Analysis and handling of biased data provided by wireless sensor networks with lack of energy

Abstract

Wireless sensor networks consist of several battery-powered microelectronic devices whose purpose is to monitor the environment in which they are deployed. When a device's battery is almost completely discharged, it can stop working or start producing distorted or biased data, making monitoring unfeasible. Thus, many works seek to control the energy level of the sensor node to avoid the described behavior. However, battery-powered wireless networks inevitably run out of energy at some point in time. In this way, the objective of this study is to propose, implement, and compare methods that treat the data which are biased due to the lack of energy of the network devices. This method consists of joining estimating mechanisms like Kalman filter and GM (1, 1) with another technique to measure the reliability of the network devices based on the available energy in the nodes' batteries.