Efficient architectures for soft-output algorithms
K.M. Chugg, Xiaopeng Chen · 2002
Algorithms which compute either the a posteriori probability (APP) or the minimum sequence metric (MSM) of an input symbol to a finite state machine based on some portion of the observation sequence are developed. Full-record (i.e., type-I) and fixed-delay algorithms are both considered and related to existing algorithms. In particular, the fixed-delay APP and MSM algorithms developed are shown to be significantly less complex than the equivalent algorithms of Li, Vucetic, and Sato (1995) i.e. the optimal soft-output algorithm (OSA) and the simplified version (SSA), respectively.