Eye Oh What?

The Internet of Things. IoT. Sensors, cameras, data, oh my! Many buzzwords have existed in IT, whether it’s “Cloud”, “Big Data” or “Automation”. Most new developments in the enterprise have started out as a great general idea, and is slowly implemented and evolved into a usable technology for the enterprise world. The Internet of Things (IoT) is no different. What classifies itself as IoT? Is it a robot that can move items down an assembly line and then report on it? Possibly, but we’ve been doing that since at least the mid 80’s. Could it be an iPhone that can pay for the grocery bill with the touch of an NFC tag? Sure could, but how does that extend into the grand picture of the internet. Better yet, how does that create a network of information that can be acted upon? Many questions, including these are warranted, but how do we implement this in a functional and repeatable sense?

I know that I’ve placed this in my projects area, and mostly because I plan on playing in the IoT space for some self learning, especially with the AWS ecosystem, but I thought it would be a great write up to get some opinions and thoughts churning. Part 2 will be my thoughts and plans for a project, but first let’s define for ourselves what IoT is/can be.

4 Steps to Clarity

IoT is a large subject, but I think that it can be broken down into a repeatable cycle of consumption. A model of deployment regardless of the use case. Like most standardized items in IT, we can programmatically design a way of utilizing these technologies to resolve some sort of issue.

IoT is more than just sensors, but rather what type of information we’re gathering from these sensors and how we’re manipulating the data we’ve gathered in order to further resolve a problem or gain more insight.

Step 1: Sensors and data gathering

IoT first starts with the data that we need to collect and the way that we collect this is through sensors. A sensor can range from a thermometer to a door-entry sensor to a video camera. Anything that can sense the physical world can be considered as a sensor in IoT.

We load our code, if anything custom is needed to be ran on the device, then we place our sensor to start collecting our data. In this state our sensor is isolated and not too much use to us.

Step 2: Data Pipeline and Landing Pads

Once we have code loaded into our sensors, we then need to take our data and place it somewhere that can be of benefit to us. More often than not we think of using a database. We’re not going to use just one, but two here! Why two? Because IoT has 2 very distinct ways of using and manipulating our data. Realtime and Deep (“Big Data”) Analytics.

Step 3: Real Time Analytics and Physical State Change

So, we have some sensors collecting some data, we’ve been able to ship that data to 2 separate repositories, but now what? This is where the fun starts. Imagine you’re a farmer and you’ve planted corn in 100 acres of field. You’ve also placed hundreds of soil testing sensors in the ground, which are currently indicating that 5 acres of your soil is dry and in poor condition. Like any good farmer, you go and investigate the soil, add water and fertilizer and continue on your day.

What we’ve done here is in real time adjust our environment based on conditions that we saw in our analytics. Imagine how else this could be implemented; dynamically adjusting air conditioning based on thermal imaging, speeding up or slowing down manufacturing robotics based on loading dock availability, etc. The point is that actions are being reported and provisions are being taken to take advantage of this data.

Step 4: Viewing the big picture

Data being gathered in real time is great, and we can achieve some wonderful short term goals with that information, but what else can we achieve? Imagine that we’re back on the farm (not down on the farm anymore, thanks Bob Evans!) and once again, we have the same 5 acre plot of land dry and damaged. Depending on the length of time between occurrences, the normal person may just view this as coincidence.

This is where big data, and historical data, can come into play. Granted, we’re using an extremely simplistic view of what analytics can do, but this example at least helps us with the big picture. If we saw that every 2 ¹⁄₂ years the same area of land dried up, we can start looking at other data and see if there’s any correlation, such as poor drainage, chemicals out of balance over a longer timespan, etc.

Back to the beginning and forward again

This workflow is not a one time solution, but rather can continue to drive into other technologies, such as machine learning and AI. The more data that we feed and iteratively work through, the more complex and mature our data models become, giving us more and more fine tuned results.

Data gathering in conjunction with being able to change the environment/interactions that our sensors are monitoring is what makes IoT so interesting to me. The applications are near endless and are tying closer and closer to robotics and the future of society overall. I feel like we’re at the beginning of what an interconnected society looks like. Imagine what the next decade will provide in change!

Matt Hoyt is a current IT Architect in Columbus, OH. All thoughts displayed here are his and his alone.