Sensors - Application and Multidisciplinary Aspects of Wireless Sensor Networks - page 13

2.2.3.2 Least Squares BFSL Linearity

The method of least squares assumes that the best-fit curve of a given type is the
curve that has the minimal sum of the deviations squared (least square error) from
a given set of data.
When defining least squares BFSL linearity, the slope of the best-fit line is defined
using the measured results in a number of calibration points with the equation:

S (actual sensor output at each data point) * (actual sensor output at each data point)

(actual sensor output at each data point)

2

S
Having mathematically determined the slope of the best-fit straight line it is then
possible to determine the maximum deviation of any point from this line.

2.2.4 Sensitivity

Sensitivity of a sensor is the ratio between a small change of the input and the resulting
change in the output signal. Mathematically speaking, it is defined as the slope of
the output characteristic curve. Sensitivity error is a departure from the ideal slope
of the characteristic curve.

2.2.5 Accuracy

Accuracy represents the largest expected error between the ideal output signal and
the actual output signal. Sometimes it is presented as a percentage of the maximum
output signal.

2.2.6 Dynamic Range

Dynamic range (or Span) is the range of the input signal that can be accu-
rately converted into the electrical output. Outside of the dynamic range sensor
produces either a predefined value, or, more commonly, undefined and
inconsistent.

2.2.7 Noise

All sensors produce noise in addition to output signal. For applications that
require high precision, sensing amount of noise introduced by a sensor can be of
at most importance.