Skip to Main Content
Many mobile radio communications systems use a "frequency division duplex" (FDD) scheme to separate the reverse link (mobile-to-base) transmissions from the forward link (base-to-mobile) transmissions. FDD poses a difficult problem in the design of forward link smart antenna systems. The performance of such systems is limited by the quality of channel estimates. An often-considered compromise is to estimate the forward link vector channel as the eigenvector associated with the largest eigenvalue of the reverse link spatial covariance matrix R, averaged over many fades. The assumption is that R obtained in this manner represents primarily the path geometry of the channel, which is nominally frequency-independent, while averaging out frequency-selective fading, which is likely to be uncorrelated over the duplex separation in an FDD system. To verify that this is a reasonable assumption, it would be useful to compare direct, simultaneous measurements of the vector channel (that is, the coefficients describing the propagation between each antenna of the base station array and the single mobile antenna) at both the reverse and forward link frequencies in field conditions. This paper presents such measurements, obtained in a complex outdoor environment at 2.4 GHz. It is found that the agreement between the eigenvectors associated with the largest eigenvalue of R at the two frequencies is quite good.