Higher Rank Principal Kronecker Model for Triply Selective Fading Channels With Experimental Validation

This paper proposes a higher rank principal Kronecker model (PKM) for simulating triply selective fading channels. To construct the PKM, the channel correlation matrices are decomposed using the higher order singular value decomposition (HOSVD) method. The proposed PKM-HOSVD model improves upon the original Kronecker model by using higher rank approximation of the channel correlation matrices rather than the rank-1 approximation. The proposed PKM-HOSVD model was validated by extensive field experiments conducted for 4 × 4 multiple-input-multiple-output (MIMO) systems in both indoor and outdoor environments. The carrier frequencies used included 800 MHz, 2.2 GHz, and 5.2 GHz. The channel correlation matrices calculated from the measured channel coefficients were then decomposed via the proposed PKM-HOSVD method. The quality of the decomposition was evaluated by not only the mean square error but also the correlation matrix distance. These results indicate that many practical channels must use higher rank approximation rather than the commonly used rank-1 approximation (or the Kermoal method) to achieve satisfactory decomposition accuracy. In addition, the predicted channel capacity by the proposed channel simulation model achieves better accuracy than the original rank-1 channel simulation model.