5 min read The Right IoT Solution : A Hard Choice to Make Reasons to [...]
Reasons to connect everyday products are no longer confined to the ‘Jetsons’ fantasy of an automated world. As technologies become more affordable and accessible, more and more consumer and industrial devices will connect to the Internet. But before halting operations and jumping into a connected IoT project, it is important to understand why connectivity is important. As previously summarized by Harvard Business Review, some of the leading reasons driving IoT solution strategies today include:
With now over 400 companies offering IoT platforms to connect and manage connected products, several industry analysts have reported on criteria that should be taken in consideration when trying to select the right solution. This list of criteria from Forbes highlights some of the most important ones:
RELATED: interested in monetizing your data? click below to read more!
While these are very relevant, they focus on the hardware-to-cloud part of the IoT stack (i.e. connectivity and device management layers). To add to this list, we will focus on the criteria that pertain to the most talked about (and important) part of the IoT stack: the data analytics layer.
An IoT project without advanced data analytics is like a football team without a coach. Every decision is taken individually, based at best on instinct, sometimes out of fear, and without any coordination and common goals. And like the coach of a football team, to be able to adequately evaluate the performance of the team, assets strengths and weaknesses, adapt to the need of the group and communicate instructions, success is based on intimate and in depth knowledge of each and every player on the team. This is why the more knowledge that is captured and readily available on the connected products and their environment, the more efficient and accurate the business decisions will be. We hence describe 5 additional criteria to be carefully weighted and discussed when selecting a data-first IoT solution:
back in the days of M2M, sharing time-series alone (e.g. temperature, humidity, vibration, etc.) was often enough for two systems to cooperate. Now in the IoT era, capturing and sharing events (e.g. button pressed, panel on, etc.) is as important as the time-series themselves to efficiently analyse connected products and to succeed when undertaking IoT projects. Furthermore, the coverage and diversity of these captured events should be carefully defined. A data-first IoT solution should be able to manage:
And finally, events triggered by humans, like operators, service technicians or customers should be distinguishable from automated events (e.g. door opened, button pressed, filter replaced, etc.).
Integration with third party solutions and existing lines of business applications is of the upmost importance to further increase the value of actionable insights. A data-first IoT solution should be able to easily (and cost effectively) stream data out in near real-time by providing appropriate publish/subscribe services. These include, but are not limited to, Web sockets, MQTT and AMQP, etc.
To achieve the high levels of accuracy required in production environments, prediction models often require large training and test sets of data that span across multiple months and often times years. A data-first IoT solution should scale enough to be able to appropriately store and manage the very large sets of data required to train advanced AI models and make accurate predictions.
With a strong need for standardization in the IoT space, and to improve the time-to-insight, data-first IoT solutions should natively embed (and comply to) libraries of out-of-the-box, industry standard data models. These include, for example, models and data structures from the OGC, the OIC, Project Haystack, oneIoTa, or the W3C. Use of standard models not only facilitates data exchange, but the rich descriptions they provide (e.g. points on the evaporator, heating modes, harmonic distortion, cooling capability, etc.) maximizes possible correlations between the sensor data and the deployed connected ecosystem.
Data enrichment refers to processes used to enhance, refine or improve raw data. Enrichment contributes to making data and ultimately insights a more valuable and actionable asset. A data-first IoT solution should be able to easily integrate with external enrichment services (e.g. geo location, weather, crime rates, population density, etc.) as well as internal services (e.g. CRM data, ERP data, asset management data, etc.).
By now it should be clear, that the importance of a data-first solution cannot be understated. But in addition to choosing the right solution, organizations must ensure they can quickly and efficiently support new data sources at larger and faster speeds.
There is great value in IoT data, and even greater value in actionable data insights. But without the right IoT solution to query, visualize, analyze and share the results, data is just another unchecked box on a growing to do list.
Find out how Mnubo is tapping into the benefits of product data by transforming raw sensor data into actionable business insights. Read more here!
Thanks for reading! If you have an opinion, issue or simply wish to elaborate on the article, feel free to participate in on Twitter!