Two-Dimensional Evolutionary Spectrum Approach to Nonstationary Fading Channel Modeling

Broadband mobile communication systems experience fading over a wide frequency band, which induces severe intersymbol interference and drastic channel gain fluctuation, resulting in a time-frequency doubly selective fading channel gain. Most existing works on fading channel modeling assume wide-sense stationarity with respect to (w.r.t.) time and uncorrelated scattering w.r.t. delay. However, due to the time-varying movement of mobile terminals, fading is usually nonstationary w.r.t. time, i.e., the Doppler spectrum is nonstationary. In addition, electromagnetic signals reflected/scattered from the same clustter/object are correlated and may induce nonstationarity w.r.t. delay due to time-varying cluster environments, e.g., wind causes the movement of leaves, resulting in time-varying scattering of leaves. To model nonstationary broadband mobile fading channels, this paper introduces a two-dimensional evolutionary spectrum (2-D ES) approach, which is compatible with the power spectral density of a stationary process. Based on the 2-D ES, we also propose an estimation method to estimate the 2-D ES parameters from a trace (sample path) of a fading process. Furthermore, we show how to set up a simulator to generate a 2-D stationary or nonstationary fading process according to our developed 2-D ES framework. Using the 2-D ES approach, we analyze a class of Nakagami-m channels which is nonstationary in both the time and frequency domains. Two examples are used to show the accuracy of our developed 2-D ES-based channel modeling approach.