Abstract
This paper presents a lattice detection strategy for spatial multiplexing (SM) which takes advantage of a pre-processing stage based on the geometric relations between the points in the primal lattice and the ones in the dual lattice. The first part of the paper clarifies this geometric relationship that will be exploited later on in the design of a pre-processing stage for the proposed receiver. This pre-processing finds a set of successive minima in the dual lattice, and is only required at each channel update. The subsequent symbol detection algorithm exclusively involves a linear transformation (the pseudo-inverse) in order to generate a list of candidate solutions for the underlying closest vector problem (CVP). The receiver outperforms ordered successive interference cancellation and, in the low signal-to-noise ratio (SNR) regime, also outperforms lattice-reduction-aided receivers at the expense of a "true" sphere-decoder that runs only once per channel update, and not for each received vector.