Method and device for estimating the reliability of a decoded symbol sequence

A method is presented for estimating the reliability of a data sequence after Viterbi decoding. Within the Viterbi decoding, corresponding to an initial time instant, the possible initial states for the Viterbi decoding are established (801). After proceeding to a subsequent time instant (804), at each state corresponding to the present time instant, a surviving trellis path is selected among the trellis paths coming into that state (805). The steps of proceeding and selecting are repeated until a final time instant (807), and at said final time instant a final surviving trellis path is selected to represent the decoded data sequence. At the selection step of the Viterbi decoding, a characteristic reliability metric is updated separately in association with each selected surviving trellis path (806). After selecting a final surviving trellis path, its characteristic reliability metric is converted to a sequence reliability statistic corresponding to the estimated reliability of the decoded data sequence (809, 810-814).