A sensor network is a network connecting multiple sensors that are measuring a given environment (Image 1). Through technological innovations, sensors have become smaller and operate with very little power. In addition, there are now smart sensors that go beyond just measuring the physical quantity of things. These smart sensors which can process simple information and communicate are currently being utilized.
Now that sensors are cheap and mass produced, we are able to measure temperature, power consumption and a variety of other things all over the place. The subsequent questions would then be how do we efficiently collect the data from the sensors that are placed all over the place, and how do we make use of this data for observations and monitoring? An example would be battery operated CO2 sensors placed in a forest, which at times need to conserve battery energy and still communicate (Image 2).
Now that we are able to place great quantities of sensors and can collect a massive amount of data, we also need to select reliable sensors, too. There are some sensors that after a long period of operation could start to have physical failures and record abnormal values as measured values. It would be necessary to remove these sensors from the sensor network and not include them in the dataset.
Case Studies with Sensor Networks
There are many areas in which research is being conducted in relation to sensor networks. One such example is the monitoring system of the elderly. In this case, multiple sensors are placed in the home of the elderly and based on the data being collected it will monitor life activities to, for example, detect if the elderly might have fallen ill or may have collapsed from illness (Image 3).
Another example is the “EXPO AMEDAS” from the World Exposition in Aichi, which has a sensor network designed to monitor the earth environment. It was a system that collected data on temperature and the concentration level of CO2 from around the EXPO site, and allowed that data to be visible in real time through the internet. There are global level issues such as global warming that people are discussing and trying to deal with; valuable data collected from such sensor networks should be able to contribute greatly to these discussions.
Research on Sensor Networks
In our lab, we conduct research on the fundamental technology to construct such sensor networks and its applications. For fundamental research, we are studying how to detect and diagnose malfunctioning sensors within a network (Image 4). For applied research, besides the elderly home monitoring systems as mentioned above, we are also researching space weather forecasts to report on the space environment using data from satellites as well as collected data from ground-based observations (Image 5).
Image 4: Sensor abnormality diagnosis network (dynamic relation network). The plus sign indicates the sensor is normal, the minus sign indicates a sensor diagnosed as abnormal.