By now you might have read about our versatile high performance and low power IoT sensor hub MAREN (if not click here). The feature-rich reference design platform which is equipped with Bluetooth and NFC as standard can be extended with additional connectivity solutions via Atmel XPLAINED PRO interface. A couple days ago we described the possibilities created by using the Atmel ATA8520-EK-3 extension in order to connect MAREN to the Sigfox network (read more here).
In this article we will take a look at a new solution based on the Atmel ATWINC1500-XSTK SoC for WiFi connectivity. EBV has partnered up with IBM and therefore can now offer full support for IBM’s IoT platform Watson as well as for the Watson Quickstart sandbox. The combination of MAREN and Watson creates a versatile platform well suited for machine learning applications.
Data Needs Watson
In order to explain the benefits of this powerful partnership let’s take a look at Watson and it’s relevant features.
The platform was developed by IBM in order to address the increasing problem of “data waste”. Let’s add some perspective to this: According to IDC from beginning of time through 2011 the world contained barely one (0.9) zettabyte of data. With the dawn of the IoT data amounts started to rise quickly and in 2011 alone 1.8 zettabytes have been created. Estimations forecast that we will be generating 35 zettabytes of data per year by 2020.
Those huge amounts of data bring several challenges. Structured data might be comparably easy to manage but too often we store more of it than we will ever use (not even to mention the amounts in excess data created through duplications).
The biggest chunk of data however is unstructured (e.g. videos, texts, audio…) and neither easy to search nor to interpret and manage.
This has lead to the phenomenon that most companies store huge amounts of data while they use only a fraction (only 32% of structured and 12% of unstructured data are used on average according to CIO Insight).
As a result we waste money and effort (data management, back ups, storage etc.), add complexity (infrastructure for huge amounts of data) and increase security risks (the more data the harder to secure).
For IoT device manufacturers as well as Industry 4.0 projects this is bad news. In order to optimise processes and to profit from data it is not enough to be able to search it and to discover files and data points. The full potential of the data can only be unlocked if we are able to make sense out of it and to analyse and connect information in order to make decisions and predictions based on all relevant data.
A Cognitive System That Understands Nearly Everything
IBM Watson has been developed specifically to manage and analyse huge streams of data. The cognitive system can be fed with documents relevant for a specific domain and is thereby “trained” to understand the topic. The system is able to process and understand information and can provide humans with suggestions based on all data consumed whether structured or unstructured.
This works with everything from sensor data, over emails, old archive files to social media posts. During it’s lifetime Watson is constantly analysing the stream of data created by your company, machines and products. Therefore the system is able to filter and interpret the huge amount of information in your enterprise in order to deliver the perfect snippet and combination of data to you to make the best decision based on all the structured and unstructured information available.
As Watson is able to connect all kinds of business layers, from sensor to the cloud the suggestions are more accurate than anything single humans could usually achieve. At the same time the system is able to learn about restrictions and security and thus to determine who can see which data (learn more about Watson in the white paper “Watson – A System Designed for Answers” and the video below). In addition, WATSON is able to correlate your data with external sources (like weather) to find another sense and help to refine your predictions and decisions.
Combining MAREN’s Sensors & Watson’s Cognitive Abilities
In other words with IBM’s IoT platform we move from (big) data analysis to data understanding. In combination with MAREN this enables massive new opportunities in regards of machine learning. IBM offers an API specifically dedicated to this task. It enables Watson to automate data processing and continuously monitor new data and user interactions. The results are ranked on learned priorities. This can be applied to any data coming from devices and sensors to automatically understand the current conditions, what’s normal, expected trends, properties to monitor and suggested actions when an issue arises.
The sensors of MAREN are the perfect prerequisite to collect important data in machine learning projects. Watson is able to learn normal and abnormal conditions based on MAREN’s information. This concept can be applied to manufacturing equipment in a production line in order to reduce machine downtime, prevent supply shortages and to optimise production steps. At the same time the data from the environmental sensors in combination with user data can be used by facility management companies in order to manage applications like AC and lighting more efficiently. Turning to the consumer segment you can also use the sensor data collected in your products to analyse customer behaviour and to identify new opportunities and room for innovation.
Ok now let’s get to the verdict of all this: why should you be excited about the marriage of MAREN and Watson? On the one hand MAREN was created with easy development for the IoT in mind. Equipped with MEMS sensors (motion, environmental) and sensor processing algorithms (sensor fusion) the reference design can be used in complex industrial and automation projects but also in simple consumer applications. Together with Watson it is feasible to analyse and connect all data created by you, other systems and your applications and to make decisions based on close to 100 percent of the information available. With Watson MAREN becomes a true end-to-end solution from sensor to cloud. As Watson is highly scalable, meaning it will constantly grow with and adapt to your projects, you are able to choose a future proven and flexible solution.
MAREN is compatible with the Watson QuickStart Sandbox thus you can get started with prototyping as quickly as possible. Whenever you’re ready it’s easy to switch to the full Watson IoT platform supported by the EBV reference design.
In addition EBV is constantly refining MAREN to create all the flexibility you need to translate your projects into reality. Contact EBV for inquiries, additional know-how, design support and hardware here.
Get access to the extensive technical resources for MAREN here.
cover: IBM Watson Headquarter in Munich (image: IBM)