The Coded Slotted Aloha (CSA) scheme has been
proposed for the uncoordinated, non orthogonal multiple access.
CSA achieves an outstanding performance by exploiting segmentation,
code diversity, and interference cancellation. In the
literature, the CSA performance has been evaluated assuming the
conventional erasure channel. Purpose of this paper is to extend
the evaluation to the more realistic capture channel. To this aim,
two approaches are considered: an ideal full capture one, in which
the position of the segments is known by the receiver, and a mixed
approach, in which the reception of a single uncollided segment
is required. A code redesign is also proposed for the situation in
which the nodes transmit with a constant power, while, the more
general scenario characterized by fading and power diversity is
investigated by extensive Monte Carlo evaluations.
