Importance of Big Data and what do with it
According to an IBM Marketing Cloud study 90% of current data on the internet was generated in the past 2 years. However what exactly does this mean and does it affect our everyday life?
Nowadays, apart from the fact that more and more end users hop on the internet and create regular content, more and more sensors, machines and devices are being connected to the IoT generating oceans of binary data.
According to industry experts we are at the break of the so called “Industry 4.0” that is commonly referred to as the “fourth industrial revolution”. One of the very interesting and peculiar things that industry 4.0 is enabling can be described with the following situation.
Imagine you are at the store buying sneakers. Then you take your phone and scan a QR code on the sneaker, immediately on the screen you get information about where the sneakers were manufactured, when they were released to the market, reviews about the sneakers, expected life duration or travel distance of the sneakers, what materials they are made of, where should you take them for recycling, how much they weigh.
Connected sneakers – why not?
Then you see a button ‘show more’. You press it! More data is shown in organized manner giving you info regarding how much embedded energy is in the sneakers. You get access to a 3D BIM model of the sneakers, you also find detailed tracking information for the transportation of the sneakers from the manufacturing plant to the store. You can even get information about the warehouse at which they were stored and how much time they spent there.
Moreover, you get to know what kind of OS the sneakers are coming with. What information, the added sensors provide for you and your feet, like feet temperature, walking imbalance, inside shoe humidity and sweating rate. The same example can be done also for construction products, clothes and home appliances.
The importance of big data management
As one can see just connecting a one pair of sneakers to IoT creates a lot of data. So much data that if it is not managed correctly, you will get lost in it and you won’t be able to extract any value.
Nowadays more and more assembled or manufactured products have an identification number, connected to data regarding its origin. Thus if screening for any defects or deviations at quality control results in a positive match, the system can get information at which assembly line the product went through and at what time. It would furthermore determine the specific process that caused the defect. Based on information channels like this one the whole system can schedule a predictive maintenance and thus the company can increase their products’ quality, decrease machine breakdowns and reduce time for doing regular maintenance. The result can easily slash the maintenance cost by more than 50%.
Furthermore, among the most promising approaches to utilizing value from so much data will be by using Artificial Intelligence (AI). In the past several years the synergy of AI with Big Data has yield amazing results because of two main factors. One of the factors is the abundance and availability of the data itself. The other factor is the computing power. In 1997 1.3 teraflops were worth 55 million dollars and took 150 m2 of space. Last year, NVIDIA released Titan worth just 3000 dollars clocking at 110 teraflops. So how can we benefit from that?
Google made digital twins for their data centres, then used AI to optimize the control of its HVAC system depending on the load of the data centre. The result was 40% increase of the Power Usage Effectiveness (PUI). In essence thanks to their detailed data centre twins they were able to perform a sensitivity AI-enabled analysis of the HVAC system and thus find the optimal working point for each load. In reality that would be close to turning all the knobs and switches of the HVAC system in different configurations and test the system behaviour for different work loads in different ambient conditions. In the digital world this happens in a matter of hours, whereas in the real world it may be not even possible to test all the different scenarios.
As more and more devices are being connected to the IoT their impact on our lives would increase and at the same time our interaction with them would become more seamless and subtle. As practice shows – you can improve only what you can define and measure. The new ‘linked’ world would enable us to track and measure more of our daily life, resulting in the desire to improve upon the daily KPIs.
The importance of big data interoperability
Finally, however, we have to raise an important issue. The issue of data interoperability. Data about products everywhere around us is abundant –yes, but it is often stored in unstructured formats that are hard to process, however big the computing power. This greatly limits the possibilities of utilising its great value. The way one manufacturer defines the inside shoe humidity of the sneakers they produce may greatly vary from how it’s done by another – the same applies to the HVAC systems manufacturers. How can a user compare between the two. The answer is through standardisation of data structures and intricate mapping between data points. This lays the fundamentals needed for all the great benefits promised by IoT, BIM models, AI and digital twins. This is exactly what we at Cobuilder are good at.
In order to get to a level where buildings take ‘conscious’ decisions about the microclimate inside, reduce their resource consumption, schedules predictive maintenances and take good care of their occupants, we need to look at the fundamentals first. How should data be structured? Where should data relationships be stored? Who is the most credible source of data? What is the solution to semantic differences?