I built a Connected Home from scratch. Here's what I learned

I built a Connected Home from scratch. Here's what I learned
French-Startup

Connected Conference - Day 1 - Image by Dan Taylor - dan@heisenbergmedia.com-166 copy

Everyone talks about the Connected Home, the opportunity to build the next great platform (after the smartphone), and yet, not many people have built one. I spent three months putting one together for Connected Conference, and I can tell you – we’re not there yet.
The rules were pretty simple: we wanted to build a house that you could live in today, and we wanted to make it as connected as possible, as long as 1) it had an actual purpose 2) a regular consumer could put it together themselves and 3) no custom development allowed (and we even broke that rule in the end).
The goal was to put together a Connected Home that you could walk through, so that, instead of hearing about this futuristic home in which some startup said they would play a central role, you could actually see what that product looks like in the wild. It was an effort to benchmark what is connected in the home today – and, more interestingly sometimes, what isn’t. Here’s what we learned:

1) The Network Bandwidth is Key.


We spent 24 hours building the walls, the doors, and all that. I thought it would be hard to put windows into a building built from scratch in the middle of the exhibition hall, but it was much harder to get a dedicated network with bandwidth equal to that of a household full of geeks to stream 1080p home security video to a mobile app. It turns out, streaming 1080p is data-heavy: go figure.
A standard home wifi network can handle this, as long as you don’t also connect 4 MacBooks, a handful of tablets and your smartphone of preference. Not to mention the 30 or so other products that were communicating either to their own cloud or with other objects in the room over Wi-Fi.
Today’s connected home objects have surpassed 4G & high-speed Internet network standards even in the most connected of locations (our venue was outfitted with multiple networks with ranging levels of bandwidth – the house had its own dedicated 10 Megabyte/s network.)

2) No new tech where there wasn’t tech before


The kitchen was pretty full – the Triby fridge magnet (on top of a Samsung fridge with an LED screen), a connected cocktail maker (MixStik), and even a connected fryer (Thanks SEB) – however, the bedroom was… minimalist. Beyond our awesome Keecker robot and a sleep-monitoring platform and associated lightbulb, the bedroom doesn’t seem to have many dedicated products. Does this reflect the consumer’s desire to disconnect from the world in the bedroom, or is this an opportunity to create the must-have bedroom device? (I know what you’re thinking – get your minds out of the gutter).
What we really noticed is that Connected Home technology that is really making strides is currently piggy-backing off of existing outlets, existing appliances, knowing that consumers already expect a speaker to connect to a power outlet, so adding wi-fi-enabled speakers contains less friction than a wi-fi-enabled plant, for example (even though we had two of those).

3) Everything you’ve heard about Connected Home interoperability is BS


It doesn’t matter what your standard of choice is – we played around with all of them. I won’t name names, because i’m not into shaming, but the core issue is that multi-object communication doesn’t work. Sure, you can hook up your music streaming service of choice to your speakers to play music from anywhere in the house, and you can even hook up that same service to a lamp which will choreograph your multi-color LED light in time with music; however, doing both at the same time just isn’t possible. Right now, both of those connections (on the devices we played with) happen inside the respective apps of those products, instead of via your central music streaming app. Short of simultaneously playing the same playlist on both apps at the same time on different devices, it’s just not possible right now.

4) It’s only going to get better


Because we restricted ourselves to products that are being on the market (or will be in 2015), and because we didn’t want to do any custom dev (Rasberry-Pi enabled curtains, anyone?), our home was as evolved as the Connected Home industry. Which is to say: it’s on its way. Sure, we could’ve included more than the 30 devices we had in the home, but it would’ve only underlined the above points even more (and potentially taken down the neighborhood’s network); however, we see what’s on the horizon – API’s for your fridge to tell your stove what items are going bad so that your phone can preheat the oven in anticipation of your proposed meal, for example.
2016 will see the Connected Home get even more interesting, which is why we will be building yet another Connected Home, and we’ll be looking at measuring how far we’ve come. For now, there’s work to be done before Connected Homes will become the norm.

4 Responses

  1. ArnaudBeaufeist (@arnaudbeaufeist)

    Thanks for sharing Liam ! Interoperability is somehow already there, though : stantard is IP 🙂
    For instance : I can run my favorite playlist on my SONOS and switch my HUE lights on when I’m back home (ie : when my iphone has connected onto my home’s wifi). I can also have a Netatmo CO2 alert that runs my ventilation. Most of these equipments propose web APIs. And with a cheap smarthome box, you can get plugins and connectors to these APIs without any need for coding (but a little config!).

    • Liam Boogar

      Arnaud,
      Appreciate your comment, but you’re definitely misunderstanding me. What you’re talking about is using multiple devices at once. My example was: I want to listen to a song, and have two different devices that can ‘interact my music’ – speakers & say a lamp. Both of these products, in order to interact with my music, require that i play my music via their dedicated app. I can’t just come home and play music on my phone and suddenly have all my lights start flickering.
      However, this means that multiple things can’t respond to the same command. This is an issue, when you want deep integration with these appliances.
      Sure, I can set up IFTTT formulas to have lights turn on when I come home – these are singular, non synchronous actions. Multiple synchronous actions from one output: we’re not there yet, and this is what the consumer expects (i.e: “it should just work”).

    • Karl Schulmeisters

      IP is a very very basic transport level protocol. And not all devices support it (Bluetooth LE devices for example have an IP stack but to move LE functionality into that stack requires software ) and IP is just a transport. There is no IP packet dedicated to the status of a lightswitch or to what entry in your SONOS playlist you are on, or even to what IS a SONOS playlist.
      Interoperability is far more than just being able to exchange data packets. Think of it this way. If I write:
      Es šaubos ka jums šis būs saprotams.
      you are highly unlikely to be able to parse that without the help of google translate – which in fact will make a grammatically incorrect (though loosely understandable) translation.
      And yet the “transport” layer is there (latinate writing)
      and even a protocol negotiation layer is there (its clearly a western language of some sort and probably a European one)
      So to say that there is interoperability at the IP layer is to completely miss the challenge here.

  2. Karl Schulmeisters

    Interoperability is the least of the Connected Home’s challenges. That’s just software protocol stacks. The real issue is that of perceptual name spaces. This was a big challenge when I built one of the first real “connected home” projects back in the mid 1990s with an almost unlimited budget, and it still is not solved today.
    What do I mean by the “perceptual name space” problem.
    If upon entering the connected home Liam and his team built, I had said to Liam
    –> ” the light that’s in the corner by the guitar”
    –> “the light on the nightstand in the bedroom”
    –> “the light opposite the fridge”
    he would have instantly understood each of those sentences and which light I was referring to. Because we as humans make a “named” mapping of the “space” we are in based on our “perceptions” (ie “perceptual name space”).
    now – lets pretend that the intercommunications problem is automatically solved. And Liam has gone and deployed all the components into the space he built.
    Not one of those above mentioned lights
    …1) knows what the spatial layout is – you need to import or create a CAD drawing to have that
    …2) understands where within the spatial layout it resides – Geolocation currently isn’t on anyone’s radar – but it is why I told the BeSpoon http://bespoon.com/ folks that without their solution the Connected Home is a pipe dream
    …3) has a clue as to what it should be called. After all at install they will have really helpful id names like “Samsung_Frigo_3F2504E0-4F89-41D3-9A0C-0305E82C3301”
    So you will have an electronic network of “names” that also have a “spatial” layout that is based on the electronic “perceptions” of the devices.
    So you are faced with the problem of two very different Perceptual Name Spaces:
    –> Liam’s
    –> The connected devices.
    And mapping the two to each other is very very complex and today – and in the fairly forseeable future – even if the devices all have the BeSpoon chip integrated (which BTW increases their costs non-trivially) – takes quite a bit of human effort.
    And while Liam and I may be willing to make that effort. And while someone well-to-do who wants to show off how “chouette” they are technically will put up with paying someone to do this – for the average household, the value-add of this does not justify the effort.
    So I disagree with Liam…. I don’t think “its only going to get better”.. Its first going to get much much worse before it gets better. After all, how many of you still have one remote control for your LiveBox, and another one to control the surround sound or TV?

Leave a Reply

You must be logged in to post a comment.