How to Make Proper Use of 1 Trillion Sensors

Feb 17, 2014
Tsuneyuki Miyake, Nikkei Electronics
Bumpei Magori, project lecturer for the University of Tokyo
Bumpei Magori, project lecturer for the University of Tokyo
[ If it clicks, the expanded picture will open ]

The number of sensors used in our society is expected to grow rapidly. Some people consider that more than one trillion sensors, which is 100 times larger than the number of sensors currently used in a year, will be used every year in the future. And there are global efforts to realize "Trillion Sensor Universe" at an early date.

Janusz Bryzek, chair of TSensors Summit LLC, who advocates Trillion Sensor Universe, expects that 45 trillion sensors will be used in 2033 (For more information, click "abstract" of his lecture on the website of Trillion Sensors Summit Japan 2014).

Even though there is a potential demand for a huge number of sensors, the diffusion of sensors will reach its limit if sensors themselves do not change, said Bumpei Magori, project lecturer, the Institute of Industrial Science, the University of Tokyo.

Magori pointed this out because he already sees problems in using various sensors in the field of architecture. He is a researcher in the field of architecture and has designed buildings for a major construction company. Based on such experience, he is afraid that human errors will occur at the time of installing sensors and make it impossible to manage a large number of sensors.

Currently, if sensors are installed in 1,000 locations in a building, residence or commercial facility, 3% of them are outputting incorrect information due to mistakes made at the time of installation, Magori said. In other words, even if high-accuracy sensors are deployed, they may not be fully used.

Intelligent sensors will gain importance

(Continue to the next page)

Intelligent sensors will gain importance

Therefore, Magori stresses that it is essential to develop intelligent sensors in an era when a much larger number of sensors are installed. In the first place, human errors occur at the time of deploying sensors because they are set up by hand.

Many sensors currently being used have a single function of detecting a target. A type and unit of data have to be set on the side of a controller that receives the data (e.g. whether sensor output is temperature or humidity). Also, wiring for controllers and sensors are done by hand, making incorrect setting and wiring inevitable. With a huge number of sensors, it is not practical to check all of them.

Intelligent sensors will help solve this problem. If sensors can transmit the kind, unit and ID of information they are outputting to a network by using a general format based on, for example, Internet protocol and they are recognized by a controller or server, it becomes possible to eliminate human errors at the time of setup and installation.

Specifically, intelligent sensors will be devices that have a sensor, signal processing circuit and communication function with TCP/IP processing capability in one package. By setting a kind, unit and ID of information in a manufacturing process, it becomes possible to eliminate mistakes in the installation process.

However, Magori considers that such sensors are not sufficient as future intelligent sensors. For example, data output from a temperature sensor has different meanings depending on installation conditions. When it is installed near a window facing south, it shows a higher room temperature than in the case when it is near a corridor on the north side.

Furthermore, the meaning of temperature data output from a sensor changes depending on whether it is near a window through which sunlight is coming, whether it is raining outside, whether it is near a heat-generating home appliance on the side of a corridor, etc. To know the correct meaning of data, information on the surrounding environment, which influences temperature data, should be taken into consideration.

Controlling various elements to save energy at stores

(Continue to the next page)

Controlling various elements to save energy at stores

Magori has been engaged in the research on the use of artificial intelligence (AI) for saving energy at convenience stores. And he developed a technology to halve power consumption by using a test store that he developed in cooperation with Lawson Inc (Shinichiro Uto (Lawson) will deliver a lecture on the details of the technology at Trillion Sensors Summit Japan 2014). Based on the results (the effects to reduce power consumption), the AI repeatedly adjusts control elements such as air-conditioning and lighting systems to improve the accuracy of a control algorithm.

Magori employed the AI because the number of control elements that influence the power consumption of convenience stores is expected to increase in the future. As control elements, he uses air conditioning facilities, lighting apparatuses and natural energies such as ventilation and daylight. It is not practical to optimize those elements by hand. And it is difficult to prepare an accurate control algorithm in advance because each store has different conditions.

It is difficult to optimally control even multiple existing air conditioners. The energy efficiency of an air conditioner usually becomes highest when it is operated at about 70% of its maximum capacity. So, it is more energy-efficient to use one air conditioner at about 70% of its maximum capacity than to use two air conditioners at about 30-40% of their maximum capacities. Magori is currently optimizing the existing system to develop a sensor system that realizes a "zero-energy convenience store."