In the past detection of dark matter was only possible via indirect
observations, but with the current technology new horizons are close
to be reached. Fundamental aspects in the current discussion are the
predictions of the dark matter phase space density and the local
velocity distribution, because of their crucial impact on the signal
prediction. The scales relevant for terrestrial instruments are of the
order of milliparsecs, too small even for the most powerful
simulation. To avoid this problem alternative approaches have to be
designed. Through a backwards-in-time evolution technique, we
calculated the velocity distribution of dark matter in the solar
neighbourhood, based on merger trees which describe the history of a
Milky Way-like halo. The method is very flexible: it allows to make
predictions, to calculate the evolution of the spatial distribution in
the Galaxy and to map out the velocity structure at any time and
position at arbitrarily-high resolution. Our current analysis shows
that although the dark matter particles are spread through all the
parameter space, structures are present even after billions of years.