The Internet of Things (IoT) is conquering the world. At the center of networking the devices are Big Data, which brings in one of the challenges to IoT. It is becoming increasingly important to process data as early as possible. For example, process the IoT data directly in the device without delay.
“The amount of data related to the Internet of Things is enormous – they have to be filtered. A processing directly in the device is already often more efficient. To achieve this, new analytics solutions are needed “, said Dr. Joachim Schaper, the Vice President & Head of AGT International, a pioneer in IoT and Social data management, Big Data integration and advanced analytics.
The challenges of Big Data have included many areas in the IT system. For example, the mass of data from the IoT devices requires immense bandwidth for data transfer and huge data storage capacity. Additionally, the heterogeneity of the unstructured data and wide geographic distribution of data sources are also make the low latency analytics work very difficult.
“Much of the data collected is not usable for Analytics, but the data is still valuable locally. Often it is more effective to provide the practical benefits to achieve a reaction in the device in real-time – with the aim that the “device” learns his typical states and may report anomalies autonomously over time. The costs associated with this type of artificial intelligence capabilities will be used more often in the future, “adds Joachim Schaper.
Robots, as the self-configuring “analytics” components in the complex systems, can process more data today. They can learn through a regular comparison with existing data to identify any critical deviations from standards, and can react immediately in case of need. For examples:
- Companies benefit from Big Data technology in the production process or in the laboratory. If there is a sensor registers open and closed position of the valve, the device can then automatically switch off immediately.
- A camera for monitoring a construction site responds, if persons are at an unusual time and with unusual patterns of movement in their focus. Thus, an alarm can be triggered before the intruder can do damage.
Solution providers have been trying to make Big Data analytics more accessible through innovative applications. Modules which are based on machine learning and are integrated into the terminals will gain much more important in the future IoT. According to AGT International the development of this form of artificial intelligence will change many fields in our daily life.
In the 2015 CeBIT, the largest and most internationally represented computer expo, the panelists made several interesting points that foretell the future of the IoT, and how we’re going to uncover its true value:
- IoT data is rich, derived from a huge number of sensors and delivered in real time; it’s significantly different from traditional transactional data, and so needs to be treated differently.
- The vast majority of IoT data amounts to “noise” if not properly filtered. On this point, Alan Southall of SAP said that most customers are not ready for IoT applications because much of the data is useless. He noted that 60% of data collected from a piece of industrial machinery is unusable for predicting faults in that machinery.
- All panelists agreed that within five years analytics will be embedded in nearly all devices: machines with embedded analytics will enable greater optimization of resources, and predictive analytics will help to significantly reduce manufacturing downtimes, as examples.
- According to David Boundy of Intel and Alan Southall of SAP, edge analytics, or analytics that occur on or near the device, are critical to the future success of the IoT. Analytics at the edge will give rise to machines that interact with backend systems at a higher level than is possible today, and become self-conscious over time.
- David Boundy of Intel mentioned the difficulties inherent in securing IoT data as it moves from the edge to the data center to the cloud, as well as the challenge of protecting people’s privacy as more and more data about them is collected. The panel agreed that there is a need to address these concerns while leveraging opportunities to create new value, such as the delivery of new services, and more efficient use of resources.
- Amr Salem of Cisco commented that our perspective on handling data, and the issues surrounding it, need to evolve, stating that “we need to apply real world terminology and logic to the digital world.” This begins with sensors and sensory data that automate and quantify pattern tracking specific to product distribution and customer behaviors in the physical world. Such data is becoming the foundation for new IoT-based applications and services.
- We need more data scientists! Implementation of IoT analytics requires skilled minds that can extract insights from so much IoT data. Alternatively, or in addition, IoT analytics platforms, such as that developed by AGT, will empower application developers to leverage the output of a smaller number of data scientists.
- Bök of Weidmueller stated that the manufacturing industry will be dealing with challenges related to the analog-to-digital transition for years to come. As the shift progresses, companies will be able to build-out existing services, enrich their customer experience and create alternative revenue streams and business models that capitalize on the new ways in which consumers will buy and use products.