What is Big Data? - Part 3
In the previous articles about big data, I pointed out that prices of data storage and transfer have dropped; it was never this inexpensive to be connected, and connectivity costs continue declining by 25% annually (CISCO, 2013). It became common that most of the mobile phones, portable devices can connect to the Internet, which was quite a novelty ten years ago. This trend is steadily continuing: a research made by Gartner showed that by 2020, there will be 26 billion devices connected to the Internet or each other (Middleton, Kjeldsen, & Tully, 2013).
Internet of Thing (IoT) – as a concept – has been introduced by Kevin Ashton in 1999, by “uniquely identifiable interoperable connected objects with radio- frequency identification (RFID) technology” (Li, Xu, & Zhao, 2015). However this definition has been extended: Pretz characterized IoT as a things-connected network, in which the objects connect via smart sensors and they can interact without human interaction (Pretz, 2013). Greengard simplifies and generalizes the definition. He states that “the Internet of Things literally means “things” or “objects” that connect to the Internet – and each other.” (Greengard, 2015). These devices must have unique identifiers and Internet Protocol (IP) addresses so they can be identified. Greengard does not limit IoT by the type of connection used, but he also agrees that physical-first objects (ABI Research, 2014) can become part of the Internet of Things by equipping them with active or passive RFID tags2 (Greengard, 2015).
The Internet of Things (IoT) means that electronic devices are connected to each other or the Internet. But not just the standard machines with a display, but also household appliances, such as washing machines, lights, wearables, and almost anything. Naturally, IoT is not limited to personal usage: in some vehicles there are already hundreds of sensors that track the condition of mechanical parts, there are smart parking solutions, intelligent street lighting, and the different industries also connect the various machines to the network to collect data like vibration, temperature or productivity (machines in factories have already collected many of these data, but the novelty is to connect them to the network so they can “know” about each other, this way enhancing productivity).
Other researches show an even higher number: a study made by Cisco Systems mentions 50 billion connected “things” (CISCO, 2013). 2 Active RFID tags need a power source, while passive tags can use the RFID reader’s electric power, so they do not need their own power source. This, their long lifetime (they can function for up to 20 years), and their low cost (a few cents per passive tag) makes them more compelling than the active one, though their functionality is also limited (Greengard, 2015).
If a device or machine has an on/off button, the chances are that it will be part of the Internet of Things.
One particular property of this new phenomenon is that machines no longer need human interaction in many cases: by connecting them to a network, we can enable machine-to-machine communication.
While IoT transforms our everyday lives, sensors, and smart connected devices become accessible in the different industries as well. Even moderate estimates show that companies will spend US$500 billion on connected equipment while creating US$15 trillion of value by 2030 (Peter & Annunziata, 2012). And there is a good reason behind this: enterprises can boost their production, deliver better products and decrease costs. Germany's National Academy of Science and Engineering forecasts that operational efficiency can increase 30% by IoT and related processes (Heng, 2014). Some of the key areas, where operations can be made more efficient are better scheduling, inventory management, increased safety, more flexible production and predictive maintenance (Daugherty, Banerjee, Negm, & Alter, 2015). In the Industrial Internet of Things (or Industrial Internet, Industry 4.0 or simply smart manufacturing) there are three typical communication forms: machine to machine (M2M), human to machine (H2M) and machine to smartphone or tablet (M2S). As devices become smarter, they do not need human intervention anymore to work with data that comes from another machine. Furthermore, they can make decisions based on the inputs and previously collected masses of data (Greengard, 2015).
Defects can occur in machines during their operation, causing delays, additional cost for the company and other negative effects. It is not surprising that enterprises were continuously investing in new maintenance technologies that were more economical and efficient. Breakdown maintenance (or reactive maintenance) is the traditional reparation technique, focused on fixing failures that have already occurred. Even though this is the least effective method, it is unavoidable in cases when the error could not have been checked. Preventive maintenance uses scheduling to replace parts on a regular basis (for example, after 600 working hours), and proactive maintenance aims to reduce the number of errors by always tracking failures to their root causes and fixing them. However, these methods are expensive and not always efficient. Scheduling can cause replacement of perfectly functioning parts while fixing already broken machines can result in downtimes. Moreover, the operation site has to keep large stocks of replacement parts that use additional areas and other resources (Scheffer & Girdhar, 2004). Advanced sensors can measure vibration, acoustic emission, corrosion or temperature, and companies can collect this data. Algorithms can be used to recognize patterns in the data, and send signals if something might break soon. Predictive maintenance has many benefits, such as an increase in machine productivity, prolonged intervals between scheduled maintenance activities, improved repair times, increased machine lives, better reparation planning, improved product quality and decreased maintenance costs (Daugherty, Banerjee, Negm, & Alter, 2015). These optimizations can result in saving up to 12% over scheduled overhauls while reducing maintenance costs by 30% and eliminating 70% of the breakdowns (Daugherty, Banerjee, Negm, & Alter, 2015). However, Fox and Do’s research (2013) shows that companies must be critical in implementing big data solutions in their operations as the hype around the technology might be more significant than the actual benefits, and managers should consider many non-trivial factors before investing in them (Fox & Do, 2013).
Equipping machinery with sensors is not a novel phenomenon: 30 years old – or even older – machines have had them. But connecting them to a network and utilizing their data to get additional insights is a paradigm shift that does not only improve the existing processes, but it can bring brand new opportunities to the company. A good example is mining companies. Previously, when a drill hit hard rock, they had to analyze the ore to decide how to proceed excavation. Now they can use predictive analyses built on big data acquired from preceding mining operations, and they have results in a fraction of the time that they needed before. Moreover, machines can make decisions themselves without human interaction to streamline processes (Daugherty, Banerjee, Negm, & Alter, 2015). Drone systems are used to analyze and monitor processing landscapes and measure stockpiles (Wilson, 2015), and they are helping exploration of new mining areas while having less environmental impact (Fiscor, 2015). These, and other innovative applications of new and existing technologies can create new business opportunities in all industries, and they can transform how production companies create values.
The connected devices, automated processes can enhance production, but they also bring new dangers that companies should consider. There can be interruptions in the operations, network problems, sabotage, cyber attacks and data theft (Daugherty, Banerjee, Negm, & Alter, 2015). As more devices are connected, they become more vulnerable to attack. There have already been several cases, where businesses have been sabotaged by hackers: in 2014, they shut down an oil rig and infected another one with a malware. Internet of Things is a new phenomenon and companies only focus on the benefits while they are not entirely aware of the related risks (Wagstaff, 2014). But besides security issues, there are social aspects as well: IoT transforms professions, and it proposes questions about privacy (e.g., tracking employee fatigue or health- related data) and responsibilities (e.g., who is to blame if an automated machine causes an accident). “At the very least the Internet of Things will deliver new challenges and problems revolving around security, privacy, and how we go about living our digital lives. It will almost certainly create new points of contention and dispute among members of society [...]” (Greengard, 2015).
Industrial Internet of Thing is transforming production and operations, but it requires a paradigm shift from companies. Besides optimizing the current processes and discovering new value creation opportunities, they have to expand their risk management practices to be prepared for cyber-security threats.