Smart devices – What are sensors?
A sensor is an artificial implementation of what is called a sense in biology. With a sensor, a machine observes the environment and information can be collected. A sensor measures a physical quantity and converts it into a signal. Sensors translate measurements from the real world into data for the digital domain. There is an almost infinite diversity of parameters that can be measured, such as location, displacement, movement, sound frequency, temperature, pressure, humidity, electrical voltage level, camera images, color, chemical composition, etc.
The goal is to detect events or changes in the environment. A sensor is always used with other electronics, as simple as a lamp or as complex as a computer. Advanced chip technology makes it possible to integrate all the required functions at low cost, in a small volume and with low energy consumption. The number of sensors around us is increasing rapidly. Estimates vary, but many expect that by 2020 more than 50 billion sensors will be connected to each other via the Internet of Things (IoT).
But why would you use sensors? Or in other words: why would you want to measure? A well-known (Dutch) statement is: “to measure is to know”. This refers to the great importance of carrying out measurements to make concrete, factual information available. You can compare, calculate, predict and check with figures. Measuring provides insight into things that go well and that do not go well. By measuring you check whether you have done what you intended to do and whether your goals were achieved. If you measure, you know where you are now; you know the current situation. From there you can always get better. You can therefore learn and improve by measuring.
The devices, which together form the Internet of Things (IoT), are equipped with sensors. With these sensors, the devices collect data about the way they are used and about the environment around them. The collected data can be as simple as a measurement of the temperature or as complex as a complete video feed. But also think of sensor data in the form of location, sound or humidity, and different measurements of machines or of our bodies. These devices have built-in (wireless) connectivity so that they can be connected to the Internet and can exchange data. Billions of connected devices are part of the IoT. A side effect of IoT is that all connected devices generate a huge amount of data (Big Data).
IoT makes an endless supply of information available that was previously unavailable. And if sensor data was already present, it was difficult to analyze because it came from all sorts of separate devices. With IoT we record continuous measurements of various sensors, which we can easily read out. This allows us to recognize trends and make predictions.
IoT makes our lives easier (the smart thermostat), shifts our focus to efficiency (the washing machine starts when the prices of electricity drop), and helps us anticipate (body monitors that provide us with information continuously). The expectation is that IoT can offer solutions for major social problems relating to energy, the environment and crime. Over time, using IoT, for example, we will use less energy, waste fewer products and spend less money. To give a very concrete example: thanks to IoT, trash cans can let us know how full they are and thus whether they need to be emptied. Those who know how to make use of such information can work much more efficiently.
The IoT is growing exponentially: there are more and more devices that collect, store and exchange data. In addition, consumers, organizations, governments and companies themselves produce more and more data, for example on social media. The amount of data is growing exponentially. People speak of Big Data when they work with one or more datasets that are too large to be maintained with regular database management systems.
More and more you hear that Big Data describes a development. It contains two components. First of all computer technology: the increasingly sophisticated hardware and software that makes it possible to collect, process, and store more data. The second component is the statistic that makes it possible to find meaning in a collection of separate data. Big Data in this definition refers to our possibility to analyze and use the ever-increasing amount of data. Big Data is essentially about realizing added value from the processing and analysis of data. Characteristic is that it concerns unstructured, varied data from different types of sources that are processed in real-time.
Big Data is playing an increasingly large role. After all, these data contain a wealth of information for various purposes, such as marketing, scientific research or preventive maintenance. In order to actually use the increasing amount of data, a good and smart analysis of the data is necessary. Big Data analysis is the process of researching Big Data – to discover hidden patterns, unknown correlations, market trends, customer preferences, and other useful information – to make more informed decisions.
The advantage of data analysis is that decisions can be based on knowledge gained from facts and thus become less dependent on intuition and subjective experiences. With this knowledge, costs can be reduced, processes can be streamlined and the quality of products and services can be increased. By combining data intelligently and by interpreting / translating, new insights are created that can be used for new services, applications, and markets. This information can also be combined with data from various external sources, such as weather data or demographics.
The IoT generates an unprecedented amount of Big Data, which greatly taxes the Internet infrastructure. According to an estimate, by 2020 there will be 5,200 gigabytes of data for all people on Earth. To support the billions of paired devices expected by then, we would have to deploy around 340 application servers per day (or 120,000 servers per year). Cloud computing offers a way to meet these dizzying requirements.
Cloud computing is the availability of hardware, software and data via a network on request. When you work in the Cloud you store and retrieve hardware, software and data in a different location than your own. Because this storage location is not visible and tangible, the term Cloud is used. Everything is stored on a server that is unknown to you. The Cloud stands for a network that with all the computers that are connected to it forms a kind of cloud, where the end user does not know on how much and on which computer the software runs, or where those computers are located exactly. The user has his own scalable, virtual infrastructure that is scalable. Without the possibility of scaling, an online service does not relate to cloud computing.
IoT devices sometimes run on their own embedded software or firmware, but they can also use the Cloud to process data. The data that is sent is stored and processed within the Cloudserver, ie in a data center using data analysis. As soon as the data reaches the Cloud, the software processes the data. This can be very simple, such as checking whether the temperature value is within an acceptable range. Or complex, such as the use of computer vision on video to identify objects (such as intruders in your home).
We are seeing explosive growth in the volume, speed, and variety of Big Data collected by the IoT. But how do you come up with ways to analyze large amounts of data and thereby unlock information? This problem is also called the “Big Data problem”: the collection of complex data sets that are so large that it becomes difficult to analyze and interpret them manually or using current applications. Big Data analyses require new processing forms for large data sets.
Big Data, the Cloud, and the Internet of Things are all parts of a continuum. Cloud computing is the structure that supports Big Data projects. You can not think about the IoT without thinking about the Cloud, and it’s hard to think about the Cloud without thinking about the analysis of the stored Big Data. Because the faster you analyze data, the faster you get results and the greater the predictive value of data. After all, the real value of Big Data lies in the insights gained through analysis – discovered patterns, derived meanings, indicators for decisions and ultimately the ability to respond to the world with greater intelligence.
Big Data analysis consists of a series of advanced technologies designed to work with large amounts of heterogeneous data. To reap the full benefits of IoT data, we need to improve the speed and accuracy of Big Data analysis. This involves the use of advanced quantitative methods such as Artificial Intelligence (AI), including machine learning, to explore the data, and to discover connections and patterns. In order to identify potential problems, the data must be analyzed in terms of what is normal and what is not. Agreements, correlations and deviations must be quickly identified on the basis of real-time data streams. In an IoT situation, AI can help bring the billions of data points down to what really makes sense. It is impossible to assess and understand all Big Data with traditional methods. It just takes too much time.
It is generally accepted that IoT and AI are very important to each other’s future. AI will make IoT viable on a scale, and through IoT the lives of most people will be influenced daily by AI. The potential for highly individualized services is endless and will drastically change the way people live.
When it comes to IoT, it is often necessary to identify correlations between input from dozens of sensors and external factors that quickly generate millions of data points. Machine learning starts with the outcome variables (e.g. energy saving) and then automatically searches for prediction variables and their interactions. Machine learning is valuable if you know what you want, but you just don’t know the important input variables to make that decision. So you give the algorithm the goal and then let it “learn” which factors are important to reach that goal.
In addition, machine learning is also valuable for accurately predicting future events. Algorithms are continuously improved as more data is captured and assimilated. This means that the algorithm can make predictions and can see what actually happens, which can be compared to make adjustments to become more accurate. The predictive analyzes made possible by machine learning are extremely valuable for many IoT applications. By collecting data from multiple sensors, algorithms can learn what is typical and then detect when something abnormal begins to happen.
In essence, IoT involves sensors that are embedded in all kinds of devices and send data streams via Internet connections to one or more central (Cloud) locations. That data can then be analyzed. These results are used to improve the life of the user. All IoT devices follow these five basic steps: measuring, sending, storing, analyzing, acting. What makes an IoT application worth buying (or making) is value in the last step of that chain, “acting”. Acting can mean an infinite number of things, ranging from a physical action to providing information. Regardless of how acting looks, its value depends entirely on the “analysis”. And AI (or rather machine learning) plays a crucial role in this analysis. With machine learning, patterns can be detected in the data. When machine learning is applied to the “analyze” step, this can dramatically change what is (or is not) done in the subsequent “acting” step.
We need to improve the speed and accuracy of Big Data analysis to ensure that IoT fulfills its promise. All data in the world is completely useless if we can not use it. The only way to analyze this data generated by the IoT is with machine learning. With machine learning, patterns, correlations and anomalies can be found, from which lessons can be learned, so that ultimately, for example, better decisions can be made. The potential of Big Data can only really be realized when it is combined with AI.
The combination of IoT solutions with AI enables real-time reactions, for example via a remote video camera that reads license plates or analyzes faces. In addition, AI processes data afterwards, such as searching for patterns in data and performing predictive analyzes. AI makes the huge amounts of data from IoT devices valuable, while IoT is the best source for the real-time data that AI needs to develop. Devices transform from “smart”, i.e. connected to the Internet with a corresponding mobile app, to “intelligent”, which is characterized by the ability for devices to learn from their interactions with users and other devices, as well as the interactions with all other devices in the network. Artificial Intelligence really helps IoT devices become intelligent.
What does the future look like for Big Data Intelligence; the convergence of Big Data and AI? What seemingly impossible challenges can we tackle? Better jobs, a more sustainable environment, smarter economy, a safer world, a cure for cancer?
Thanks to Big Data, data scientists gain unhindered access to – and work with – huge data sets. Instead of relying on representative data examples, data scientists can now rely on the data itself. This is why many organizations have switched from a hypothesized approach to a “data first” approach. We can let data determine the direction and tell the story. Big Data enables an environment that stimulates the discovery through iteration. As a result, we can learn faster.
The information from IoT devices must somehow be made useful for the end-user. Smart objects must be able to communicate with people. We usually have access to the results on our mobile devices or computers via apps or browsers. Information is displayed in the form of graphs or diagrams in a user-friendly interface. The user can then perform an action and influence the system. The adjustments or actions that the user makes are then sent via the system: from the user interface to the Cloud and back to the sensors/devices to make the changes.
However, some actions are performed automatically. Instead of waiting for you to adjust the temperature, the system can do this automatically via predefined rules. And instead of calling you to warn you of an intruder, the IoT system can also automatically inform the relevant authorities. They can take measurements of the environment and use the data to change their own settings and to signal other devices to do this. Many of the objects perform actions based on algorithms, which take place either within their own processors or on Cloud servers.
IoT has various gradations: 1) you can have an object measure and, based on that, have an unambiguous action to it, 2) you can also have an object interpret certain information and have it act on it, and 3) you can let an object understand data and set new goals independently. Things and products become “intelligent” by adding computing power and are thus able to make decisions themselves. They can therefore exchange information at any time and initiate physical actions. This means that soon there will be little or no human interaction with these devices. Especially when there are more devices that can work together with other devices, we can automate many everyday tasks.
In the end, it is not about the product itself, but about the digital, data, and information-driven added value that arises. Through IoT, we get much more insight in and influence on situations. Insight and influence arise because there is continuous meaningful information that can be converted into (automatic) action.
