In the first of a series of blogs on the Internet of Things, Larry Mittag outlines the landscape of concerns around constrained resources, network alternatives and security.
Some of you may remember me from the articles that I have written for Embedded Systems Programming or the column that I wrote for a few years for Communications System Design magazine. EE Times has graciously (or perhaps foolishly) provided me a blog platform for a bit here. I will be working under the general topic of the Internet of Things, and my specialty will be the network architecture and software stacks involved in that wonderfully trendy topic.
I must admit to a bit of amusement at the attention that IoT is drawing these days. Years ago I headed up the embedded systems division of a services company named Stellcom, which had an Internet division. The combination was relatively arbitrary, but turned out to be a great ‘peanut butter and chocolate’ mix that gave us a point of view that was relatively unique for the time. We were on the front edge of the Internet-enabled device wave, which produced a number of very useful devices and a larger number of silly ones like Internet-enabled toasters.
That time also was the development era for smartphones. There was much discussion about how Internet protocols would have to adapt to the limitations of small devices. Processing power was limited, as was network bandwidth, so obviously special protocols were needed. The most popular at the time was Wireless Application Protocol (WAP), which was a stripped-down set of HTML protocols that wouldn’t stress these devices.
Fast-forward to today and my cell phone is faster and has more storage than the departmental PDP-11/70 at the first company that I worked for out of college. Cellular bandwidth is such that I could use up my entire monthly allotment of data in a couple of days. The WAP protocol has completely disappeared from the scene, and the acronym has been recycled into Wireless Access Point.
Now we come to the new frontier of the Internet of Things. From one point of view this is just a simple extension of the Internet and, of course, it will be using the same protocols as the rest of the network. Granted, the sheer number of devices will require the use of IPv6 protocols instead of IPv4, but this conversion is already well underway. (By the way, has anybody noticed this? Both my cell phone and home ISP now have IPv6 available. No muss, no fuss, it just works. I was watching for it and I am not sure when it happened, but it did.). From this point of view the IoT expansion is simply a matter of scaling up the infrastructure to handle the explosion of devices.
The other side of the argument is that there are fundamental limitations to the devices that are being enabled on the network that make this impractical. The sheer size and complexity of a TCP/IP stack are too much of a burden for a simple sensor, and the wireless communications will require too much power. What we need are special stripped-down networking protocols, adapted to the embedded environment. Many in this camp are also very skeptical about the whole idea of an IoT expansion that encompasses such a large universe of devices. Why do we need that, anyway?
Whenever there are two such polar points of view the reality is usually somewhere in the middle. People are just starting to come to grips with what they can do with the fine-grained and detailed data that can be collected from a sea of distributed sensors. There are real gains to be hand from the patterns that emerge from this type of big data analysis which increases our understanding of the world around us.
Next page: Toll takes and security threats