An Extended Jointly Gaussian Approach for Iterative Equalization

A novel equalization scheme for signals transmitted over multipath Multiple-Input Multiple-Output (MIMO) channels, well-suited for iterative processing, is proposed in this work. This method, dubbed Extended Jointly Gaussian Approach (extended JGA), provides an interesting trade-off between complexity and performance for equalizers based on the JGA. It works by first performing a marginalization over a set of interfering terms, and then using a jointly Gaussian assumption on the remaining interference. It is shown that, with this extension, performance can be greatly improved for some scenarios at the expense of a manageable increase in computational complexity. In order to reduce the computational burden of the detection process, complexity saving techniques are discussed. For Single-Carrier Frequency-Division Multiple Access (SC-FDMA) schemes, the computational burden of the equalization process can be further reduced by using frequency-domain versions of the classical JGA, or the extended JGA proposed in this work. The potential of the extended method is assessed for non-iterative schemes via analysis of Signal to Interference-plus-Noise Ratios (SINRs) at the output of the equalizer. This figure of merit shows that a significant increase in througput can be obtained by removing some terms from the interference pool, specially for MIMO channels. For iterative equalization, the convergence behaviour of systems applying equalizers based on both the classical and the extended JGA is analyzed by means of EXIT charts. Simulink models of uplink Long Term Evolution (LTE) communication systems, applying both classical and extended JGA equalizers, are used to produce Monte Carlo simulations. These simulations are used to confirm the performance gains indicated by SINR analysis and EXIT charts for realistic MIMO scenarios.