Chapter Summary

Chapter Summary

Key Points

  • 1.

    The cascaded channel model. For a BS with NtN_t antennas, an RIS with NN elements, and a UE with NrN_r antennas, the effective end-to-end channel is Heff=Hd+H2Ξ¦H1\mathbf{H}_{\text{eff}} = \mathbf{H}_d + \mathbf{H}_2 \boldsymbol{\Phi} \mathbf{H}_1, a NrΓ—NtN_r \times N_t matrix linear in the RIS phase vector Ο•\boldsymbol{\phi}. The Khatri-Rao identity vec(H2Ξ¦H1)=(H1TβŠ™H2)Ο•\text{vec}(\mathbf{H}_2 \boldsymbol{\Phi} \mathbf{H}_1) = \big(\mathbf{H}_1^T \odot \mathbf{H}_2\big) \boldsymbol{\phi} is the workhorse that every optimization algorithm exploits.

  • 2.

    Rank and keyhole. The rank of H2Ξ¦H1\mathbf{H}_2 \boldsymbol{\Phi} \mathbf{H}_1 is bounded by min⁑(rank(H1),N,rank(H2))\min(\text{rank}(\mathbf{H}_1), N, \text{rank}(\mathbf{H}_2)). Pure LoS on both hops gives a rank-1 cascade β€” the RIS keyhole, which eliminates multi-stream multiplexing regardless of BS or UE antenna count. Rich scattering on either (or both) sides is required for multi-stream operation.

  • 3.

    LoS RIS phases are a spatial sawtooth. Under LoS on both hops, the optimal RIS phase is a linear (for planar arrays, bilinear) function of element index: ΞΈn1,n2⋆=Ο€[n1Ξ”u+n2Ξ”v]\theta_{n_1, n_2}^\star = \pi[n_1 \Delta u + n_2 \Delta v], where Ξ”u,Ξ”v\Delta u, \Delta v are determined by the BS and UE directions. This is the phased-array "generalized reflection" pattern.

  • 4.

    Cascaded Rayleigh has fatter tails. For small NN, cascaded Rayleigh fading yields higher outage probability than direct Rayleigh of the same mean SNR. At Nβ‰₯256N \geq 256, the CLT approximates the cascaded distribution back to Gaussian, and the "keyhole penalty" in outage vanishes. Another reason to design with hundreds of elements.

  • 5.

    Near-field at sub-THz is the rule, not the exception. The Fraunhofer distance dF=2D2/Ξ»d_F = 2D^2/\lambda can exceed a kilometer for 1-m apertures at 140 GHz. Inside dFd_F, wavefront curvature across the aperture enables distance-specific focusing and adds extra degrees of freedom absent in the far-field. Near-field modeling is essential for sub-THz RIS design; it is the technical foundation of the array-fed RIS architecture (Chapter 11).

Looking Ahead

We now have the full signal model in hand. Chapter 4 attacks the first practical question the model raises: how does the system estimate H1\mathbf{H}_1 and H2\mathbf{H}_2 in the first place, given that the RIS is passive and cannot transmit pilots? The answer β€” ON/OFF switching patterns, DFT codebooks, and compressed sensing β€” sets up the optimization chapters 5–8, which finally exploit the cascaded model to design Ξ¦\boldsymbol{\Phi} and v\mathbf{v} jointly.