A Lattice Precoding for Flat-Fading MIMO Channels Based on Eigenvalue Decomposition

A novel Lattice Precoding scheme based on Eigenvalue decomposition (EDLP) is proposed for single user multiple- input multiple-output (MIMO) communication systems transmitting over slowly-varying flat-fading channels. This technique uses linear precoding to orthogonalize the channels, thus nullifying the interference. It employs a perturbation vector as well to reduce the transmitted power. The performance is enhanced by joint optimization of the linear precoder and the perturbation vector. As a result, the EDLP achieves not only full diversity order, but also significant power gain over comparable techniques. Although this structure enables EDLP to perform a global optimization on the closest lattice point search problem, the computational complexity of this nonlinear lattice precoding is comparable to linear techniques. Furthermore, the complexity at the receiver is kept constrained. When channel estimation errors are present, the proposed EDLP technique exhibits its robustness compared to other lattice precoding techniques.