A recursive spline-based algorithm for sensor calibration design

An interpolation algorithm is reported in the present paper. It is applied for the calibration of a low cost analog to digital converter. The algorithm was formulated to be applicable for the sensor calibration problem in other electronic devices. It consists in the approximation of the calibration function to a piecewise linear function, where each piece is chosen to give the absolute error within each subinterval smaller than a reference value. The two main advantages of the present method are the decrease of the number of points used in the interpolation scheme and the evaluation's complexity. The algorithm starts by fitting a linear function through the two end points of the data set. Then, the absolute error maximum point is chosen. This point is used to divide the initial interval into two other intervals and two new linear functions are fitted to these intervals. The successive choosing and division processes are repeated until the absolute error in each subinterval is smaller than the absolute reference error.<<ETX>>