An engineer at the University of Connecticut is developing a more efficient and adaptive system that could be used by the U.S. Navy to track enemy submarines and ships, and other targets of interest.

The Navy deploys thousands of underwater acoustic sensor networks to monitor oceanographic conditions or track enemy ships, for example. But these undersea sensors are always on, always tracking, even when there's no target, said Shalabh Gupta, a UConn engineer and researcher at the National Institute for Undersea Vehicle Technology headquartered at UConn's Avery Point campus.

Gupta has a devised an algorithm that would essentially turn individual sensor nodes on and off. When there's a target of interest, the sensor nodes closest to the target would switch to high-power sensing to track the target, predict its trajectory, and alert other nearby sensors within its path so that they also switch to high power.

Sensor nodes that are farther away from the target would cycle between low-power sensing and sleep mode to reduce the amount of energy used.

Gupta likened it to lights that detect sensors and turn on when they sense movement.

Current sensor nodes tend to run on full power, meaning their batteries have a short life span, and given their location in the depths of the ocean, battery replacement can be time-consuming or nearly impossible. If sensor nodes fail, there's a coverage gap and targets could be lost, Gupta said.

Gupta is also looking into implementing a classification system to inform the sensor networks. A straight moving target is most likely an underwater drone or submarine, whereas a target that is rapidly moving back and forth could be a big fish, he said. By establishing a classification system, "you only track when it's a target of interest so that will also save energy."

Gupta has done some initial tests of ground sensor prototypes and is in discussion with Navy officials about using them for underwater sensor networks. But he is still looking for funding to build the underwater sensors.

j.bergman@theday.com