Sensor calibration using density estimation-based error model

Abstract

We propose a sensor calibration approach that is based on constructing statistical error models that capture the characteristics of the measurement errors. The approach is generic in the sense that it can be utilized in any arbitrary sensor modalities. The error models can be constructed either off-line or on-line and is derived using the nonparametric kernel density estimation techniques. Models constructed using various forms of the kernel smoothing functions are compared and contrasted using statistical evaluation methods. Based on the selected error model, we propose four alternatives to make the transition from the error model to the calibration model, which is represented by piece-wise polynomials. In addition, statistical validation and evaluation methods such as resubstitution, is used in order to establish the interval of confidence for both the error model and the calibration model. Traces of the acoustic signal-based distance measurements recorded by in-field deployed sensors are used as our demonstrative example. Furthermore, we discuss the broad range of applications of the error models and provide a tangible example on how adopting statistical error model as the optimization objective impacts the accuracy of the location discovery task for wireless ad-hoc sensor networks.