Faceless sensors and tiny routers needed for the internet of things

Texas Instruments has unveiled a Wi-Fi module that lets people connect devices without screens to a Wi-Fi network via a smartphone. For anyone with a Sonos system or other connected devices in the home, this may not seem so novel, but as we add more connected devices inside the home with lower price points, a standard module that enables a connection via a smartphone makes sense.

The TI gear lets companies add Wi-Fi connectivity to garage door openers, door locks, lamps, cameras or anything else without requiring manufacturers to bear the added expense of including a display or input device. The user will connect devices containing the module via their smartphone, which is the great enabler for the internet of things. That means it costs less to build connected devices, and presumably will make them easier to connect using a standard radio technology. Of course, there are other radio technologies that someone might want to use.

But, while figuring out a way to add connectivity to dumb devices without adding expensive components like screens or input devices (those also consume a lot of power, which is bad for battery-powered devices) is important, it’s not the only element we need to develop intelligent and connected environments. Another element that might make sense is a lightweight router that connects a network of sensors. For example, ThingSquare, which has created an open source hardware and software platform for the internet of things, called ThingSquare Mist that includes a lightweight router. The founder of ThingSquare, Adam Dunkels, developed Contiki a programming language designed for connecting and communication between sensors.

Routers, gateways and more!

The ThingSquare Mist router, which was released right before Christmas, requires a mere 5 kilobytes of ROM and 1 kilobyte of RAM. The router runs on tiny, low-cost microcontrollers that can be on the sensor itself: no need for a computer gateway or a bulky Linux board. While this setup won’t work for every network, it’s great for sensor networks in remote locations where power is limited or even as a router to coordinate multiple sensors for some kind of mesh network to monitor physical health.

Once you have sensors, connectivity on those sensors, and some kind of router to collect and pass information between the sensors, you then need either a brain accessed through the cloud or on a device, like the Raspberry Pi or maybe even your smartphone. Or maybe your router will look more like a gateway that speaks several different radio standards, which is what the MobiPlug box or SmartThings hub offers. Figuring out what people will use depends on whether or not it makes sense to have your intelligence at the edge or in the cloud — a debate I covered yesterday.

Dunkels says that a smarter router/gateway device adds costs without providing visible functionality, which can be a hard sell for buyers. “It’s hard to sell a router,” he says. “It doesn’t do much. It just sits there. Suppose you want a system that monitors your plants. You’d rather buy the sensors for $ 10 and a service than buy the gateway that 20 times as expensive so you can have a moisture monitoring system.”

The Tado system

The Tado system

He added that putting a little bit more intelligence at the end point and then tapping into a cloud-based service is more viable, although if one looks at the comments on our post about the Tado thermostat, which was built using the ThingSquare system, people have a hard time buying an expensive service too. The thermostat is adaptive and basically a cheap box controlled by your smartphone, but also costs $ 126 a year for the overall service. In contrast, the Nest thermostat offers less of a “service” but it is a one-time purchase.

Oh, and what about security?

Finally, once you have the brains, the network and the storage of the data somewhere, there’s a final building block — security. ThingSquare, for example, lets sensors talk to the internet, but the internet can’t talk to the end points on the network. Of course there are different entry points for malicious actors. They could infiltrate the the network or device itself or access the data traversing the network or once it is stored in the cloud.

Questions remain about whether or not the hardware will require security built into the chips as some people seem to believe, or if there’s enough bandwidth (or power) available for encrypting data in some of the networking standards that the internet of things will use. Of course, we’ll still need the apps themselves and the availability of data or APIs that will make building services easy, but that’s a topic for another post.


GigaOM