Multiple Differential Detection with Channel Prediction Employing Soft-Output Per-Survivor Processing

Abstract

In order to improve tracking performance on fast time-varying channels, this paper proposes a soft-output Viterbi algorithm (SOVA) for multiple differential detection (MDD) employing channel prediction and forward error correction. Although QPSK are generally demodulated based on LLR (log-likelihood ratio) for its soft-decisions, it is difficult for joint detection (JD), which does not explicitly estimate channel state information, to derive LLR. This paper proposes soft-output PSP (per-survivor processing), which derives metrics difference between the two paths in the Viterbi algorithm (VA) of PSP, where PSP is JD based on the VA. The proposed scheme can improve tracking performance for time-varying channels without an increase of computational complexity. Finally, computer simulation results confirm that the proposed scheme can track four times larger maximum Doppler frequency normalized by symbol rate in fading than the conventional schemes.