Summary: | 碩士 === 國立中正大學 === 通訊工程研究所 === 100 === Vector perturbation (VP) precoding, the precoding technique which perturbs the signal
by some perturbation vector before transmission, can effectively improve the performance
of a communications system. When combined with the use of multiple-input-multipleoutput
(MIMO) technique, VP has been shown to achieve the full diversity gain. In this
thesis, we propose several practical approaches to improve this technique.
In the first part of the thesis, we introduce the lattice reduction (LR) algorithm to the
recently proposed VP transceiver design with iterative power loading. Unlike the original
algorithm which uses a sphere-decoder to search for the optimal perturbation vector, we
propose to solve it via a number of LR-aided precoding schemes. Simulation results show
that the proposed algorithms can significantly reduce the computational complexity at the
expense of only marginal performance loss.
In the second part of the thesis, we investigate ways of improving the vector perturbation
techniques by using a candidate list. Two kinds of algorithms have been proposed.
The first kind is based on the forward and backward LLL, which has been shown to
provide improved error rate performance. The second kind reduces the computational
complexity by exploiting the structure of the unimodular transformation matrix. The algorithms
have been evaluated and compared via extensive computer simulations.
|