Third time is the charm - Why the World just might be ready for the Internet of Things this time around

The technology to connect 'things' to the Internet has existed for more than 20 years, so if we take a look back at recent history we might well be tempted to ask the question why will IoT ‘happen’ this time around. In this paper we examine the origins of the Internet of Things, answer the question "Why Now?", and look forward to the next wave of disruptive technologies that will be coming to a device near you in the next few years. Index Terms—Internet-of-Things, IoT History. I. A Short History of the Internet The term “Internet of Things” (IoT) was first documented by British visionary, Kevin Ashton, in 1999. He used the phrase to describe a system where the Internet connects to the ‘real world’ via an ubiquitous network 
of data sensors. Of course the use of this term has grown somewhat beyond the original intention and today it means many things to many people. But to get back to the root of it all we should also consider the “Internet” itself in order to understand the full context of the IoT. The origins of the Internet go back to the Arpanet in the late 1960’s. By 1970 there were five permanent nodes on the Arpanet at several of the largest US. Learning from these early days the researchers realized they needed to build a robust data protocol that could recover from transmission errors. In 1974 Vint Cerf introduced TCP/IP but it was a decade later before it was broadly adopted across the network and the real growth could start. In 1984 the c.1,000 active network nodes on the early “Internet” switched over to adopt TCP/IP for their core data transmission and networking protocol. And since then the network has continued to grow unabated. There are two key things to remember about the Internet: (i) the “Internet” is not the Web; you can think of the Web as a GUI for the display & publishing of data carried by the Internet but the underlying data transports, in particular TCP/IP, are what have allowed the Internet to scale; (ii) the “Internet” was designed to military specifications as a ‘battlefield’ protocol; it is designed to be able to adapt to unreliable channels and to recover from data loss. This last point has allowed the “Internet” to grow consistently over the last 4-5 decades and the introduction of mobile devices has further driven this demand. In fact you could say that today a computer is pretty useless without Internet connectivity and that underlying connectivity is provided by TCP/IP. II. A Personal Perspective on the early IoT The technology to connect ‘Things’ to the Internet has actually existed for more than two decades. I can confirm this personally as I was blissfully connecting CEBus devices to the Internet 3-4 years before Ashton’s revelations [1], [2] and even developing Java user-interfaces for these ‘Things’ [3]. But let us begin this journey back in time with a workshop I gave in 2002 at the IEEE International Conference on Consumer Electronics (ICCE 2002) entitled:. A. ICCE 2002 – Home Networks for the 21 Century This tutorial session that I organized covered a range of topics from the physical layers, to connectivity technologies available at that time and implementing TCP/IP on 8-bit embedded microcontrollers. The conclusion at that time was that TCP/IP was the way to glue things together at the lower layers of the communications stack but there was a need for a middleware layer to sit on top that would understand the nature and capabilities of individual devices. The final presentation in the tutorial considered OSGi and UPnP for this middleware, concluding that UPnP was a good foundation, but needed additional capabilities, especially for the UI. In fig 1 you’ll note a slide from this tutorial giving an overview on how to build a ‘Thing’; note that at that time 16bit micros were state-of-art and low cost devices would have likely employed 8-bit micros. And the optimal connectivity technologies for low-cost devices were Bluetooth for wireless, which was difficult to use, or wired Ethernet. Fig. 1. How-to build a ‘Thing’ in 2002 This was based on the cost factors shown in Figure 2. Note the cost of a hardware Ethernet was low enough to be considered for mass-market products and Bluetooth was almost cheap enough to be a contender. Wifi technology was far too expensive at this time for genuine consumer products, being 10’s of USD even in high-volumes. Fig. 2. The costs of connectivity in 2002. At this point you might start to wonder why the IoT didn’t ‘happen’ in 2002. The enabling technologies were clearly available and there were many people interested in how to developed connected devices – we had strong attendance at this tutorial. But somehow the pieces didn’t fit together in a compelling way. Over the following decade it became clear to me that simply connecting “Things” to the Internet simply doesn’t create enough value to sustain practical business models – if it did then IoT would have happened back in 2002. B. The Age of CEBus – the mid-1990’s If you know what CEBus [4]–[6] is then you are likely a CE industry veteran like myself. It was a networking standard developed originally by the Electronic Industries Association (EIA) and then adopted by the Consumer Electronics Association (CEA). The standard was very much ahead of its time and supported multiple physical layers including twistedpair, coaxial cable, powerline, wireless and even RF. I had discovered CEBus in the early 1990’s and managed to find resources to purchase some development kits – it was my first exploration of advanced networking protocols on embedded controllers. At the time there was also a new OS based on Unix available called Linux and in my youthful enthusiasm I had started playing with this new toy. It was a breath of fresh air to be able to compile the OS from scratch. And of course it also gave access to the TCP/IP stack. Perhaps you can now begin to understand how I ended up delivering that tutorial nearly a decade later in 2002? It seemed natural to build a Linux-based ‘gateway’ between our CEBus nodes and the Internet 1], [2]. Then a new technology called Java arrived on the scene – it was networkcentric and provided a powerful UI capability, for that time. Again it made sense to build UI-components that would be activated by the presence of CEBus devices [3]. Not only could we connect devices to the Internet, but they could have dynamic user interface elements so people could control and interact with these ‘Things’. C. Waiting for GODOT Apologies to Samuel Beckett, but I must confess that I’ve been waiting to “get our devices out there” (GODOT) for more than two decades. In the beginning I was so keen I even set up a small company in the late 1990’s with a view to developing the next generation of connected consumer devices. That company eventually moved in a different direction (www.fotonation.com) but anecdotally, its original business was “connectivity for devices” which then transformed into “connectivity for cameras” and continued to grow and evolve towards the technology needs of the nascent digital consumer imaging industry. But I digress a bit. It is clear that the enabling technology for IoT existed not only in 2002, but in fact 6-7 years earlier in the mid-1990’s it was also available and arguably in a more capable, functional and scalable form than many of today’s IoT solutions. So if the technology existed and there has not really been any disruptive breakthrough then we arrive at the same question why did IoT not go mainstream before now? III. IoT in 2016 – What is Different? The world wasn’t ready for IoT in the mid-1990’s or in the early 2000’s, so what has changed in the meantime? If you know a little about me you may already know that I have been an IoT skeptic in recent years, but now I begin to see some things that are changing my views. To get a better understanding and context lets recap what we know hasn’t changed a lot: The Internet is still pretty much the same; it got bigger and more technologies have been layered on top of TCP/IP and its little brother, UDP. We have a lot of support now for audio and video traffic which shows that the capacity and capability of the infrastructure has increased, but there have not been any radical changes – just a constant growth of nodes and data traffic. Embedded devices haven’t changed radically either. Yes, we have moved to 32 bit systems and most embedded devices can easily support a sophisticated OS, but the underlying connection is still via a TCP/IP stack which was available nearly 20 years ago for 8/16 bit devices. Connectivity technologies have improved but not radically. Yes, we can now have more sophisticated Wifi connections for less than 5 USD, but you could have achieved a wireless link in 2002 using Bluetooth and an Ethernet bridge. There isn’t anything very disruptive here that would argue that IoT will become a commercial success today. So now lets consider what is new and has changed the technology landscape.