Scalable Channel Estimation in FDD Massive MIMO
Doctor of Philosophy
Massive MIMO brings in key benefits that make it a key design in the next-generation wireless network. To fulfill the potential benefits, channel state information is essential to realize effective user selection and beamforming. In this thesis, we design and analyze scalable channel estimation schemes for FDD massive MIMO systems. First, to make downlink channel estimation scalable with the number of base-station antennas, one of the key ideas is to exploit the inherent sparsity of wireless channels, driven by two main assumptions: (i) the cardinality of channel models in propagation domain is much smaller than the expected base-station array sizes (64+ antennas), and (ii) uplink and downlink channels share the same spatial space. However, based on our channel measurement data, we find that the two assumptions may not always hold and hence FDD channel estimation schemes with the above assumptions may not result in maximal achievable performance. To understand the performance gap, we analyze the modeling mismatch regarding the above two assumptions to quantify the modeling error of approximating downlink channel with uplink dominant angles in the propagation domain. We derive modeling error convergence with growing base-station array size and provide both numerical and experimental results. We observe that modeling error increases with the number of base-station antennas before converging to a value that is independent of the base-station array size, and more distributed channel power leads to larger modeling error. Utilizing the modeling error, we then investigate the resulted beamforming performance rate loss. Accordingly, from both numerical and experimental results, we observe that rate loss increases with the number of base-station antennas before converging to a value that is independent of the base-station array size, and more distributed channel power results in higher rate loss. Second, to make downlink channel estimation scalable with the number of users, we propose a novel propagation domain-based user selection scheme, labeled as zero-measurement selection, for FDD massive MIMO systems. The key idea of approximate selection is to infer downlink user channel norm and inter-user channel orthogonality from uplink channel in propagation domain, which is proven effective with both experimental and numerical results. In zero-measurement selection, the base-station performs downlink user selection before any downlink channel estimation. As a result, the downlink channel estimation overhead for both user selection and beamforming will be independent of the total number of users. Then we evaluate zero-measurement selection with both measured and simulated channels. The results show that zero-measurement selection achieves up to 92.5% weighted sum rate of genie-aid user selection on average and scales well with both the number of base-station antennas and the number of users. We also employ simulated channels for further performance validation and the numerical results yield similar observations as experimental findings.
FDD; Massive MIMO; Channel Estimation; User Selection