In this correspondence, the detection of trellis-coded modulated signals co
rrupted by intersymbol interference, co-channel interference, and nonlinear
impairments is treated as a classification task by means of a clustering b
ased sequence equalizer decoder. The receiver performs jointly decoding and
equalization of trellis-encoded signals. No specific model is required for
the channel or for the interference and the noise, and no code knowledge i
s needed at the receiver. Complexity reduction of the equalizer is obtained
through two suboptimal techniques: a) clusters' grouping and b) the M algo
rithm. The robust performance of the proposed scheme is illustrated by simu
lations.