Adapting a Downlink Array from Uplink Measurements
Bertrand M. Hochwald
Thomas L. Marzetta
Abstract:
In a frequency division duplex system, a transmitter adaptive
antenna array can potentially improve the performance of a wireless
downlink. But because the uplink and downlink channels have different
wavelengths, and therefore different responses, direct downlink adaptation
based on channel estimates of the uplink is generally not feasible.
Instead, there has been some recent interest in adaptations that require
only the second-order statistics of the uplink and downlink to be similar.
These algorithms derive adaptive weights from the covariance of the
received signal to apply to a downlink transmitter array. We make two
contributions to this area. First, we introduce an array configuration
employing M+1 elements with log-periodic spacing, that comprises two
overlapping subarrays, each with M elements, that are scaled versions of
each other, with the scale factor equal to the ratio of the uplink
wavelength to the downlink wavelength. This array has identical
beampatterns at the two wavelengths, thus helping to fulfill the
requirement that the uplink and downlink second-order statistics be the
same. Second, we demonstrate that obtaining a good estimate of the uplink
covariance matrix is not essential for the successful operation of the
adaptive scheme: even when the mobile is at rest and the uplink information
comprises only a single snapshot from the receiver array, an adaptive
scheme can improve the signal-to-noise ratio.
Status:
Appears in IEEE Transactions on Signal Processing, vol. 49, pp. 642-653,
Mar. 2001.
Files:
Compressed PostScript (.ps.gz) (84K),
PostScript (.ps) (366K),
PDF v3.0 (.pdf) (180K).
Bert Hochwald <hochwald@bell-labs.com>