Security Sensor Networks
Sensors That Organize Themselves
Tomorrow's smart sensors will communicate with each other and provide early warning of hazards such as forest fires, floods and avalanches.
It's 2020, and the devastating bush fires that used to plague Australia are history. Drones have dropped thousands of miniature sensors over high-risk areas. Some of these sensors are equipped with a tiny GPS (Global Positioning System) module, allowing them to know their own locations, both in absolute terms and in relation to the other sensors. The sensors use minuscule antennas to communicate with one another. Integrated heat sensors measure changes in temperature, and this information is, in turn, communicated from sensor to sensor and finally transmitted via a data line to a control center in Sydney. If the control center registers a sudden spike in temperature, or if several sensors break down entirely, a fire alarm is triggered (see picture below). In other parts of the world, similar sensors issue warnings about avalanches in mountainous areas, floods in low-lying river valleys, or toxic leaks in chemical factories.
How a network of sensors detects a forest fire
A visionary scenario? Not by a long shot. Prototypes of the components of such self-organizing sensor networks are already available. "All that's missing is the evidence that the components can work together effectively," says Michael Bahr, an information and communication expert at Siemens Corporate Technology (CT) in Munich. Several pilot projects have yielded initial results. For example, in 2001 the University of California at Berkeley and the Intel Berkeley Research Lab tested a network consisting of 800 radio sensors that measured the light intensity from a stage.
Drawing their power from solar cells or batteries, the sensors in such networks are autonomous. But their key feature is the way they route their messages. The system works like this: A sensor transmits a message via its antenna and waits for another sensor to respond. When it receives an answer, it notes which sensor has responded. Because each sensor's range is only a few meters, the data must be passed on from sensor to sensor until it reaches a central fusion point.
According to Dr. Wendelin Feiten, an expert in intelligent systems at Siemens CT, the measurements registered by the sensors must be "fused" at a central point before they can be evaluated. That's because each sensor can "see" only a small segment of its environment. It transmits this information to its neighbor, which adds its own information before passing the data package to the next sensor. If the number of sensors reporting on an event is larger than necessary, the data received is often redundant or contradictory. Statistical processes are then used to create an overall picture and evaluate its credibility.
Many types of sensors could operate in such a network: simple sensors for measuring temperature, humidity or pressure, gas sensors and tiny video, infrared or 3-D cameras.
Larger sensors connected by cables are already being used to monitor production lines and to maintain secure data networks. But tomorrow's self-organizing radio sensor networks will be based on radio modules and other components that are not only small and cheap, but also power stingy, says Feiten's colleague Dr. Gerd Völksen. "If a sensor is to sell for five euros, then the individual components, such as the battery, antenna, positioning system, chip configuration and other parts have to cost far less than one euro," he says. Obviously, developments in sensor research are heading in exactly this direction. However, it will take some time before the scenario depicted by Michael Crichton in his novel Prey becomes a reality. In Crichton's story, flying cameras made up of self-organizing nanoelements communicate with one another via radio and actually evolve into something new.
Sylvia Trage