IEEE Access (Jan 2023)

Hybrid Precoding/Combining for mmWave MIMO Systems With Hybrid Array Architecture

  • Mohamed Alouzi,
  • Faisal Al-Kamali,
  • Claude D'amours,
  • Francois Chan

DOI
https://doi.org/10.1109/ACCESS.2023.3321681
Journal volume & issue
Vol. 11
pp. 110736 – 110749

Abstract

Read online

Utilizing hybrid precoding/combining for subarray (SA) architectures in millimeter-wave (mmWave) multi-input multi-output (MIMO) systems offers reduced hardware cost and power consumption when compared with that for full array (FA) architectures, although the spectral efficiency is lower. Hence, this paper introduces a new hybrid array (HA) architecture for mmWave MIMO systems, designed to achieve a balance between spectral efficiency, cost, and power consumption. Initially, the proposed HA architecture partitions the antennas into distinct subarrays. The number of these subarrays equals the count of radio frequency chains at the transmitter/receiver. These subarrays are subsequently organized into separate subsets referred to as groups. Ultimately, within each group, the antennas are connected with a corresponding group of radio frequency chains, employing a connection method similar to that of the FA architecture. Compared to the SA architecture, the proposed HA architecture provides higher spectral efficiency by exploiting spatial diversity through grouping of antennas. Furthermore, the HA architecture achieves cost and power reduction in comparison to the FA architecture by connecting a subset of antennas to each group of radio frequency chains. Two efficient iterative hybrid precoding/combining algorithms are also proposed, studied and compared for HA architecture mmWave MIMO system. During the design derivations, the proposed algorithms consider the block structure inherent in the analog precoding and combining matrices within the HA architecture. This consideration makes it possible to independently optimize hybrid precoding/combining for each group. Results show that the suggested HA hybrid design using Algorithm 2 provides better performance than using Algorithm 1 and both algorithms can achieve higher spectral efficiency than the conventional SA and FA hybrid designs while being less complex.

Keywords