This thesis work has been based on an existing directional
hearing system for pilots. Its purpose has been to add distance
simulation abilities to it. To simulate different
distances, a FIR filter technique has been used.
When fighter pilots are receiving many alarm signals at the same
time,
they cannot react properly to all of them. The large amont of
information makes them feel stressed. To reduce this stress, a
directional hearing system has been developed. This system is
positioning sound sources in three dimensions and when doing so
with the
error signals, it makes it easier for the pilot to cope with with every
single one. This reduces the so called "information stress" pilots
often are experiencing.
Measurements were made at a field with large loudspeakers
and a microphone. White noise was used as input signal
to the system and the microphone recorded it while it was placed
at different distances.
Afterwards, a FIR filter model for each distance was
estimated.
One of the problems encountered with system identification
was non-stationarity. In this case, the wind
caused the transmission channel to be nonstationary
so the estimated time-invariant
filters failed to describe the system
properly. It is suggested that this type of experiment
should rather be performed indoors.
Another problem was that the loudspeaker and the microphone
introduced poles to the system.
An ARMAX-process was identified and the result became
slightly better. (If the characteristics of the loudspeaker and the
microphone had been analysed before the identifaction, that
information
could have been used to improve the final result even more.)
Using thes resulting filters to simulate distances does
provide some distance cues, but it does not give
a true subjective
sense of a distance.
