The Analytics of IoT

If you look at the pace of growth of the Internet of Things (IoT) devices today, it is all pervasive.  Everything is smart – from pens, lanyards, and switches to air conditioners, traffic lights, and refrigerators. Today almost every retail company has smart tags associated with every SKU on every aisle, which monitor when it is getting depleted so that it can trigger of an alert to an inventory management system which will trigger a replenishment procedure and enabling the item to be on the way into the supply chain in a real time.

Likewise, we could talk about a smart refrigerator which could detect when the milk carton is empty then place an order by connecting to the order management system. A smart city would enable real time capture of the traffic scenario across major streets and warn office goers on the conditions in real time to enable them to take alternate routes.

On the scale of data generated by these devices, to get an estimate let us just count the number of IoT end points or a number of devices with each one capturing different parameters (for example there are sensors which capture multiple parameters like temperature, a pressure of a room etc.).  This number runs into billions worldwide.

Imagine the final volume of the data that is generated by these devices including sensors, actuators, tags etc., at regular intervals of a couple of seconds gap, i.e. with very high velocity. These numbers lead to a data nightmare, namely of how to make use and sense of all the data generated.


The sheer volume of data presents quite a few challenges and raises more questions than answers.

  1. Is all the generated data going to be stored?
  2. Will it be possible to derive meaning from the data generated by higher level analytics?
  3. Is it possible to preserve meaning from the data without losing any data?
  4. Can we handle the high velocity of the data stream from the IoT devices?
  5. Can we derive insights via analytics on fast-paced data?
  6. How do we secure IoT data and prevent misuse?
  7. How do we handle the low power of the devices in IoT while preserving data?
  8. How do we enable reliable data communication between sensors, from sensors to applications, or even from sensors to cloud?

These questions need to be answered in depth if we were to take IoT utility to next level via absorbing data analytics as a logical extension to all the innovations that have happened in IoT world on devices front.


IoT Analytics is broadly an emerging area of data science and data analytics which is able to provide answers to all the questions raised above.  In particular, the challenges which are of focus in IoT analytics include but are not limited to:

  • Streaming Analytics Algorithms
  • Real-Time Analytics Algorithms
  • Sampling Techniques for Real-Time Data
  • Storage Techniques for Real-time data
  • High-Performance Analytics
  • Reliable Data Messaging
  • Handling multiple communication protocols
  • Security of Streaming Data
  • Anomaly Detection of Streaming Data

In any use case of IoT, be it in retail, transportation, supply chain, energy efficiency, sports, healthcare, or agriculture, its successful deployment now lies in completing the loop of data science as applied to the data generated by the sensors/tags/actuators/beacons etc. in the IoT ecosystem.

Let us take an example of how IoT is used in sports. In the 2016 edition of the Olympics, several use cases of IoT were visible, be it the boxing sensors used by boxers or Google glass like sensors used by cyclists. In case of the boxers, the sensors were measuring primary data like punch intensity, punch numbers etc. The derived analytics on these primary data included a relative comparison to personal benchmarked best, an analysis of the relative ratio of the types of punches, a comparison to the benchmarked ratio of shots.

This is a good example to show how basic IoT data captured by a sensor is combined with higher level analytics to evaluate an individual boxer’s performance in terms of both personal best and comparison with competitors. Such intuitive and innovative use cases can be seen across sectors applying IoT.


The full capability of IoT for business use lies in applying analytics to the data generated by the IoT ecosystem. However, this area of IoT analytics is just about emerging and needs to evolve to handle all the complexities related to streaming and real-time data. At the same time, there is a need to understand that for data science itself this is posing a new challenge in terms of the need for new kind of algorithms capable of dealing with high velocity and high volume streaming data.

Let the synergy of IoT and Analytics begin.

Dr. Srinivas Padmanabhuni, President of ACM India and co-founder of Tarah Technologies
Dr. Srini is a co-creator of our Internet of Things learning path. He has a rich experience of over 15 years in IT Industry, having given more than 100 expert invited talks across universities in US, China, Australia, Canada, Singapore, UK, and India, including universities such as Carnegie Mellon, Purdue, RUC etc. He is a prolific and astute researcher who has seven granted patents, 15 filed patents, a published book by Wiley, several book chapters, and 70+ refereed journal and conference papers to his credit, in addition to marquee invited talks and editorial positions. He also holds a Ph.D in Artificial Intelligence from University of Alberta, Edmonton, Canada.

Related Articles

Please wait while your application is being created.
Request Callback