Abstract—Waterfilling solutions provide optimal powerdistribution in multiple-input multiple-output (MIMO)system design. However, the optimal distribution is usuallyobtained through costly computational processes, such asthe determination of the system eigenvalues. Forcommunication channels in a fast paced environment, thecosts are even higher due to the necessity of trackingchannel changes. In addition, the computational costsincrease with the number of inputs and outputs, i.e. the sizeof the MIMO channel matrix. A solution for reducing thecomputational burden is to utilize pre-determinedwaterfilling based on the channel’s statistics. No updates arerequired unless the channel statistical characteristicschange. This work studies waterfilling estimations based onrandom matrix theory. The results can be applied when thechannel coefficients follow a Rayleigh distribution and thenoise is additive, white, and Gaussian.

Index Terms— MIMO systems, random matrix, eigenvalues,waterfilling

Cite: Victor M. Vergara, Silvio E. Barbin, and Ramiro Jordan, "Waterfilling Estimation for AWGN MIMO Channel Modeled as a Random Matrix," Journal of Communications, vol. 3, no.5, pp.12-22, 2008.