The Internet of Things gets in the news these days, which is nice. But we’re still a ways away from where it gets really interesting, mostly due to cost.

Sure, you can buy a refrigerator that connects to the Internet (don’t ask why you would want such a thing, no one knows yet, but that’s the beauty). You can get a telephone that does so, too. Heck, I’m sure someone’s even made a lock for your front door that lets people in via the web… Oh, wait….

But most of these “things” that connect to the Internet today use Wi-Fi and Wi-Fi is a horribly expensive way to connect to a network for “things.”

Wi-Fi implies quite powerful hardware, both from a true power consumption point of view and from a software stack point of view. Even things like Zigbee or 802.15.4/6LoWPAN require quite decent hardware specs (IPv6 has been said to require at least 16 kB of RAM to function well, afaik). This means not only are rather large batteries required, many times also needing recharging, but the minimal hardware to attach to such a network needs at least $3 of hardware.

Antenna, PCB, match network, transceiver, micro, and battery. If you can do that, and get decent battery life, for less than $3 in volume, you’re doing quite well. I’m not saying it’s not possible, just very very difficult with today’s parts.

Hence, I think the really interesting Internet of Things applications happens when the cost gets below $1 to manufacture. We’re probably not that far away from it but I don’t think Wi-Fi or 6LoWPAN as specified today will be how we get there. The really interesting uses of an Internet of Things is when everything has access to the net, and that won’t happen with commodity items till the extra cost to buy an Internet enabled device is almost non-existent compared to the non-Internet enabled device.

Take for instance, my toaster. It’s not Internet connected and really all I want it to do is to toast things. But if I’m going to spend $30 on a new toaster, a 10% increase to $33 (see above for costs) seems silly to me, what exactly does my toaster need the net for? But if the increase in cost is closer to $0.50 then why not? Maybe I can do some neat little project with it or something.

Toasters are a good judge of cost, I think for this.

But to get the net into all these inexpensive things we need a different physical layer than is available today. I think it’ll look a lot more like passive RFID than an active radio. The need for a power source is very limiting for many things but passive RFID doesn’t need one. Passive RFID often is available with “battery assist” in order to increase range, which takes care of RFID’s often cited limited range (think 30+ meters with battery assist).

The cost of passive RFID is cheap, lots of tags are available for < $1. Often these tags have no microprocessor but low power micro designs are getting to the point where they can consume tiny amounts of power and do useful things, which will be a requirement in order to harvest RF and run. Some of the new memory technologies, like FRAM, might help here too in order to boot fast and persist RAM.

Obviously IPv6 will be the middle layer for this new tech, hopefully not consuming 16 kB of RAM, and we’ll need cheap (think < $50) access points like our Wi-Fi has today, for any place where these devices want to get online. But the real hard part is standardizing this stuff and finding a way to get decent data rates and functionality for < $1 per thing.

I give it 2 years till we see something like this start to emerge. 6LoWPAN and Zigbee and 802.15.4 are leading the way today but they will soon be surpassed by a new generation of very low cost batteryless (and battery assisted) passive “things” that are net connected.

It’ll be fun to watch RFID make this transformation.


01 July 2013