IoT: Different Visions, Different Truths
If a real, IT-oriented Internet of Things model deploys, it will change everything about how we work and live, and change every technology decision we make.
We all know about the Internet of Things (IoT), but its name has led us astray. The Internet of Things should really be called the "Network of Things," and our first problem with IoT comes from the difference in these terms.
Most vendors take the term "IoT" literally; they see all of our "things" directly on the Internet. Cisco is one example; it has always seen the IoT as being explicitly about the Internet. Network operators are opportunistic, too; Verizon sees IoT even more narrowly, as evidenced in its recent IoT announcement. In Verizon's view, IoT is "things on LTE." You can understand both companies' perspective would suit their own business goals, but the danger is that these views threaten IoT's underlying business case.
IoT advocates say that we'll have billions of "things" networked. We already do. Virtually every factory, public utility, transportation giant, and many offices and homes already deploy sensors to control things. You can buy a simple home motion detector for less than $40, and run it for six months to a year for the price of a couple of batteries. A sensor for window/door opening can be even cheaper, and like all of these sensors is wireless. Best of all, you don't hear of people hacking or subjecting such sensors to denial-of-service attacks.
Putting these sensors on LTE or the Internet would certainly raise their prices significantly. Powering the sensor with an LTE radio takes more than a couple of triple-A batteries. In addition, the cellular/Internet path to IoT is what generates all the concerns about security and privacy. A sensor directly on the Internet is addressable, and can be hacked. If you want to enforce public policy restrictions on use or security, you have to build software into the sensor to protect itself. This adds to the cost of the device and to the difficulty in powering it.
Then there's the problem of sensor context and authenticity. Suppose you want to know if you're going to encounter traffic ahead. A dozen sensors might be "ahead," depending on your definition of what that means, but how would your smart car know which they were among the teeming IoT "thing" masses? How would you know that somebody didn't plant a false sensor to lead you astray for fun, or to do you harm? How do you even know the data format in use, or how to interpret it? Remember, we have billions of these out there.
Controllers are even worse. If you put a control element directly online -- a traffic light or furnace, for example -- you run the risk of someone taking control of it directly. Even if the device requires an encrypted control message, you could still swamp it with junk. Do we then have gridlock, frozen pipes, or worse?
Any smart IT planners would look at "thing networking" a different way. They'd ask what sensor data their company already collects. They'd pull that data, time-stamped, into a big-data repository where it could be subject to analytics and correlation to generate events. They'd see IoT not as the Internet connection of sensors and controllers, but as an Internet-connected set of big-data applications.
In this real IoT opportunity, everything that is trusted could contribute sensor data to the repository and impose policies to control access to it. These are enforced in the repository not the sensor, so we don't need to change out what we have or impose expensive software controls at the sensor level. Controllers can be accessed through related cloud APIs that go through similar policy filters. All of a sudden we have manageable risks, and we can get to the real IoT opportunities.
If deployed successfully, IoT would give us a way of understanding the context of our lives and work. We would know where things are, what's happening based on geography, and if we expand our definition of "things" a bit, we could also understand the social and geographic context of everyone. We could confidently answer the classic Siri question "What's that?" by combining all of IoT's contextual information.
And there's more. A T-shirt or golf club is a "thing," and so is a dinner special or a musical. If we collected the attributes of these things, we could change shopping for stuff forever. Imagine virtually walking down the main street of your town looking for a deal or something to do. When your virtual self, interacting with the thing-ness out there, gets the right answer, you follow it to where you want to go.
Work changes too. We've all heard that IoT could help field personnel and salespeople find junction boxes or prospects, but it would also enable tighter integration of software-automated diagnosis and remediation with human interaction. People could be guided to a specific place and given virtual images of what to find, what to do.
Collaboration also changes. The success of online meeting software like WebEx, GoToMeeting, or Lynx demonstrates that we collaborate with each other around something shared -- like slides. IoT, with its almost-complete view of the world as a context map, could give us a lot more to collaborate around -- any experience, any "thing," in fact.
If a real, IT-oriented IoT model deploys, it will change everything about how we work and live, and change every technology decision we make in the data center, our homes, and networks globally. What kind of changes? That's what I plan to explore in the next blogs of this series.
Follow Tom Nolle on Google+!
Tom Nolle on Google+