Variation prediction-based energy saving scheme in status monitoring WSNs

Monitoring status with wireless sensor networks (WSNs) usually needs to perceive and transmit status values periodically. The periodical sampling will continuously consume sensor nodes´ energy, which is rather limited in WSNs. Meanwhile, most monitored status values change slowly, continuously and self-interrelated in time, result in that a portion of periodical samples seem to be redundant. This paper proposes a novel energy saving scheme based on variation prediction for status monitoring WSNs, which substantially turns periodical sampling into virtual event triggered sampling based on the status variation prediction. Specifically, according to the past status changing characteristics, our scheme can predict the time point when the status variation will be larger than a predefined threshold, and trigger a perception and transmission at the predicted time point. Based on such scheme, the sensor node can sleep for more time to save energy, especially in time when the monitored status value is not (or slowly) changing. The experiments deployed in LongMen Mountain show that our scheme can save considerable energy with similar accuracy in status monitoring comparing to traditional periodical scheme.