Design of Wideband Mobile Radio Channel Simulators Based on Real-World Measurement Data

In this article, we analyze different ways to estimate the temporal autocorrelation function (TACF) and the frequency correlation function (FCF) of wideband mobile radio channels from real-world channel measurements. The way the TACF and FCF are estimated depends on whether the measured channel is autocorrelation-ergodic with respect to time and/or frequency. In either case, it is shown that double averaging, i.e., averaging of measurement snapshots over time and frequency, improves the estimates of the TACF and FCF. We provide two examples of applying the iterative nonlinear least squares approximation (INLSA) algorithm to determine the parameters of measurement- based channel simulators from the estimated TACF and FCF. The presented examples correspond to real-world outdoor and indoor propagation environments. The distinctive feature of the designed channel simulators is that their temporal and frequency correlation characteristics are sufficiently close to those that have been estimated from the measurement data. For comparison reasons, we also show the corresponding results obtained by applying the well-known subspace-alternating generalized expectation-maximization (SAGE) algorithm.