Channel-Aware Bayesian Model for Reliable Environmental Monitoring Sensor Networks

Abstract

We provide a novel statistical-based approach to address the reliable data reception problem for environment monitoring sensor networks. In this paper, a channel-aware Bayesian model is designed to choose the most likely quantization level as the value of the monitored phenomenon at the sink node such that the probability of a correct decision is maximized. Simple methods are presented to formulate both the temporal correlation of the monitored phenomenon and the error information of the data transfer channel between the sensor node which monitors the phenomenon and the sink node as the prior input to this Bayesian device. Evaluation based on real sensor data shows that the proposed model can monitor physical phenomena much more accurately than channel-unaware algorithms.