Home » IoT Embedded Systems » Featured Stories » IoT Analytics is a Different Breed of Analytics

IoT Analytics is a Different Breed of Analytics

Analytics purpose-built for the IoT and tied to IoT platform technology is the only practical way to mine the IoT data.

By David Friedman, CEO, Ayla Networks

Now that objects of all kinds are getting sensors, and embedded networking software is becoming part of the Internet of Things, there’s growing realization that the data generated by connected products could represent a goldmine—for manufacturers, sellers and users of the IoT-enabled products.

At the same time, the market is filled with a number of very sophisticated analytics software approaches. Given that analytics is all about turning raw data into usable, useful forms, the obvious question is: Why not just apply existing analytics packages to IoT data? Or is there a need for a subset of data analytics that’s specific to the IoT?

In the IoT, there’s a huge challenge in collecting and ingesting data from the diverse range of connected products out there. In most cases, these are not products that were designed with networking and data generation in mind. They are the coffee makers, water heaters, washing machines and light switches built to perform very specific functions, typically predating the IoT and the concept of connected products. Even when they are IoT-enabled, these products will necessarily generate vastly different kinds of data in many formats. Before that data can be fed into a sophisticated analytics packages, it will need to be “translated” into a form recognizable by these analytics solutions.

That translation is one of the important functions performed by an IoT platform. IoT platforms are necessary because when it comes to the Internet of Things, it’s the “things” part that will trip up the “Internet” folks every time. And when it comes to analytics, IoT analytics is a breed all its own.

It’s All About the Things

Fragmentation is a fact of life for the IoT. Unlike the world of computing and communications devices, the Things of the IoT will never coalesce into a handful of nice, manageable groups. Every category of product connected to the IoT has its own domain, which likely has nothing to do with the technologies, ecosystems or expertise required for IoT connectivity.

The disparate product types connecting to the IoT are not like the computers and smartphones and tablets that connect to the Internet of Computers, which were designed from the outset with connectivity built in. The Things of the IoT connect in multiple ways, use multiple networking protocols, have different purposes, and generate different kids of data from one another. Not just temporarily, but always.

Turning traditional products into connected products requires specialized, deep knowledge and expertise in networking, software development, cloud computing and other technologies. And once manufacturers’ connected products are deployed, they need appropriate data analytics solutions to make sense of the data generated by their now-connected products.

Smart Things and Dumb Things

Many IoT-connected products have limited capabilities for computing and communications—sometimes due to physical size, other times due to the prohibitive cost it would take to add sufficient memory, processing power, protocol support and other technologies.

For such constrained products, the best way to make them smart is to leverage the cloud for intelligence. With the right IoT platform in place, even the dumbest products can “catch” intelligence from the cloud as needed. This is important for everything from scheduling—e.g., making sure that sprinkler systems that lose Internet connectivity while running don’t keep running indefinitely—to security—at the product, cloud and software control levels, as well as through all the hand-offs along the way.

IoT analytics solutions need to be able to deal not only with the connected products’ full range of protocols and data types, but also with various cloud architectures, mobile and web-based application types, and security measures. IoT analytics approaches also need a data architecture able to handle the constant, perhaps overwhelming, stream of data generated by always-on IoT products. And they need to do all this at scale, even when operating on data from millions or tens of millions of connected products.

Finally, IoT analytics solutions must account for the fact that their users—mostly manufacturers of traditional products that only recently have joined the IoT—are unlikely to have teams of data scientists on staff. Manufacturers have to be able to ask and get answers about their connected products’ performance, as well as their end users’ real-world experience using the products, easily and cost-effectively.

Navigating the Jungle of Connectivity to Find the Gold

Appreciating the potential goldmine of IoT data is easy. Getting there is another story. IoT analytics solutions able to handle all the connectivity challenges presented by the multitude of IoT Things will be needed to unlock the full potential of the IoT.

We believe that analytics purpose-built for the IoT, and ideally tied to comprehensive IoT platform technology, is the only practical way to mine the IoT data for all the gold that it represents.

David Friedman is CEO and co-founder of Ayla Networks in Sunnyvale, Calif. Contact him at david@aylanetworks.com .

Great information delivered straight to your inbox

One comment

Leave a Reply

Your email address will not be published. Required fields are marked *

*