Channel estimate-based performance prediction for coherent linear modulation on slowly fading channels

This paper describes a framework for predicting the error rate performance of linearly modulated coherent systems operating over frequency selective, slowly fading channels, when the statistics of the bandlimited two-sided discrete FIR model for the channel are available from a channel estimator. This work supports an upcoming experiment in which channel estimates obtained with one modulation scheme will be used to predict the performance of higher order modulation schemes. The focus of this paper is the mathematical framework for the theoretical performance prediction. Linear memoryless modulation is assumed, providing a a very general signal model that includes M-PSK and M-QAM. The channel model is slowly fading, with correlated complex Gaussian taps, incorporating both the flat and frequency selective cases. Expressions for the error probability are presented, and the cases for which closed form or acceptable numerical solutions exist are delineated. A new recursive method for solving the Rician-fading main path case is presented. The high-data-rate maritime line-of-sight channel is used to demonstrate the use fullness of the method.