Big Data, Analytics, and the Internet of Things

Big Data, Analytics, and the Internet of Things

Mohak Shahanalytics leader and research scientist at Bosch Research, USA

Technological advances are making it possible to embed increasing computing power in small devices as well as capture live streaming data from them at near-live speeds.
These advances in the areas of sensing technology in conjunction with communication and computing technologies are resulting in a world of interconnected devices typically referred to as the Internet of Things-IoT (also the Internet of Things and Services – IoTS, and the Industrial Internet, depending on the context). While this ability allows data-generation at scale, the advances in machine learning have allowed building models on this ever-increasing amount of data. Data analytics in general and big data technologies (for big data not just in volume but also in velocity and variety) in particular are already playing a crucial role and are poised to become increasingly important as the IoT related technologies and penetration of connection-ready devices increase.
A recent Gartner report estimates that, by 2020, this network of interconnected devices will grow to about 26 billion units with an incremental revenue generation in excess of $300 Billion. Most of this revenue gain will appear in services. Further, global economic value-add through sales into diverse end markets would reach $1.9 trillion.
Consequently, the data resulting from these devices will grow exponentially too resulting in new business opportunities as well as posing novel challenges to managing and processing it for value gain. The data of the digital universe is slated to grow 10 folds by 2020. Various research and analysis firms have confirmed the scale of these projections in addition to the Gartner report. IDC further notes that data just from embedded systems, i.e. sensors and physical systems capturing data from physical universe, will constitute 10% of the digital universe by 2020 (this currently stands at 2%) and represent a higher percentage of target-rich data.

Big data technologies and the IoT are playing a transformative role in the society. The pervasive and ubiquitous nature of such technologies will profoundly change the world as we know it, just as industrial revolution and then Internet did in the past. Various visions of the IoT including the things- oriented vision (focusing on devices), internet-oriented vision (focusing on communication and interconnectivity) and semantic-oriented vision (focusing on data management and integration) have been presented in the past.
However, from a value perspective, I believe a functional vision of the IoT is necessary, where the resulting data and insights, and not the enablement mechanisms, play a central role.
This vision goes beyond the enablement stack composed of data acquisitions and protocols, big data infrastructure, data management and integration challenges, and developments in the machine learning and data mining capabilities.
The functional vision enables an integrated view of the value- creation that can happen as a result of bringing these technologies together leading to novel business models as well as revenue sources, diversification of revenue streams in addition to increasing visibility and operational efficiency. Further, this interconnectivity offers opportunities for enhanced services that can supplement each other by means of derived and abstracted insights – higher level of observations and inferences that can be made from data arriving from multiple interconnected devices.

Given the data acquisition capabilities already in place in the context of monitoring of various physical assets, the immediate opportunities are bigger from the industrial perspective. Data analytics capabilities are already impacting a broad range of areas from manufacturing, asset and fleet management, operations management, resource exploration, energy, healthcare, retail and logistics, as well as connected assets. On the consumer end, we are currently undergoing a transformation, as more and more physical devices capable of advanced sensing become part of our routine life.
Consumer applications will start witnessing a rapid growth in integrated services and systems, which I believe will generate much more value in contrast to one-offs, once a critical mass of such interconnected devices is reached in various domains.

The capabilities in leveraging big data in both the industry-, and the consumer-contexts are already transitioning from performing descriptive analytics to predictive analytics. For instance, based on real- time sensor data, we can predict certain classes of field events (e.g., failures or malfunctions) for heavy assets such as aircraft engines and turbines more reliably, and complement physics-based models employed in such cases. As these technologies mature, they will enable another transition from predictive to prescriptive analytics whereby recommendations on resolutions of such events could be made even to the extent of devices themselves taking corrective actions making them self-aware and self-maintaining.

Even though we are already witnessing transformational changes, much more needs to be done on various fronts be it advancements in the big data technologies, analytics, privacy and security, or policy. These challenges can broadly be categorized into technology challenges and adoption challenges. The former category covers relatively well understood, even if not yet resolved, issues such as security, interpretability of models and data quality. On the other hand, the adoption issues are more profound and our understanding of their scope as well as scale will grow as this transformation progresses. One such important adoption challenge is that of human-analytics interaction. As decision-making becomes more automated, it is not clear what role humans will play and how will they even react to condoning or correcting these automated decisions. For instance, automated approaches certainly have the potential to bring down the variations resulting from manual approaches from multiple sources and hence are robust in this respect. However, in some cases such variations are desired, even required, so that we can advance our understanding through a multitude of perspectives. While it may seem a bit early, it is timely to start seriously discussing how these issues will evolve and have an impact on adoption of respective technologies.

For a detailed overview of the above topics, please refer to the upcoming book chapter on the subject.
Preliminary draft available here.

You may also like...