A Correlation Tensor–based Model for Time Variant Frequency Selective Mimo Channels

In this contribution we present a new analytical channel model for frequency selective, time variant MIMO systems. The model is based on a correlation tensor, which allows a natural description of multi–dimensional signals. By applying the Higher Order Singular Value Decomposition (HOSVD), we gain a better insight into the multi–dimensional eigenstructure of the channel. Applications of the model include the denoising of measured channels and the possibility to generate new synthetic channels displaying a given correlation in time, frequency, and space. The proposed model possesses advantages over existing 2–dimensional eigenmode–based channel models. In contrast to them, the tensor–based model can cope with frequency and time selectivity in a natural way.