IoT Analytics: Scope and Applications

Ever since its inception, IoT has been transforming the way day-to-day tasks are being completed, businesses are being run and services are being provided. The effect of IoT is so pervasive that it has given rise to  terms like Cloud Capitalism and Cloud Economy, probably referring to staggering projected size of $100 trillion for digital economy, the scale and grandeur earlier reserved exclusively for oil industry. Unless you are living under a rock, chances are there that you have heard and probably used some of the IoT consumer devices like Amazon Echo, Google Home, Logitech Pop, NestCam, Ring which lash you with comforts like digital concierge, smart button control and safety and smart temperature control.

Given the wide range of applications of IoT in the enterprise set-ups as-well-as ample opportunities to create consumer electronic goods based on IoT, the sky is the limit. IoT is going to give rise to a $250 billion industry out of which services, analytics and Application development would account for more than 60%.

Every opportunity comes with an imminent threat of squandering it. Same is true with IoT. If the upstream and downstream technologies required to harness the power of IoT is not learnt and used, the ultimate potential of IoT wouldn’t be realized.

Pre-requisite

The problem solving on IoT use-cases requires a holistic skill-set including but not limited to the following: i) tech stacks needed to deploy IoT systems, ii) store/stream the high frequency data, iii) understanding & implementing predictive analytics algorithms, iv) understanding the business use cases involving IoT to deliver the maximum impact.

The technology stack required to successfully execute an Iot Analytics project includes the following among other things:

• Enterprise tools – expertise with applications like IBM Watson IoT platform, Azure IoT suite
• Allied tools – expertise with R, Python and other open sourced tools to build and deploy custom algorithms
• Databases – expertise with NoSQL databases like Cassandra, MongoDB, Oracle NoSQL, PI Historian Time Series database, Splunk, OpenTSdb etc. which are amenable to storing IoT kind of data

There are two service lines or career paths an organization or an individual can tread onto to make the most of the IoT revolution

• Data Science & Consulting – a team of data scientists who have implemented the algorithms like SVM, Neural Networks, Shapelets, Random Forests and consultants who understand and take keen interest in your business
• Product development – product development acts like an essential ligand which connects all the complex maths and tech in the background to the fingertips of the end-user. That’s why we have a dedicated team of developers who would bring any application to the user in a jiffy.

Scope

An organization taking up IoT Analytics offerings could i) develop new numerical methods or customize the existing ones to enhance performance, ii)scout for new enterprise and consumer problems which can be solved through IoT, iii) develop plug-and-play API either through leveraging the existing IoT tools or creating standalones and iv) create a knowledge base for using existing IoT tools.

In terms of tech infrastructure, the offerings and research in IoT Analytics can be categorized in the following verticals:

Case Study

At Noah Data, we have delivered several projects for enterprise clients wherein we have:

1. Case1 – Helped an automotive client set-up big-data infrastructure for real-time data streaming and processing and performed analytics based on past trips for a connected car application
2. Case2 – developed the data infrastructure to handle IoT data coming from sensors installed to look after the health of electrical submergible pumps(ESP) and a Time-to-Failure model to predict in advance as to when a pump would fail reducing a lot of downtime and repair cost.

The analytics & tool development involved the following steps:

Step 1 – ingestion of time-series IoT data from ESP sensors to NoSQL database

Step 2 – implementing Shapelet algorithm to predict Time-to-Failure

Step 3 – automating push notifications about failures to stakeholders

Step 4 – creating a user interface to monitor and manage the entire process

These experiences exposed us to the bountiful world of IoT and positive feedback we received from our clients encouraged us to explore further and transform it as one of our core strengths.

Please stay tuned to our blog for the next instalment of the IoT series wherein we would take a detailed approach of some of the techniques and algorithms, such as kNN for time-series classification, we have used to solve some cutting-edge IoT-based industrial problems.