While the main interest in digital apparatus is focused on
increasing the performance in terms of output data rates
and resolution in the measurement of the event energy, the
digital methods lend themselves also to accomplish other
measurement tasks, such as the determination of the arrival
time of events. In this latter case, severa1 methods based on
least squares routines have been demonstrated [1]. But, the
high potential of digital least squares methods in high-resolution nuclear spectroscopy cannot be fully exploited in
practical due to its heavy calculation burden which makes
difficult a real-time implementation. These methods have been
widely deployed on temporal computing architectures,
especially on PC. In embedded systems, where for obvious
reasons of power dissipation, consumption, size, it is not
possible to have the computing power of a PC processor, least
squares routines can be run in real-time only on decimated
data sets or data at low bit-rate. Least squares problems fall
into two categories, linear and non-linear. A regression model
is a linear one when the model comprises a linear combination
of the parameters. Non-linear least squares problems arise for
instance in non-linear regression, where parameters in a model
are sought such that the model is in good agreement with
available observations. The linear least squares problem has a
closed form solution, but the non-linear problem does not and
is usually solved by iterative refinement; at each iteration the
system is approximated by a linear one, so the core calculation