Performance of Soft Decision Metrics and Diversity Combining with Imperfect Channel Estimation

In this paper, we derived two new decision metrics for the non-coherent maximum a posteriori (N-MAP) detector and the generalized-likelihood ratio (GLR) detector. The proposed detectors are optimally designed to make use of the available estimated channel. The obtained results show that the proposed detectors can outperform a system using classical training-based maximum-likelihood (TBML) detector. The performance is evaluated by means of computer simulations with both convolutional coded and LDPC-coded systems.