For the network-level perspective, we considered cell-free massive MIMO as the primary physical-layer technology to realize scalable device-centric communications. Cell-free massive MIMO systems conveniently combine elements from massive MIMO, small cells, and user-centric joint transmission coordinated multi-point, which can potentially eliminate the inter-cell interference and ensure uniformly good service for all the user equipments. Prior to our work, non-cooperative precoding strategies were generally preferred in the cell-free massive MIMO literature as they do not require any channel stat information exchange via backhaul signaling, although cooperative precoding design can bring considerable performance gains. Hence, we developed efficient distributed optimization algorithms and signaling schemes to implement cooperative precoding in cell-free massive MIMO systems in a scalable manner. In doing so, we considered both unicasting and multicasting scenarios, which can accommodate different types of wireless applications.
For the device-level perspective, we mainly considered fully digital massive MIMO architectures with low-resolution analog-to-digital/digital-to-analog converters (ADCs/DACs) as a means to reduce the hardware power consumption and complexity at the base station when operating at high frequencies. Hybrid analog-digital arrays do not scale well beyond the low mmWave spectrum due to the need for complex analog circuitry and resource-consuming beam management schemes. Few-bit or 1-bit fully digital massive MIMO can outperform its hybrid analog-digital counterpart in terms of beamforming flexibility and spectral/energy efficiency. 1-bit ADCs/DACs are particularly attractive as they minimize the complexity and power consumption of the data conversion, although the 1-bit quantization inevitably deteriorates the spectral efficiency. Hence, we provided new analytical studies on the channel estimation, data detection, and spectral efficiency of massive MIMO systems with 1-bit and few-bit ADCs/DACs. In addition, we considered fully digital unquantized massive MIMO and devised innovative methods to boost the spectral efficiency, i.e. by means of statistical beamforming at the user-side for highly overlapping user equipments, and the data detection performance, i.e. via low-complexity detection algorithms based on variational Bayesian inference.
For the integration of network- and device-level perspectives, to guarantee the necessary physical-layer paradigm shift with respect to the current 5G systems, the increase in the operating frequencies must go hand in hand with the introduction of flexible device-centric mechanisms. In this respect, there is a plethora of interconnected design parameters at the network and device level that need to be considered jointly in order to satisfy application-specific quality-of-service requirements. Hence, we aimed at determining a unified view of the infrastructure, spectrum, and protocols/algorithms that are necessary to accommodate specific wireless applications in future 6G systems. In addition, we analyzed multi-antenna multicasting aided by device-to-device communications to accommodate safety-critical applications under different channel state information assumptions.
