With the successful deployment of fifth-generation (5G) systems, both industry and academia anticipate the evolution of 6G systems to meet future data-intensive needs. 6G aims to enhance coverage, capacity, and spectral efficiency, with extremely large-input multiple-output (XL-MIMO) base stations (BS) being central to this vision. In addition, the mid-band spectrum (7-15 GHz) becomes a strategic choice, providing sufficient bandwidth and reasonable path loss characteristics. However, the implementation of XL-MIMO in the mid-band will bring fundamental changes to the electromagnetic signal propagation characteristics, especially due to the increase in Fraunhofer distance. Therefore, it is necessary to model the waves as spherical for near-field scenarios. Furthermore, the efficient operation of communication systems depends on accurate knowledge of channel state information (CSI). However, existing CSI feedback schemes (Type-1 and Type-2) were originally designed based on far-field assumptions and exhibit obvious limitations when applied to XL-MIMO systems operating within the mid-band spectrum. To address this issue, this paper proposes two innovative CSI feedback frameworks: Type-3 for single-user MIMO (SU-MIMO) scenarios and Type-4 for multi-user MIMO (MU-MIMO), providing comparable uplink overhead compared to Type-1 and Type-2 respectively. Furthermore, the performance of the proposed Type-3 and Type-4 feedback schemes in terms of spectral efficiency and overhead is also evaluated and compared with existing Type-1 and Type-2 feedback schemes. Results demonstrate the superiority of the proposed CSI feedback schemes over existing feedback schemes, remaining resilient to performance degradation with an increase in the oversampling factor while maintaining similar uplink overhead. Notably, Type-4 demonstrates significant performance improvements compared to existing Type-2 feedback, and the gain amplifies as the number of UEs increases in MU-MIMO scenarios.

https://ieeexplore.ieee.org/abstract/document/10565892