump on these...
 
Worst case, you resell them, at this price you'll even make a profit ;)  Can't wait, that's for sure!
 
mdonovan said:
If this all pans out they would be great. And the fact that they will offer an API, even better. They do look a bit deep for a normal box, but it's hard to tell from a picture.
 
Matt
Matt, keep in mind the photos are of our prototypes.  We are engineering them to fit into a 118 carlon box.  We will update the kickstarter with engineering docs as we get further along.
Utz
 
Lou Apo said:
Looks pretty nice.
 
2 questions:
 
1) How is this going to work when you have 50 or 100 of these all crammed onto a wifi router?  Are you going to need an enterprise grade setup?
2) What is standby power consumption? 
1)Good question!  Only some WiFi routers (err Time Capsule) limit the number of WiFi devices.. in apples case its 50 .  most however will allow up to the 254 available ip addresses in the subnet.  I will bet money though that performance will drop with 100 of our dimmers on your WiFi network.  They just aren't designed for that (meaning very large homes)(yet).  This is not just a ube problem, but an industry problem... That is why we formed the Internet of Things Consortium along with 10 other emerging companies in the connected home space..  (www.iofthings.org) The Internet of Things WILL drive innovation in routers and access points simply due to the number of IP enabled devices we will soon have in our homes.
Then you can buy hundreds of them from me for you projects. :)
 
2)  The 5mW is the draw from the processor and WiFi chip.  this does not account for the triac / dimmer.
 
Utz
 
I'll start off by saying this is an interesting project; In the past I've supported why people have avoided going TCP/IP for every single device in the home but if they're able to make it work cost effectively and find a market of people savvy enough to keep their light switches functioning after changing from UVerse to Comcast to FIOS - then hey, more power to everyone!!  Honestly, I'll consider the technology if we start seeing the integration possibilities.
 
That said, I want to add some technical info on Wifi - as that's an area that most people just don't seem to understand.  The problem about the number of client devices really has nothing to do with the Access Point or the devices - it comes down to a problem with the 802.11 protocol in that it's a Carrier Sense technology.  Here's a snippet regarding how it works:
Wifi is CSMA so every radio has to wait until it sees an opportunity to transmit (*Carrier*Sense*). The more radios, the more waiting. when you get past 30 radios, the wait time begins to take up most of the radio time. You might get 100 clients to connect, but as soon as one starts to transfer data, many of those radios are going to lose their connection. The radio doesn't matter, the combo of CSMA and the wifi standard adds up to this limitation.
With something like TDMA (many cell phones), each client device knows its timeslot and there's no collisions but with CSMA each device has to wait for an opening then try to jump in - think of it this way - in something like TDMA, everyone can line up in a circle and each one gets 1 second to transmit (example) - so everyone knows when their turn is and how much they'll get through; with CSMA everyone is trying to bum-rush at the same time leaving the radios and the clients to fight over who gets to make a connection and that overhead alone can end up taking down the radio keeping *anyone* from getting throguh.
 
It's not insurmountable - but the solution to this problem directly conflicts another common application - seamless roaming throughout your home.
 
The way you fix it so more clients can connect via wifi is by turning the power down making the coverage cell smaller and using more AP's - that in turn increases the density available.  However, there's no real roaming built into Wifi - so say you make the cells smaller - great - your stationary devices like light switches wii's and TV's will be happy... but say you then connect to your phone to watch a youtube video - as you walk from one end of your house to the other, it won't seamlessly jump to the strongest/closest AP just to keep a good signal; in fact it'll try to hold onto its connection to the weaker signal until it just can't hold it anymore (dropping speeds, causing excessive buffering, etc until it finally loses connection) then it'll finally jump to the stronger AP...  So fixing one problem will screw up the other.
 
This is the very core of enterprise wifi engineering and what we have to deal with on a day to day basis when covering a campus or venue that has to support dozens or hundreds of client devices in a given space.  There are of course some commercial options that have trickery to make this more seamless - which is very necessary for things like VOIP roaming between AP's while in the middle of a phone call - but those solutions aren't practical for 99% of the homes out there.
 
All that said, if I were designing a wireless infrastructure to combat these issues and provide support for my 75-ish light switches plus outlets plus Security Cameras, Media Players, and other devices around the home - I would do a hybrid solution.  I would probably put in a total of 5 AP's... I'd use 4 of them alternating between channels 1 and 11 in opposite corners of the home to provide complete coverage with minimal overlap (by turning the signal strength down) and use them for stationary devices - then I'd use one high-powered AP central to the home to provide access on Channel 6 for devices like laptops, tablets and smartphones that like to move around the house a lot and need to be able to roam as seamlessly as possible.  I can cover my entire 4000sq ft house on one strong AP if it's placed in the right location (as I do today).  With even a basic management solution like the Ubiquiti Unifi solution, you can also set up multiple SSID's so that the main Channel 6 AP also can cover the stationary devices in the middle of the home but through the management software you can set it so the devices prefer to stay on the outskirt AP's keeping radio time maximized for mobile devices in the center. To be clear, there'd be one SSID that's on all 5 radios for stationary devices - then a separate SSID on the main radio for roaming devices that should reach the whole property (note: adding SSID's in no way allows more radios to connect; it's just a way to control what they connect to in this scenario - in fact, it adds to the beacon transmits that the radio must do which is also why even the best radios are limited in how many SSID's they support).
 
Again - this is me - but this comes from the experience of engineering large scale Wifi solutions - and it's one of the reasons I've warned that going Wifi for everything would introduce yet more technical challenges that are just too much for the average homeowner.
 
I agree, but remember, these switches shouldn't have to communicate over the WiFi but for very short intervals right?  And then, only a couple at one time?
 
Maybe we would have to avoid scenes that control multiple switches, or space them out accordingly.
 
Work2Play,
 
Thanks for the great explanation of a problem I knew existed but didn't know the nuts and bolts behind.
 
I suppose the traffic frequency of each device would make a big difference as well.  You might have 100 switches but if each one is silent except when someone pushes the button, perhaps it won't overwhelm the system.  I guess it depends on if the switches have any kind of heart beat and if so how often.
 
It's my understanding that they'll still have the heartbeat going on as a keep-alive and that will contribute to the overall issue but by not being chatty that'll definitely help; honestly the data being sent should be less than the overhead of the TCP/IP packet itself.  
 
To be honest, this is something I'd like to see the manufacturer address - build a test-house case with 100 light switches and toss a few computers on there and some sniffer tools and see what happens.
 
If they want to lend me enough devices to test with, I'll even do it just for fun - I must have 30 different WAP's available to me to test in my spare stuff pile...
 
Looks like I will be a guinea pig. I am will be putting ~25 in a new home build with an HAI OP II.

I am considering ubuiqti APs. Used them once on another project and was impressed.
 
25 switches with 2-3 computers/tablets should be fine with a good AP; and the Ubiquiti Pro line has multiple radios which helps accept more - along with 2.4 and 5ghz frequencies which will let you put some devices on the 5Ghz one.
 
I guess I should be clear - I don't think the switches alone are going to do it unless you're talking 75+ switches like my house or keep going wanting to add more - if this trend continues, our alarm clocks, sound bars, TV's, refrigerators, washer and dryer, power strips, thermostats, and everything else will be on IP/Wifi 
 
I just saw some comments that mentioned the "cloud" when away from the local network, but talk locally when on the network. That sounds smart like a smart idea - the local network skipping the cloud that is.
 
Back
Top