Low complexity polynomial expansion multiuser detector with strong near-far resistance

Abstract

The polynomial expansion (PE) multiuser detector (MUD) can be seen as a general iterative approach to approximate other MUDs. A PE MUD is presented which converges rapidly to the linear decorrelating MUD while having a low computational complexity (no matrix inversion is needed). This PE MUD iteratively approximates the matrix inversion, otherwise necessary for the decorrelating MUD. Furthermore, the PE MUD is analyzed in a near-far scenario. Many iterative approaches, show severe degradation in a near-far scenario. In the case of the PE MUD, the near-far degradation is due to the fact that the matrix elements, which correspond to each user, all include an absolute estimation error of similar size. Therefore, the matrix elements which correspond to the weak users experience a much higher relative error. It is shown how the PE MUD can be enhanced to eliminate this effect and therefore ensure a stronger near-far resistance.