A Fast Constrained Sphere Decoder for Ill Conditioned Communication Systems

This letter proposes a fast constrained sphere decoder for ill conditioned communications systems that exhibits less complexity than but similar performance to the generalised sphere decoder. The operational principle is based on i) the reduction of the search space by setting the hypersphere initial radius to be equal to the distance to a semidefinite program (SDP) estimate; and ii) the introduction of a heuristic pruning rule to limit the GSD spanning tree. The new algorithm achieves significant reduction in the required computational effort at the expense of a small error penalty for large dimensional systems in low signal to noise ratio (SNR) regimes.