Abstract
As a typical instance of the metasurface, intelligent omni-surfaces (IOSs) emerge as a potential technique for coverage extension benefiting from the symmetric reflection and refraction. For large-scale IOSs, the enlarged near-field region allows users to be randomly distributed in both the near and far fields of IOSs. To perform beamforming in such a near-far (NF) field communication system, in this paper, we propose an NF-field codebook design and beam training scheme, which avoids the high complexity of accurate channel state information (CSI) acquisition. We reveal that the traditional Rayleigh distance based NF-field boundary may cause redundant training overhead. Thus, an effective NF-field boundary is introduced in terms of the beamforming gain. We then utilize this NF-field boundary and the symmetric characteristic of the IOS reflective-refractive signals to design an IOS-tailored codebook consisting of multiple codewords covering both the near and far fields. On this basis, a joint reflective-refractive beam training mechanism for IOS-aided systems is presented, where beam training is simultaneously performed in the symmetric regions of the IOS, thereby reducing training overhead. Simulation results show that the proposed scheme achieves a higher sum rate than the traditional codebooks given the same number of codewords, and performs close to the perfect CSI case.
Funding Information
  • National Science Foundation (62322101)
  • National Science Foundation (62271012)
  • National Science Foundation (62227809)