Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Super friends: IoT, AI and cloud create a powerful team

What do blockbuster movies have to do with IoT? More than you might think.

Superhero movies are the big box office draw right now, with superhero team movies leading the way. Even if you’ve never seen one, you probably know the formula: A group of superheroes with different powers and vulnerabilities band together to defeat the biggest, baddest villains imaginable. It’s not a new formula, but it works. And I’ll suggest it also works to tell the future story of IoT. Here’s how.

Think of IoT like a superhero team: The trick to unlocking its greatest potential is to bring together a diversity of technologies — the strengths of each perfectly compensating for the vulnerabilities of others — then deploy them against the biggest, most intractable business problems. The super team of IoT, AI and cloud is a great candidate for such a story.

Tech superpowers

Everyone knows superheroes have certain vulnerabilities, like kryptonite, and dependencies, like a magic ring. Tech superheroes are no different. No single technology can do everything, and every technology has its dependency: lightbulbs need electricity, cars need gasoline and so on.

So, while we know that IoT can find powerful new insights in almost every aspect of daily life, there’s an implied dependency: the ability of human analysts to find those insights among mountains of data. According to some estimates, IoT-connected devices produce almost 1,000 times more data each day than the entire internet in 2005. And while there will be ever greater, more useful insights in all that data, no human could hope to uncover them in a single lifetime. This is our cue for the next technology super friend: AI.

Where IoT’s superpower is to sense and generate tons of data, AI supports by sorting through that data, making it actionable, and getting smarter as it does so. With AI, we can now process human speech in real time, determine where time-sensitive shipments need to be routed in a supply chain and many other applications that would have seemed like the stuff of comic books just a decade ago. But AI’s superpower has its dependency too — it often needs more processing power than most edge devices can provide. That’s where the call goes out for the cloud.

The cloud’s superpower is ubiquity, expansiveness and being on-demand — which means it doesn’t just store all the data coming from IoT’s sensors, it can also host AI and machine learning tools. With this new team member in the mix, IoT users can rely on the processing power and sophistication of cloud providers and even select from a variety of AI tools that suit their particular needs in the moment. In fact, as edge devices become more sophisticated, users can increasingly run simple AI tools locally at the edge, without the need for costly backhaul communications from edge devices back to central servers.

IoT, AI and cloud on a mission

IoT, AI and the cloud’s strengths and dependencies fit together nicely on paper, but their real power comes to play when they work together to tackle real-world challenges.

Like managing a power grid. Traditionally, power grids were centrally managed by utilities that controlled both the production and transmission of electrical power. The only variable was customer demand. But today’s modern grid has many more variables: power production from renewable energy sources vary depending on conditions, while small-scale solar and other generation technologies add to the complexity as utility customers both consume their own power and sell their surplus back to the grid. And with electric vehicles of all kinds proliferating, this complexity — and the challenges it brings — will likely grow.

But together, IoT, AI and cloud can help utility managers not only address all of these new variables, but make use of them to create a more efficient, greener grid.

In the United Kingdom, the independent energy supplier OVO provides one example of how IoT, AI and cloud can come together to create an entirely new business model based on data.

OVO is trialing a service for residential consumers with solar panels that combines a variety of sensors and data sources with in-home (or in-car!) batteries. The IoT sensors track the home’s energy usage and battery power levels, communicating that data to the cloud where it’s combined — by AI — with data like public grid load and current electricity prices. With this enhanced data, the AI can then direct the customer’s batteries to store excess energy when grid demand grid is low and release it to the grid when demand is high.

This business model gives homeowners stable, fixed prices on their energy, while the grid (and the public it serves) gets access to more energy in distributed storage. This kind of smart management of demand is a way to a transition to greener sources of energy. (See our article on disruptive innovation in the power sector to learn more.)

Not the only solution

But just as a blockbuster may not be a good fit for an art house festival, the combination of IoT, AI and cloud may not be right for every problem. Communicating data collected back to a central cloud may not be cost-effective for simple IoT applications in austere environments, like pipelines. Similarly, the time required for communications and analysis may make this combination difficult to use in applications — like monitoring pipeline valves — that are always-on and require minimal delay. Such analytically simple tasks don’t benefit much from AI, and the valves are often in geographically isolated locations making communications costly.

So while IoT, AI and cloud can accomplish incredible things together, the first step for any business should be to think through the following three criteria:

  1. Does the volume of data collected from a potential IoT system require more than simple human analysis or could it benefit from deeper analysis?
  2. Can backhaul communications be established from the edge back to central servers? If that communication cannot be constant, can it at be established at regular intervals or paired with simple at-the-edge AI?
  3. Can reliable training data be found for the AI?

If the answers to these questions are yes, you may want to reach for whatever serves as your bat-signal and call in the super team of IoT, AI and cloud.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.

Data Center
Data Management