How IoT will Disrupt Enterprise Computing

Mu Sigma defines itself as a category-defining decision sciences and big data analytics company, helping enterprises 'systematize' better data-driven decision making. The company believes that its interdisciplinary approach and integrated ecosystem of platform, processes and people are redefining how companies approach problem solving in areas of marketing, risk and supply chain gives it a distinct edge in the market. With more than 3,500 decision scientists working across 10 industries Mu Sigma says that it has been consistently validated as the preferred decision sciences and analytics partner for large enterprises. In an interview to DATAQUEST, Ganesh Moorthy - Apprentice Leader, Mu Sigma, talks about  IoT evolution and its impact on enterprise IT organizations. Excerpts

Can you talk about the multi-pronged impact of IoT and how it will disrupt enterprise computing? 

As devices get smarter and handle more computing on the edge, the volume of data being routed through the enterprise network will drop. But as IoT devices become ubiquitous, overall data volumes will increase tremendously. Enterprises will have deal with this deluge of information at scale and increasingly in real time. Machine data will drive even greater demand for big data storage, and new, more adaptable scale out models for server farms. Now add in the wide variety of communication protocols, and complexities in areas of network management and information management are compounded.

Owing to varied locations of devices (within or outside of the firewall) network security will be critical and will require new policies be put in place. Analysing disparate data will become the main focus towards insights for predictive and/or preventive purpose. Enterprises will need to invest on real time complex event management systems to keep up with information stream.

Do you think IoT has passed the hype curve and on its way to adoption? 

IoT has certainly passed the hype and surely on its way to enterprise adoption. We are seeing a lot more use cases for IoT ranging from accessing information from fitness devices to smart cities, where every device is interconnected for most optimized living conditions. Telematics is probably one of the biggest use case right now.

How will analytics aid in better IoT? 

IoT and analytics go hand in hand. Take fitbit for example. While the device itself has gyrometer, accelerometer and pulse monitor to measure your activities and heartbeat, you need a corresponding application to analyse the pattern. A more complicated application is facial recognition through videos. The video camera captures live feed, which are decoded and run through a number of modelling techniques to detect facial features for identification. We have been experimenting with developing analytics on small computers such as Raspberry PI and Arduino towards edge analytics where most analysis takes place on the device itself. Results are then sent to the server for historical pattern detection. We have successfully used similar approaches to determine on-shelf availability of products in real-time, sending out notifications to appropriate stakeholders for restock. This approach resulted in significant saving in combination with operational efficiency and reduced human effort.

What role is Mu Sigma playing in IoT? 

Mu Sigma continues to innovate in the development and application of intelligent devices that incorporate edge computing for variety of business purposes. We’re building enterprise signal processing systems to process streaming, real-time data and visualize insights in a way that makes intuitive sense to business decision makers. We’re working on streaming video data and analytics capabilities for large retailers and hospitality companies, RFID sensor data and machine level data from wearable devices, using a custom big data platform built on JADE, Spark and Storm. We’re also helping clients collate machine data from legacy industrial appliances, and using this information for preventive maintenance purposes.

If you say that you are building enterprise signal processing systems – can you throw more light on the specifics?

Our Enterprise Signal processing system is a state of art framework to help build real-time complex event processing based applications. This can be used to detect signals from data coming from IoT based devices or any real-time feeds and perform the necessary actions as required. This platform allows for users to build custom event processers using flow based paradigm and comes with highly configurable interface for front end visualization. It supports both parallel and distributed computing and scales on demand.

What according to you are some of the inhibitors on the road to IoT adoption?

While some segments like manufacturing have started adopting IoT significantly, others are facing challenges understanding the business impact of IoT itself. For those who do make a valid justification for IoT, lack of experienced skillset is the next big hurdle as it involves dealing with hardware, embedded software and advanced machine learning / deep learning techniques sometimes. Finally, data security and privacy concerns need to be addressed as Machine to Machine intelligent communication becomes ubiquitous in the years to come.