A Low-Complexity Sphere Decoder with Noise-Predictive Radius Initialization

J.L. Lagunas Morales and S. Roy (Canada)

Keywords

Sphere decoder, underdetermined systems, MIMO, virtual MIMO, radius initialization, soft-output.

Abstract

In this paper, we present an enhancement of the sphere de coder (SD) algorithm that uses a novel radius initialization based on a noise-predictive Minimum Mean Squared Error (MMSE) filter. This technique decreases considerably the complexity of the lattice search. Our SD can to be used over underdetermined channels and it outputs soft infor mation for the bits transmitted, with practically the same level of complexity as the hard output algorithm while im proving the BER performance of the decoder. Our detec tor may be used in a broader communication system along with Error Correcting Codes (ECC) and Space-Time Codes (STC) due to its characteristic of being a universal lattice decoder. Since it doesn’t compute the soft-output in an it erative fashion, its overall complexity is not so significant within a large system, yet it can deliver a high throughput by exploiting the benefits of its MIMO architecture.

Important Links:



Go Back