Best Paper Award

  Vector-perturbation (VP) MIMO precoding is a major type of nonlinear MIMO precoding that can significantly improve error rate performance of linear precoding, such as zero-forcing beamforming (ZFBF). This benefit is attributed to improvement of transmit power efficiency through searching for a perturbation vector (PV) that reduces the norm of a precoded modulation symbol vector and, then, by adding such a PV to the modulation symbol vector. Link adaptation, i.e., modulation and coding scheme (MCS) selection, for throughput maximization requires an estimate of a VP gain. However, the VP gain depends on not only an instantaneous channel matrix but also coded bit sequences, modulation orders, perturbation variables, and PV search parameters. The VP gain is thus determined after the PV search for channel-coded and modulation-mapped symbol vectors. Therefore, this chicken-or-egg problem makes it difficult to apply link adaptation to VP-MIMO precoding systems.
   In this paper, an efficient iterative MCS selection procedure is devised for VP-MIMO precoding on the basis of systematic code structures possessed in channel coding schemes used in current systems; e.g., turbo codes. Starting from MCS selection based on SINR provided with conventional ZFBF precoding, the MCS index is incremented by alternately conducting modulation symbol vector mapping of information bits and PV search for the generated modulation symbol vectors. After the MCS index is updated for a specified number of times, the residual VP gain is passed on to a higher coding rate as a last update. On the basis of the selected coding rate, the parity bits are generated by turbo coding and PV is then searched for the modulation symbol vectors mapped from the parity bits. The proposed link-adaptable VP-ZFBF precoder was applied to multi-point three-dimensional (3D) beamformers deployed in a circular-layout, generalized distributed antenna system with a 28-GHz urban NLOS model. The directivity of each 3D beamformer was controlled toward a user terminal equipped with multiple antennas on the basis of the inter-subarray phase shift minimization scheme as practical analog-digital hybrid beamforming for a massive antenna array. The throughput performances of the proposed scheme were evaluated under various conditions by means of computer simulations. Simulation results show that the proposed scheme suitably passed on the VP gain to the selection of an appropriate higher-rate MCS index and thus dramatically improved throughput performances of the conventional ZFBF MIMO precoding scheme. Therefore, this paper is expected to contribute to achieving high throughputs and, thus putting VP-MIMO precoding to practical use.