High-throughput implementation of tree-search algorithms for vector precoding

摘要

This contribution analyzes the architecture design and FPGA implementation of high-throughput multiuser vector precoders. The most complex task of such precoders, i.e. the search of the perturbing vector over an infinite lattice, has been carried out by means of two tree-search algorithms: a novel fully-pipelined complex-valued K-Best scheme and a fixed-sphere encoder (FSE). Both designs have been implemented in a Virtex-6 device, leading to precoding throughputs of up to 2.4 Gbps with 16-QAM modulation.