Building an Internet of Things

We love buzz words at Levvel. Cloud, crowd, PaaS, Big Data. All of them get bandied about fairly regularly in our hallways. But arguably my favorite one is Internet of Things. Every day I read a new mainstream news source mentioning the term. I’ll add my two cents to the discussion since I spent two years as the COO of one of the finest examples of an Internet of Things company, Nexgrid.

Image

There are plenty of great definitions of Internet of Things. I’ll focus on systems connecting various analog measurement devices onto a network that can then store the readings from these devices. Once these readings are stored, other services can then access, aggregate, and report on these readings in ways that were not possible until recently.

In the case of Nexgrid, we build a dual wireless mesh network that allows power companies and their customers to interact in real time with electric, water, or gas meters, with load control switches, with smart thermostats, smart light switches, and ultimately any device that has an IEEE 802.15.4 (“ZigBee”) or 802.11 (“WiFi”) radio on it. Nexgrid builds communication gateways, or EcoNets that link with an optional 802.3 (“Ethernet”) connection or with one another to form a patent pending wireless dual mesh network. They also transmit ZigBee traffic. Nexgrid also manufactures a variety of end point devices that communicate via WiFi or ZigBee with the ecoNet gateways or with one another. Nexgrid writes firmware that runs on these devices to securely interact with the meters or switches to read or update each device. Cloud-based servers connect securely to these devices and receive reads or issue commands.

Image 2

The result is that power companies can do new things in 4 key areas: measure, monitor, manage, and optimize. They can take measurements in real time on all sorts of interesting statistics: power, voltage, variance, and many harmonics on the line that typically would require expensive equipment and elaborate installations. Because these readings are stored in sophisticated software systems, they can be easily monitored. For instance, a power outage can be detected in milliseconds and a notification sent to a dispatcher who immediately knows where a line problem occurred. The power companies can also manage their assets, whether this means shifting load dynamically during a hot day by turning off water heaters when no one is using them, or by turning off every other street light on a street during a low traffic period. And the power companies can optimize their grid via analytics that illustrate when a given transformer, circuit, or substation is under or over utilized.

I believe there are a few key things that a truly useful Internet of Things has to embrace.

  • Massive Redundancy – if there is a single point of failure or a small number of potential bottlenecks, the system will fail as new nodes are added and complex pathways established.
  • Ubiquitous Standards – proprietary links will limit the ability to integrate new devices and will hamper efforts to build security into the network.
  • Open Source Software – security, reliability, and scalability need to be built into the network at every layer. No single proprietary software vendor can anticipate the myriad of configurations that occur on a massively parallel network connecting millions of devices supplied by thousands of vendors. Open source software has been proven to provide the resiliency and adaptability required for this type of platform.
  • Big Data – another nebulous term, but the reality is that an Internet of Things will always generate massive quantities of data. Real time, always on connections are great, but the real power comes when these values are stored for years and decades and incorporated into decision making. The volume of data involved WILL at some point break conventional data storage and retrieval tools. It also requires new ways of presenting and visualizing data.
  • Application Programming Interface (API) – the network must surface a well thought out and clean API so that other applications can be built on top of the platform to enhance its functionality. Standards such as ZigBee’s Smart Energy Profile (SEP) are a good start, although lower level technologies like RESTful XML and JSON web services that follow industry best practices will provide functions that can be combined in unimagined ways by creative developers to drive true innovation (think of Apple opening up the Apple Pay API to allow app developers to build future exciting ways of paying for all sorts of services).

Companies in every step of the eco system need to think of their products as services that comprise much large eco systems as described in this post. This includes device manufacturers, component manufacturers, network providers, software vendors, and software developers.