Higher-Order Tensor-Based Joint Transmit/Receive Beamforming and IRS Optimization

This paper proposes a new tensor-based algorithm for joint active and passive beamforming optimization for intelligent reconfigurable surface (IRS)-assisted systems. The proposed algorithm is based on the Truncated Higher-Order Singular Value Decomposition (T-HOSVD), and can independently compute the active (transmitter and receiver) and the passive (IRS) beamforming vectors. In addition, we show that the proposed algorithm can also control the signaling overhead in IRS-assisted communications by factorizing the IRS passive beamforming vector into multiple smaller factors. Simulations show that the proposed T-HOSVD achieves a spectral efficiency similar to the state-of-the-art solution while having a lower processing delay and feedback overhead.