Seung Won Choi

Implementation of an MU-MIMO system with GPU modem for non-codebook-based TDD LTE-A

Recently, 3GPP disclosed MU-MIMO as an enhancement of Release10 for the target data rate of LTE-A. In this paper, we present a prototype implementation of 4×4 MU-MIMO system using a GPU-based modem for TDD LTE-A. As GPU handles a number of data simultaneously with the parallel processors, we have verified that the implemented MU-MIMO test-bed system can provide a real-time processing of LTE-A data with nearly 4 times larger throughput compared to Release10 SU-SISO system or 2 times compared to Release8 2×2 MU-MIMO system.

On Downlink Beamforming Methods: Capacity and Error Probability Performance

In space-division multiple access (SDMA), different beamforming or space-domain precoding techniques can be applied. We investigate two different space-domain precoding methods, the maximum capacity (MC) and the minimum mean square error (MMSE) precoders, for the downlink channel. It is shown that the MMSE precoding, which is practically implementable, can provide a reasonable performance in terms of the capacity and error probability, while the MC precoding is not practical (although it is optimum in terms of the capacity). Space-domain precoding methods are also applied to code-division multiple access (CDMA) systems.