Get started Bring yourself up to speed with our introductory content.

IoT and the case for intelligence at the edge

There are six billion things connected to networks and sharing data right now. An extraordinary number for sure, and yet we’re still at the beginning of what promises to be a long-term, geometric expansion in volumes of connected devices. Gartner estimated that within four years more than half of major new business processes and systems will include “some element” of the internet of things, which means more connected things.

There’s good reason to believe the hype. Sensor data pent up in IoT can be unleashed to make homes, cars, businesses, utilities and even cities smarter. Our challenge is to develop systems that capture and process data fast, efficiently and at the point where it can have maximum impact.

The case for adding intelligence at the edge

Much of today’s sensor networks are positioned near or have fast access to clouds. No one should be surprised by that; much of IoT, especially consumer IoT, has formed up in areas where it’s convenient to deploy. But as it grows, we should expect companies to bring sensors and data acquisition systems to things in remote areas, resulting in use cases that the cloud would be ill-suited to support.

There seems to be a frenzy today around “sensor to cloud, sensor to cloud.” And there’s good reason for this, as the cloud has proven to be a fine progression from traditional data center IT. So, why not impute the benefits of the cloud on the processing of sensor data from the IoT. Well, we should, have and will. But there are just seven problems:

  1. Latency is too important. A smart car that uses sensors to detect items on the road ahead needs to process information instantly. The cloud can’t provide that, and shouldn’t be expected to.
  2. Bandwidth is too sparse. Rural areas are generally connected via expensive satellite systems that don’t have much bandwidth to begin with. Sending information to the cloud this way can take far too long or be impossible to yield useful results.
  3. Compliance requirements would complicate data sharing. Governments and corporations have put restrictions on how far data can travel. Thus, collecting data in London and shipping it to Paris for analysis isn’t an option anymore.
  4. Transmitting data would create security problems. Attackers target data that is in motion because it’s usually tougher to secure information that’s in transit. What’s more, data that’s being sent for analysis is also usually worth encrypting, which can make files fatter and transmission materially slower.
  5. Pushing data to the cloud costs. Bandwidth isn’t free and sometimes it isn’t cheap.
  6. Cloud access duplicates data collection efforts. There’ll be some, not all of course, duplication of software and hardware if both the edge and the cloud are equipped for massive IoT data.
  7. Distance creates data corruption that pollutes analysis. Have you ever participated in a cross-border phone call? Between the static and the dropped words you get a sense of what you’re hearing on the other end, but the connection is still anything but clear. Too much data is being lost during transit. Cloud connections can suffer from this same problem.

In summary, we can’t assume that pervasive connectivity will always allow the proper sharing, combining and processing data. Pervasive connectivity may not be available or suitable on an oil platform floating in the Gulf of Mexico, a secure manufacturing floor or in the wide-open fields of an Iowa farm. And, likely not performant enough for the astoundingly massive data to come from future autonomous vehicles.

Thus, instead of having to transmit data to faraway servers, in many cases it’s more practical to install portable, rugged, data center-grade computing systems onsite. Processing data at the time and place it’s collected means information becomes insight faster, leading to intelligent action that can save resources or even lives.

Future of IoT: From edge-to-cloud

IDC predicted that as much as 45% of IoT-created data will be stored, processed, analyzed and acted upon near or at the network edge by 2019. I think it’ll be more, I believe the data from IoT will be faster and bigger than all other types of big data combined.

Enterprises seeking to seize opportunities created by IoT will have to employ intelligence at many stages of the end-to-end IoT solution — from the thing, to the edge, to the data center and cloud, and lots of places in between.

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.