Is your wireless sensor feeling okay? Is it going to live a long, full life? Considering one can't exactly cram vitamins down its throat (or in this case, cord) one can really only wonder. New technology by the electronic engineering team, A*STAR (News - Alert) Institute for Infocomm research, is looking to change that with a new adaptive classification system (ACS (News - Alert)). 

A Singapore-based organization, A*STAR (Agency for Science, Technology and Research) holds that, like any electronic device, wireless sensors often require a battery, which automatically means a limited lifespan. By halting monitoring when a machine is not operating, ACS reduces the amount of power a sensor uses, which in turn creates longevity.

The system also vitalizes machinery by decreasing network traffic along with data storage requirements.

Incorporating an adaptive management system of nested sensors, the ACS works by reducing sensor sampling rates to a minimum, and by sampling parameters at a low rate to provide data to a model whose purpose is spurring more intensive sampling of other sensors when a problem is detected. The ACS also integrates a number of methods of classifying data patterns to determine those that are problematic, or whether they need higher levels of sampling. 

As a result, ACS not only makes wireless network systems healthier/more robust, but can actually be trained to identify problems using a minimal amount of data.

Minh Nhut Nguyen, who led the research team that developed the ACS, used a machinery fault simulator to test ACS. The simulator allows key components such as bearings to be replaced by weak ones that are faulty or worn.

On average, the ACS outperformed the simulator. 

Edited by Braden Becker