Noisebergs in Z Gaussian interference channels

We propose an efficient scheme to transmit information over a Z Gaussian interference channel. The scheme uses the concept of water filling to provide optimal power sharing among orthogonal dimensions. The model under investigation is an one-sided Gaussian interference channel with interference parameter a in the range (0,1), which can be recast as a degraded Gaussian interference channel. In the proposed solution, the notion of noisebergs (noise icebergs) arises, where noise power floats above signal power in a water filling representation of the problem, providing an improved allocation of power and degrees of freedom. The solution is best characterized by a graphical representation in the frequency domain.