Soft-Decision Symbol Estimation Using Viterbi Path Metrics

The patent introduces a Viterbi algorithm-based maximum likelihood sequence estimation as an intermediary process between the received signal and the final symbol estimation. This intermediary process uses branch metric calculation and path metric accumulation to systematically evaluate multiple possible symbol sequences, thereby improving estimation accuracy in the presence of signal distortion and noise compared to direct inverse function estimation.

The patent changes the estimation approach from direct inverse function to a multi-parameter optimization process using the Viterbi algorithm. It calculates branch metrics based on distance functions, accumulates path metrics, and selects the most likely symbol sequence by comparing multiple parameters (branch metrics, path metrics, and likelihood values), thereby achieving higher accuracy under distorted signal conditions.

The patent segments the complex estimation process into distinct modular stages: branch metric estimation process, path metric estimation process, and bit likelihood estimation process. Each stage performs a specific function (calculating transition likelihoods, accumulating path likelihoods, and determining final symbol probabilities), which organizes the complexity into manageable segments and enables systematic implementation.

The patent performs partial action by calculating metrics only for relevant candidate symbols and paths rather than exhaustively evaluating all possible combinations. The Viterbi algorithm prunes unlikely paths early in the process, performing calculations only on promising candidates, thereby reducing overall computational complexity while maintaining high estimation accuracy.