AUTOREGRESSIVE MODELING OF WIDE-BAND INDOOR RADIO PROPAGATION

Based on frequency domain measurements in the 0.9-1.1 GHz band, an autoregressive (AR) model for the frequency response of the indoor radio channel is introduced. It is shown that a second order process is sufficient to represent the important statistical characteristics of the channel both in the frequency domain and the time domain where each pole identifies the arrival of a cluster of paths. A comparison is made between the statistical characteristics of the empirical data and of the channel responses regenerated from the second-order AR processes. Four methods to regenerate the indoor radio channel responses from a second order AR model are proposed. The accuracy of the methods are examined by comparing the cumulative distribution functions (CDF) of the rms delay spread and the 3 dB width of the frequency correlation function with that of the measurements performed in global, local, and mixed indoor radio propagation experiments. In a global experiment, the measurements are taken at spatially distributed locations throughout the test area. In a local experiment, the measurements are taken in or around a fixed location. A mixed experiment is a combination of global and fixed measurements.