17-03-2014, 05:33 PM
Abstract
We present an adaptive heuristic algorithm for data
filtering aimed to measurements of accelerations for
probes entering a planetary atmosphere. The filter is
based on properties of the median and the quantiles,
and its design takes into account the presupposed
properties of the signal, in order to achieve a higher
precision of the recovered signal. The adaptive
parameter is the filtering window width. The
adaptation is based on the statistical properties of the
signal and on heuristic rules.
Keywords: adaptive filter, heuristic procedure,
statistical filter, acceleration
1. Introduction
We present a method of adaptive filtering specifically
tailored for the accelerometric signal acquired by a
planetary probe during the initial phase of the entry in
the atmosphere of a planet. Such measurements are
performed to determine the density profile of the
atmosphere [1,2,3,4]. The probe is decelerated due to
the friction with the atmosphere. The deceleration is a
function of the atmosphere density; therefore the
deceleration signal can be used to derive the density
profile (See Annex 1).
This elegant indirect method has the drawback
that, in the upper atmosphere, the very low density
produces low friction and the deceleration is of the
order of 1μ g (g is equal to the gravity on Earth, 9.8
m/s2). In contrast, in the lower atmosphere, the
decelerations can be of the order of 10 g. The low
deceleration produces minute signals, which, for
typical accelerometers, are much below the noise level.
Today, commercial accelerometers have a noise
threshold of 200…1000μ g. Some high cost ($1000+)
commercial accelerometers have a sensitivity threshold
of 1μ g. Custom made accelerometers have a threshold
of the order of 10-8 g [5], but they are bulky, heavy and
costly. Our aim is to use a “smart” signal processing
method, such that high precision is obtained even with
commercial, medium level accelerometers. The method
can be used with higher-end accelerometers too, to
improve their accuracy by orders of magnitude,
because the filtering method is not dependent on the
type of the accelerometer.
We present an adaptive heuristic algorithm for data
filtering aimed to measurements of accelerations for
probes entering a planetary atmosphere. The filter is
based on properties of the median and the quantiles,
and its design takes into account the presupposed
properties of the signal, in order to achieve a higher
precision of the recovered signal. The adaptive
parameter is the filtering window width. The
adaptation is based on the statistical properties of the
signal and on heuristic rules.
Keywords: adaptive filter, heuristic procedure,
statistical filter, acceleration
1. Introduction
We present a method of adaptive filtering specifically
tailored for the accelerometric signal acquired by a
planetary probe during the initial phase of the entry in
the atmosphere of a planet. Such measurements are
performed to determine the density profile of the
atmosphere [1,2,3,4]. The probe is decelerated due to
the friction with the atmosphere. The deceleration is a
function of the atmosphere density; therefore the
deceleration signal can be used to derive the density
profile (See Annex 1).
This elegant indirect method has the drawback
that, in the upper atmosphere, the very low density
produces low friction and the deceleration is of the
order of 1μ g (g is equal to the gravity on Earth, 9.8
m/s2). In contrast, in the lower atmosphere, the
decelerations can be of the order of 10 g. The low
deceleration produces minute signals, which, for
typical accelerometers, are much below the noise level.
Today, commercial accelerometers have a noise
threshold of 200…1000μ g. Some high cost ($1000+)
commercial accelerometers have a sensitivity threshold
of 1μ g. Custom made accelerometers have a threshold
of the order of 10-8 g [5], but they are bulky, heavy and
costly. Our aim is to use a “smart” signal processing
method, such that high precision is obtained even with
commercial, medium level accelerometers. The method
can be used with higher-end accelerometers too, to
improve their accuracy by orders of magnitude,
because the filtering method is not dependent on the
type of the accelerometer.