By Ravishankar Kandallu, Data Scientist and Statistical Modeller, Tata Consultancy Services, MESA Gold Keystone Sponsor
With the advent of digital era,
the world is currently witnessing a constant connectivity growth among people
and equipment units due to rapid technology evolution. A scenario unimaginable a few decades ago,
where a large array of smart and intelligent home appliances and vehicles
communicate and transmit data about their usage is slowly gaining prominence. Internet of Things (IoT) forms the
cornerstone technology behind this evolving mammoth technology shift.
- World’s renowned community of technology practitioners and consultants namely, Wikibon, estimates a 14 percent economic value addition growth amounting to USD $ 1279 Billion from IoT by 2020.
- A Mckinsey report also emphasizes an explosive trend in IoT citing a 300 percent growth addition in the number of connected products in recent years. Also, this trend is set to accelerate in future.
- A recent KRC end-customer survey in advanced countries highlighted a growing market for smart home appliances, smart energy meters, wearable devices, connected cars among other IoT devices.
- In the manufacturing front, IDC reports that manufacturing industry world-wide invested USD $ 178 Billion in IoT during 2016.
Growing IoT spends in
manufacturing signifies a future laden with overwhelming existence of smart IoT
enabled products. A Verizon survey indicates that large section of early
adopters of IoT among manufacturers are keen to apply IoT to understand the
customer preferences and product reliability. Such an evolving landscape of
smart products present a plethora of opportunities for manufacturers to excel
in customer service. With IoT, manufacturers can proactively avert all
products’ malfunction through predictive maintenance, thereby, delight the
customers. However, achieving this feat
requires a whole new technology infrastructure as explained further.
Evolution of smart and connected
products is driving manufacturers to completely overhaul, rethink and redesign
the conventional value chain. Reason,
smart products consist of three components, namely:
- Physical component: Refers to basic mechanical product such as a car or refrigerator.
- Smart components: Includes sensors, microprocessors, data storage, controls, software, embedded operating system and digital user interface.
- Connectivity components: Ports, antennae, protocols, networks, cloud with servers and product’s external operating system.
- These three components are crucial for a smart product to function in an extra-ordinary manner by enabling smart products to:
- Monitor and report their own environmental conditions and offer insights to their current performance
- Enable users to remotely control complex operations based on their usage
- Facilitate manufacturers to apply algorithms based on existing performance and usage data (stated in above points) to improve the performance further
Thus, data on usage and
performance collected from the products-in-use form the core element that
drives business operations in IoT era.
For example, a smart air-conditioner can sense raising levels of ambient
temperature that vary with seasons. By
correlating these readings with external weather data and warranty claims, a
manufacturer can ascertain the possible growth in usage of air-conditioners in
locality. This growing usage in a
prolonged manner in turn causes possible part failures that can be again
ascertained by applying algorithms on a combination of performance, usage,
weather and warranty claims data. This
enables manufacturer to plan and stock spare parts at right levels in a given
locality and also gear-up the field service team to proactively avert major
failures.
Thus, driving factors for
customer service excellence in an IoT era primarily hinge on the following
elements of technology stack that forms the foundation of IoT:
- Designing products embedded with right type of sensors that transmits precise performance measures in a required manner.
- Choice of right type of IoT architecture and protocol that enables manufacturer to access the data from smart products is another factor that impacts the overall performance.
o
A typical IoT architecture includes several
layers such as object abstraction layer, service management layer, application
layer and business layer.
o
Renowned international organizations such as
Institute of Electronics and Electrical Engineers (IEEE) and European
Telecommunications Standards Institute (ETSI) have defined several protocols related
to IoT architectures
o
The prominent protocols include Constrained
Application Protocol (CoAP), Message Queue Telemetry Transport (MQTT),
Extensible Messaging and Presence Protocol (XMPP), and Advanced Message
Queueing Protocol (AQMP)
o
Each protocol has its own merits and
demerits. Thus, ingenuity of
manufacturer in making the best choice plays a vital role
ü
Application of right algorithms on the streaming
data from smart products is crucial in spearheading customer service.
o
Performance
analysis and predictive maintenance:
Extraction of performance and usage data from smart products on a
continuous basis creates a perfect environment for conditional monitoring of
products-in-use. Algorithms such as
Naïve Bayes Classification, Hidden Markov Models and Reliability Modeling offer
an excellent approach to predict the impending failures by correlating the
streaming usage and performance data with warranty claims. Under complex
circumstances where numerous factors drive a failure, neural networks act as a
handy tool to predict part failures leading to product malfunction.
o
Understanding
customer preferences through sentiment analysis: Along with usage and performance data,
customer expressions in various internet blogs related to product performance
provides a real-time feedback on product performance to the manufacturer. Thus,
web scrapping and analyzing such blog comments offers opportunity for
manufacturer to measure the customer satisfaction levels. Algorithms that perform sentiment analysis
followed by association analysis offers a means to ascertain the customer
preferences.
o Tools to
Run the algorithms: Both open-sourced and licensed versions of the tools
are available to apply algorithms on data from smart products. Open sourced
machine learning libraries such as Tensorflow, Keras, Theano that operate on
Python environment are prominently used to build algorithms. Also, R is the other major tool used to build
algorithms in IoT. Apart from that,
licensed tools such as SAS, SPSS and Matlab are also used by manufacturers to
analyze and gain insights from IoT data.
To summarize, this blog emphasizes the enormous
opportunities presented by IoT enabled smart and connected products for
manufacturers to excel in customer service.
Predictive maintenance is a key weapon that manufacturers can bank upon
to attain customer service excellence.
At an operational level, the technology stack that enables manufacturer
to achieve this feat is outlined. More
importantly, the factors that drive manufacturers towards instilling a
responsive and efficient predictive maintenance environment is detailed.
Contact
For more information, contact manufacturing.solutions@tcs.com.
About the Author
Ravishankar Kandallu
Data Scientist and Statistical Modeller
Tata Consultancy Services
Ravishankar Kandallu is a statistical modeler and data scientist within the Business Intelligence and Analytics Center of Excellence in the Manufacturing business unit at Tata Consultancy Services (TCS). He has worked on analytical modeling projects in the domains of warranty analytics, forecasting, and network optimization for high-tech, manufacturing, and consumer products industries. He has a doctorate in supply chain management from the Indian Institute of Technology (IIT) Madras. He is also a certified six-sigma green belt.
About the Author
Ravishankar Kandallu
Data Scientist and Statistical Modeller
Tata Consultancy Services
Ravishankar Kandallu is a statistical modeler and data scientist within the Business Intelligence and Analytics Center of Excellence in the Manufacturing business unit at Tata Consultancy Services (TCS). He has worked on analytical modeling projects in the domains of warranty analytics, forecasting, and network optimization for high-tech, manufacturing, and consumer products industries. He has a doctorate in supply chain management from the Indian Institute of Technology (IIT) Madras. He is also a certified six-sigma green belt.
No comments:
Post a Comment