betaFeedback

WorksEngineering

Movable Beyond-Diagonal RIS for ISAC: Optimal Placement and Pareto Boundary

Abstract

This letter investigates integrated sensing and communication using a movable beyond-diagonal reconfigurable intelligent surface (BD-RIS). Under a rank-one line-of-sight BS–RIS channel, we derive the exact communication–sensing Pareto boundary of a fully-connected BD-RIS in terms of user–target channel correlation. For line-of-sight RIS links, maximizing this correlation yields the optimal placement. We characterize the minimum track aperture that eliminates the illumination trade-off and show that a movable diagonal RIS then attains the same optimum. Below this threshold, a dynamic program offers certified placement accuracy. Numerical results examine hardware connectivity, angular mismatch, and Rician fading.

Introduction

INTEGRATED sensing and communication (ISAC) lets one waveform and one aperture serve users and probe targets [1]. Beyond-diagonal reconfigurable intelligent surfaces (BD-RISs) interconnect their elements through a reconfigurable impedance network, which generalizes the diagonal phase-shift model to symmetric unitary scattering matrices [2], [3]. BD-RIS-aided ISAC outperforms diagonal RIS in sensing SNR and throughput [4], [5], but with elements at fixed positions and through alternating optimization.

A fixed surface, however, faces a limitation that no choice of scattering matrix can remove. When the BS reaches the surface over a line-of-sight link, this link has rank one, and the BS can only scale the incident field, not shape it. The surface must then serve the user and the target from a single field. With λ/2\lambda/2 spacing, the steering vectors toward two well-separated directions are nearly orthogonal, and their correlation decays as the surface grows. In this case each function can be favored only at the expense of the other. Interconnections widen the set of reflected fields but do not change this geometry, and adding elements makes it worse. Moving the elements changes the geometry itself, which raises two questions this letter answers: where should the elements go, and how much hardware do we still need once they are placed? The second question also has a cost dimension: a fully connected (FC) BD-RIS needs M(M+1)/2M(M + 1)/2 tunable impedances, whereas a diagonal RIS needs only MM.

The element position is a second degree of freedom [6]. RISs with movable elements improve ISAC [7], [8]. Xu et al. [9] combined movement with BD interconnections and found numerically, for the multiuser sum rate, that moving and interconnecting each win in a different regime. For ISAC, it is not yet known how the BD-RIS elements should be placed, nor whether moving and interconnecting complement or replace each other.

This letter answers these questions for a movable BD-RIS that serves one user and illuminates one target. Our contributions are:

  • We show that an FC BD-RIS turns the surface into a field synthesizer, so the Pareto boundary takes the two-direction form known from MISO beamforming [10] and is governed only by the user–target correlation ρ\rho. For group-connected and diagonal RISs, the supported part of the boundary follows from an exact two-dimensional search.

  • Under LoS, placements with larger ρ\rho Pareto-dominate those with smaller ρ\rho, so maximizing ρ\rho is optimal for the FC BD-RIS. The boundary reduces to the single-function optimum iff the aperture exceeds D∗=(M−1)λ/∣Δs∣D^{*} = (M - 1)\lambda/|\Delta_s|, using a lattice that places a grating lobe on the target; there, a movable diagonal RIS is also optimal. Below D∗D^{*}, a dynamic program (DP) with a certified accuracy bound and a closed-form clustered lattice give the placement.

  • Simulations validate the analysis, compare against a fixed sparse array of the same aperture under geometry mismatch, and evaluate Rician fading with an absolute user-SNR requirement.

Notation: (⋅)∗(\cdot)^{*}, (⋅)T(\cdot)^{\mathsf{T}}, (⋅)H(\cdot)^{\mathsf{H}} denote conjugate, transpose and Hermitian transpose; vG\boldsymbol{v}_{\mathcal{G}} is the subvector of v\boldsymbol{v} indexed by G\mathcal{G}; [a]+=max⁡(a,0)[a]^{+} = \max(a,0).

C.K. Sheemar and S. Chatzinotas are with the Interdisciplinary Centre for Security, Reliability and Trust (SnT), University of Luxembourg, L-1855 Luxembourg (e-mail: [email protected]; [email protected]). S. Solanki is with the National Institute of Technology Warangal, TS 506004, India (e-mail: [email protected]).

System Model

In Figure 1, a BS with NN antennas serves a single-antenna user and senses a point target through an MM-element BD-RIS, since the direct links are blocked. The elements slide along a linear track of length D≥(M−1)d0D \ge(M-1)d_0, with positions x=[x1,…,xM]T\mathbf{x}=[x_1,\ldots,x_M]^{\mathrm{T}} in

A movable BD-RIS serving a user and illuminating a target

Figure 1. A movable BD-RIS serves a user and illuminates a target whose direct links to the BS are blocked; the target echo returns to the BS over the same surface.

XD={x:x1≥0, xm+1−xm≥d0, xM≤D},(1)\mathcal{X}_D=\{\mathbf{x}:x_1\ge0,\ x_{m+1}-x_m\ge d_0,\ x_M\le D\}, \tag*{(1)}

where d0≤λ/2d_0\le\lambda/2 is the minimum spacing. Let [e(x,θ)]m=ej2πxmsin⁡θ/λ[\mathbf{e}(\mathbf{x},\theta)]_m=e^{j2\pi x_m\sin\theta/\lambda}. The BS–RIS link is LoS, G=αˉe(x,φ)aH\mathbf{G}=\bar{\alpha}\mathbf{e}(\mathbf{x},\varphi)\mathbf{a}^{\mathrm{H}}, where a∈CN\mathbf{a}\in\mathbb{C}^{N} is the BS steering vector. The RIS–target link is ht=βte(x,θt)\mathbf{h}_t=\beta_t\mathbf{e}(\mathbf{x},\theta_t), and the RIS–user link is hu=βue(x,θu)\mathbf{h}_u=\beta_u\mathbf{e}(\mathbf{x},\theta_u) unless stated otherwise. We assume θu≠θt\theta_u\ne\theta_t; otherwise the two channels are collinear for every placement and no trade-off exists.

The scattering matrix Θ\boldsymbol{\Theta} is lossless and reciprocal, Θ=ΘT\boldsymbol{\Theta}=\boldsymbol{\Theta}^{\mathrm{T}} and ΘHΘ=I\boldsymbol{\Theta}^{\mathrm{H}}\boldsymbol{\Theta}=\mathbf{I}, and is realized by a real symmetric reactance matrix X\mathbf{X} as Θ=(jX+Z0I)−1(jX−Z0I)\boldsymbol{\Theta}=(j\mathbf{X}+Z_0\mathbf{I})^{-1}(j\mathbf{X}-Z_0\mathbf{I}) [2]. A group-connected BD-RIS splits the elements into groups G\mathcal{G} of size MgM_g, making Θ\boldsymbol{\Theta} block diagonal; Mg=MM_g=M is the FC BD-RIS and Mg=1M_g=1 the diagonal RIS. Unlike [9], where interconnected elements move as one unit, every element moves individually and the lossless network is assumed to follow it1.

The BS transmits s(t)s(t) with covariance R\mathbf{R}, under the power budget tr⁡(R)≤P\operatorname{tr}(\mathbf{R})\le P. The user receives yu=huHΘGs+nuy_u=\mathbf{h}_u^{\mathrm{H}}\boldsymbol{\Theta}\mathbf{G}s+n_u, nu∼CN(0,σu2)n_u\sim\mathcal{CN}(0,\sigma_u^2). By reciprocity, the target echo at the BS is

ys=ξGTΘhthtTΘGs+ns,(2)y_s=\xi\mathbf{G}^{\mathrm{T}}\boldsymbol{\Theta}\mathbf{h}_t\mathbf{h}_t^{\mathrm{T}}\boldsymbol{\Theta}\mathbf{G}s+n_s, \tag*{(2)}

with reflection coefficient ξ\xi and ns∼CN(0,σs2I)n_s\sim\mathcal{CN}(0,\sigma_s^2\mathbf{I}).

Lemma 1 (Transmit design): For G=αˉeaH\mathbf{G}=\bar{\alpha}\mathbf{e}\mathbf{a}^{\mathrm{H}} and any Θ\boldsymbol{\Theta} and x\mathbf{x}, a single stream s=wss=\mathbf{w}s with maximum-ratio transmission (MRT), w=Pa/∥a∥\mathbf{w}=\sqrt{P}\mathbf{a}/\lVert\mathbf{a}\rVert, maximizes both the user SNR and the echo SNR.

Proof: Write R=Rd+Rs\mathbf{R}=\mathbf{R}_d+\mathbf{R}_s, where Rd\mathbf{R}_d carries the user data and Rs\mathbf{R}_s dedicated sensing streams. The user receives the useful power ∣αˉ∣2∣huHΘe∣2aHRda|\bar{\alpha}|^2|\mathbf{h}_u^{\mathrm{H}}\boldsymbol{\Theta}\mathbf{e}|^2\mathbf{a}^{\mathrm{H}}\mathbf{R}_d\mathbf{a} and the interference ∣αˉ∣2∣huHΘe∣2aHRsa|\bar{\alpha}|^2|\mathbf{h}_u^{\mathrm{H}}\boldsymbol{\Theta}\mathbf{e}|^2\mathbf{a}^{\mathrm{H}}\mathbf{R}_s\mathbf{a}, while the echo energy in (2) is proportional to aHRa\mathbf{a}^{\mathrm{H}}\mathbf{R}\mathbf{a}. Hence Rs=0\mathbf{R}_s=0, and aHRda≤P∥a∥2\mathbf{a}^{\mathrm{H}}\mathbf{R}_d\mathbf{a}\le P\lVert\mathbf{a}\rVert^2 with equality iff Rd=PaaH/∥a∥2\mathbf{R}_d=P\mathbf{a}\mathbf{a}^{\mathrm{H}}/\lVert\mathbf{a}\rVert^2.

Lemma 1 is a direct consequence of the rank-one BS–RIS link; it shows that all spatial processing happens at the surface. The incident field is g=αˉP∥a∥e(x,φ)\mathbf{g}=\bar{\alpha}\sqrt{P}\lVert\mathbf{a}\rVert\mathbf{e}(\mathbf{x},\varphi), with ∥g∥2=Pr≜∣αˉ∣2PNM\lVert\mathbf{g}\rVert^2=P_r\triangleq|\bar{\alpha}|^2PNM. With the reflected field u=Θg\mathbf{u}=\boldsymbol{\Theta}\mathbf{g}, the user SNR and the target illumination are

γc=∣huHu∣2σu2,Gt=∣htTu∣2.(3)\gamma_c=\frac{|\mathbf{h}_u^{\mathrm{H}}\mathbf{u}|^2}{\sigma_u^2},\qquad G_t=|\mathbf{h}_t^{\mathrm{T}}\mathbf{u}|^2. \tag*{(3)}

Since eTΘht=htTΘe\mathbf{e}^{\mathrm{T}}\boldsymbol{\Theta}\mathbf{h}_t=\mathbf{h}_t^{\mathrm{T}}\boldsymbol{\Theta}\mathbf{e} for Θ=ΘT\boldsymbol{\Theta}=\boldsymbol{\Theta}^{\mathrm{T}}, the echo becomes ys=ξαˉ2(htTΘe)2(aHw)a∗s+nsy_s=\xi\bar{\alpha}^2(\mathbf{h}_t^{\mathrm{T}}\boldsymbol{\Theta}\mathbf{e})^2(\mathbf{a}^{\mathrm{H}}\mathbf{w})\mathbf{a}^{*}s+n_s. Combining with a/∥a∥\mathbf{a}/\lVert\mathbf{a}\rVert over TT known snapshots gives SNRs=∣ξ∣2TGt2/(Pσs2)_s=|\xi|^2TG_t^2/(P\sigma_s^2). For each placement x\mathbf{x}, we study the achievable region R(x)={(γc,Gt):u=Θg, Θ∈U}\mathcal{R}(\mathbf{x})=\{(\gamma_c,G_t):\mathbf{u}=\boldsymbol{\Theta}\mathbf{g},\ \boldsymbol{\Theta}\in\mathcal{U}\}, where U\mathcal{U} is the feasible set of Θ\boldsymbol{\Theta} for the given architecture, its Pareto boundary, and the placement x∈XD\mathbf{x}\in\mathcal{X}_D that is best for it.

Analysis

Reachable Fields and the Pareto Boundary

Lemma 2: For a group-connected BD-RIS, {Θg:Θ∈U}={u:∥u∥=∥g∥ ∀g}\{\boldsymbol{\Theta}\mathbf{g}:\boldsymbol{\Theta}\in\mathcal{U}\}=\{\mathbf{u}:\lVert\mathbf{u}\rVert=\lVert\mathbf{g}\rVert\ \forall\mathbf{g}\}. In particular, an FC BD-RIS reaches every u\mathbf{u} with ∥u∥=∥g∥\lVert\mathbf{u}\rVert=\lVert\mathbf{g}\rVert, and a diagonal RIS every u\mathbf{u} with ∣um∣=∣gm∣|u_m|=|g_m|.

Proof: Necessity follows from the unitarity of each block. For sufficiency, take one block and drop the index. Choose z\mathbf{z} with ∥z∥=∥g∥\lVert\mathbf{z}\rVert=\lVert\mathbf{g}\rVert and zTz=gTu\mathbf{z}^{\mathrm{T}}\mathbf{z}=\mathbf{g}^{\mathrm{T}}\mathbf{u}, e.g., z=∥g∥ejϕ[cos⁡τ, jsin⁡τ, 0,…,0]T\mathbf{z}=\lVert\mathbf{g}\rVert e^{j\phi}[\cos\tau,\ j\sin\tau,\ 0,\ldots,0]^{\mathrm{T}} with ϕ=∠(gTu)/2\phi=\angle(\mathbf{g}^{\mathrm{T}}\mathbf{u})/2 and cos⁡2τ=∣gTu∣2/∥g∥2∥u∥2<1\cos2\tau=|\mathbf{g}^{\mathrm{T}}\mathbf{u}|^2/\lVert\mathbf{g}\rVert^2\lVert\mathbf{u}\rVert^2<1. The pairs (z,z∗)(\mathbf{z},\mathbf{z}^{*}) and (g∗,u)(\mathbf{g}^{*},\mathbf{u}) have the same Gram matrix, so a unitary Q\mathbf{Q} exists with Qz∗=g∗\mathbf{Q}\mathbf{z}^{*}=\mathbf{g}^{*} and Qz=u\mathbf{Q}\mathbf{z}=\mathbf{u}. Then Θ=QQT\boldsymbol{\Theta}=\mathbf{Q}\mathbf{Q}^{\mathrm{T}} is symmetric and unitary, and Θg=Q(QHg)∗=Qz=u\boldsymbol{\Theta}\mathbf{g}=\mathbf{Q}(\mathbf{Q}^{\mathrm{H}}\mathbf{g})^{*}=\mathbf{Q}\mathbf{z}=\mathbf{u}.

Lemma 2 concerns ideal scattering matrices. The reactance X=−jZ0(I+Θ)(I−Θ)−1\mathbf{X}=-jZ_0(\mathbf{I}+\boldsymbol{\Theta})(\mathbf{I}-\boldsymbol{\Theta})^{-1} is finite iff 11 is not an eigenvalue of Θ\boldsymbol{\Theta}. When it is, e.g., for u=g\mathbf{u}=\mathbf{g}, the field is approached with growing reactances. Since (3) depends on u\mathbf{u} only up to a common phase, u\mathbf{u} can be rotated, and Q\mathbf{Q} is free on the orthogonal complement.

Let p=hu∗/∥hu∥\mathbf{p}=\mathbf{h}_u^{*}/\lVert\mathbf{h}_u\rVert and q=ht∗/∥ht∥\mathbf{q}=\mathbf{h}_t^{*}/\lVert\mathbf{h}_t\rVert, so that γc=γˉc∣pHu∣2/Pr\gamma_c=\bar{\gamma}_c|\mathbf{p}^{\mathrm{H}}\mathbf{u}|^2/P_r and Gt=Gˉt∣qHu∣2/PrG_t=\bar{G}_t|\mathbf{q}^{\mathrm{H}}\mathbf{u}|^2/P_r, where γˉc=Pr∥hu∥2/σu2\bar{\gamma}_c=P_r\lVert\mathbf{h}_u\rVert^2/\sigma_u^2 and Gˉt=Pr∥ht∥2\bar{G}_t=P_r\lVert\mathbf{h}_t\rVert^2 are the single-function maxima. Define

ρ=∣pHq∣=∣htHhu∣∥hu∥∥ht∥,ψ=arccos⁡ρ∈[0,π/2].(4)\rho=|\mathbf{p}^{\mathrm{H}}\mathbf{q}|=\frac{|\mathbf{h}_t^{\mathrm{H}}\mathbf{h}_u|}{\lVert\mathbf{h}_u\rVert\lVert\mathbf{h}_t\rVert},\qquad\psi=\arccos\rho\in[0,\pi/2]. \tag*{(4)}

Theorem 1: For an FC BD-RIS and arbitrary hu,ht\mathbf{h}_u,\mathbf{h}_t with 0<ρ<10<\rho<1, the Pareto boundary of R(x)\mathcal{R}(\mathbf{x}) is

{(γˉccos⁡2ϑ,Gˉtcos⁡2(ψ−ϑ)):0≤ϑ≤ψ},(5)\{(\bar{\gamma}_c\cos^2\vartheta,\bar{G}_t\cos^2(\psi-\vartheta)):0\le\vartheta\le\psi\}, \tag*{(5)}

attained by u∗(ϑ)=Pr[sin⁡(ψ−ϑ)p+sin⁡ϑq~]/sin⁡ψ\mathbf{u}^{*}(\vartheta)=\sqrt{P_r}[\sin(\psi-\vartheta)\mathbf{p}+\sin\vartheta\tilde{\mathbf{q}}]/\sin\psi, with q~=e−j∠(pHq)q\tilde{\mathbf{q}}=e^{-j\angle(\mathbf{p}^{\mathrm{H}}\mathbf{q})}\mathbf{q}, and Θ\boldsymbol{\Theta} from Lemma 2. Formula (5) also holds for ρ=0\rho=0, with q~=q\tilde{\mathbf{q}}=\mathbf{q}, and for ρ=1\rho=1, where the boundary is the single point (γˉc,Gˉt)(\bar{\gamma}_c,\bar{G}_t) attained by u=Prp\mathbf{u}=\sqrt{P_r}\mathbf{p}.

Proof: For ∣pHu∣2=Prcos⁡2ϑ\lvert\mathbf{p}^{\mathrm{H}}\mathbf{u}\rvert^2=P_r\cos^2\vartheta, and write u=Pr(cos⁡ϑp+sin⁡ϑr)\mathbf{u}=\sqrt{P_r}(\cos\vartheta\mathbf{p}+\sin\vartheta\mathbf{r}), with r⊥p\mathbf{r}\perp\mathbf{p} of unit norm. Then q~=ρp+1−ρ2p⊥\tilde{\mathbf{q}}=\rho\mathbf{p}+\sqrt{1-\rho^2}\mathbf{p}_{\perp} with p⊥⊥p\mathbf{p}_{\perp}\perp\mathbf{p} of unit norm. Then ∣q~Hu∣/Pr≤ρcos⁡ϑ+1−ρ2sin⁡ϑ=cos⁡(ψ−ϑ)|\tilde{\mathbf{q}}^{\mathrm{H}}\mathbf{u}|/\sqrt{P_r}\le\rho\cos\vartheta+\sqrt{1-\rho^2}\sin\vartheta=\cos(\psi-\vartheta), with equality iff r=ejϑp⊥\mathbf{r}=e^{j\vartheta}\mathbf{p}_{\perp}, i.e., u=ejϑu∗(ϑ)\mathbf{u}=e^{j\vartheta}\mathbf{u}^{*}(\vartheta). Points with ϑ>ψ\vartheta>\psi are dominated by ϑ=ψ\vartheta=\psi.

This is the two-direction structure of the MISO Pareto boundary [10]. The new element is Lemma 2, which makes a passive FC BD-RIS behave like a fully digital transmitter.

Corollary 1: (a) For η∈(0,1]\eta\in(0,1] and a requirement γc≥ηγˉc\gamma_c\ge\eta\bar{\gamma}_c, the maximum illumination is Gt⋆=Gˉtcos⁡2(ψ−arccos⁡η)G_t^{\star}=\bar{G}_t\cos^2(\psi-\arccos\sqrt{\eta}), which equals Gˉt\bar{G}_t iff ρ≥η\rho\ge\sqrt{\eta}. (b) R(x)\mathcal{R}(\mathbf{x}) is convex, hence a weighted-sum boundary point of R\mathcal{R} and conj⁡R\operatorname{conj}\mathcal{R} [11], and max⁡wγc/γˉc+(1−w)Gt/Gˉt=12[1+1−4w(1−w)(1−ρ2)]\max w\gamma_c/\bar{\gamma}_c+(1-w)G_t/\bar{G}_t=\frac{1}{2}[1+\sqrt{1-4w(1-w)(1-\rho^2)}]. Both increase with ρ\rho.

Proof: (a) γc≥ηγˉc\gamma_c\ge\eta\bar{\gamma}_c means ϑ≤arccos⁡η\vartheta\le\arccos\sqrt{\eta} in (5), and GtG_t increases with ϑ\vartheta on [0,ψ][0,\psi]. (b) The maximum is the largest eigenvalue of wppH+(1−w)qqHw\mathbf{p}\mathbf{p}^{\mathrm{H}}+(1-w)\mathbf{q}\mathbf{q}^{\mathrm{H}}, whose nonzero eigenvalues are those of a 2×22\times2 Gram matrix with trace 11 and determinant w(1−w)(1−ρ2)w(1-w)(1-\rho^2).

For group-connected and diagonal RISs, Lemma 2 turns the weighted sum into max⁡∥uG∥=∥gG∥uHBwBwHu\max_{\lVert\mathbf{u}_{\mathcal{G}}\rVert=\lVert\mathbf{g}_{\mathcal{G}}\rVert}\mathbf{u}^{\mathrm{H}}\mathbf{B}_{w}\mathbf{B}_{w}^{\mathrm{H}}\mathbf{u} over a product of spheres, with Bw=[w p,1−w q]\mathbf{B}_{w}=[\sqrt{w}\,\mathbf{p},\sqrt{1-w}\,\mathbf{q}].

Proposition 1: For any group partition,

max⁡∥uG∥=∥gG∥uHBwBwHu=max⁡z∈C2,∥z∥=1(∑G∥gG∥∥[Bwz]G∥)2(6)\max_{\lVert\mathbf{u}_{\mathcal{G}}\rVert=\lVert\mathbf{g}_{\mathcal{G}}\rVert}\mathbf{u}^{\mathrm{H}}\mathbf{B}_{w}\mathbf{B}_{w}^{\mathrm{H}}\mathbf{u} = \max_{\mathbf{z}\in\mathbb{C}^{2},\lVert\mathbf{z}\rVert=1} \left(\sum_{\mathcal{G}}\lVert\mathbf{g}_{\mathcal{G}}\rVert\lVert[\mathbf{B}_{w}\mathbf{z}]_{\mathcal{G}}\rVert\right)^{2} \tag*{(6)}

and a maximizer is uG=∥gG∥vG/∥vG∥\mathbf{u}_{\mathcal{G}}=\lVert\mathbf{g}_{\mathcal{G}}\rVert\mathbf{v}_{\mathcal{G}}/\lVert\mathbf{v}_{\mathcal{G}}\rVert with v=Bwz⋆\mathbf{v}=\mathbf{B}_{w}\mathbf{z}^{\star}, where any uG\mathbf{u}_{\mathcal{G}} with ∥uG∥=∥gG∥\lVert\mathbf{u}_{\mathcal{G}}\rVert=\lVert\mathbf{g}_{\mathcal{G}}\rVert may be used if vG=0\mathbf{v}_{\mathcal{G}}=0. Since z\mathbf{z} has two real degrees of freedom up to a common phase, every supported point of the boundary follows from a two-dimensional search.

Proof. uHBwBwHu=max⁡∥z∥=1∣(Bwz)Hu∣2\mathbf{u}^{\mathrm{H}}\mathbf{B}_{w}\mathbf{B}_{w}^{\mathrm{H}}\mathbf{u}=\max_{\lVert\mathbf{z}\rVert=1}|(\mathbf{B}_{w}\mathbf{z})^{\mathrm{H}}\mathbf{u}|^{2}. For fixed z\mathbf{z}, the Cauchy–Schwarz inequality on each block gives ∣∑GvGHuG∣≤∑G∥vG∥∥gG∥|\sum_{\mathcal{G}}\mathbf{v}_{\mathcal{G}}^{\mathrm{H}}\mathbf{u}_{\mathcal{G}}|\le\sum_{\mathcal{G}}\lVert\mathbf{v}_{\mathcal{G}}\rVert\lVert\mathbf{g}_{\mathcal{G}}\rVert, with equality at the stated u\mathbf{u}. Exchanging the two maximizations gives (6).

When the region is nonconvex, as for the diagonal RIS in Figure 2, its unsupported points are computed numerically: we maximize GtG_{\mathrm{t}} under γc≥ηγˉc\gamma_{\mathrm{c}}\ge\eta\bar{\gamma}_{\mathrm{c}} over the same family v=Bwz\mathbf{v}=\mathbf{B}_{w}\mathbf{z} with local refinement, and check the result against random-restart SQP over all MM phases. We make no global-optimality claim for these points.

Pareto boundaries for two movable apertures and the fixed $\lambda/2$ array

Figure 2. Pareto boundaries for two movable apertures and the fixed λ/2\lambda/2 array (M=16M=16). Thick lines: FC (Theorem 1); thin dashed: diagonal (Proposition 1), whose largest gap to FC is 0.16, 0.04 and 0.003; markers: direct optimization of X\mathbf{X}; green: D≥D⋆D \ge D^\star.

Line of Sight: Positions Act Only Through ρ\rho

Theorem 2: Under LoS, ∥g∥\lVert\mathbf{g}\rVert, ∥hu∥\lVert\mathbf{h}_{\mathrm{u}}\rVert and ∥ht∥\lVert\mathbf{h}_{\mathrm{t}}\rVert do not depend on x\mathbf{x}, and

ρ(x)=1M∣∑m=1Mej2πΔsxm/λ∣,Δs=sin⁡θu−sin⁡θt,(7)\rho(\mathbf{x})=\frac{1}{M}\left|\sum_{m=1}^{M}e^{j2\pi\Delta_{\mathrm{s}}x_{m}/\lambda}\right|,\qquad\Delta_{\mathrm{s}}=\sin\theta_{\mathrm{u}}-\sin\theta_{\mathrm{t}}, \tag*{(7)}

independently of the BS–RIS geometry. For an FC BD-RIS, if ρ(x)≤ρ(x′)\rho(\mathbf{x})\le\rho(\mathbf{x}'), every point of R(x)\mathcal{R}(\mathbf{x}) is weakly Pareto-dominated by a point of R(x′)\mathcal{R}(\mathbf{x}'). Hence every criterion that is nondecreasing in γc\gamma_{\mathrm{c}} and GtG_{\mathrm{t}} is maximized by

x⋆=arg⁡max⁡x∈XD ρ(x).(8)\mathbf{x}^{\star}=\underset{\mathbf{x}\in\mathcal{X}_{D}}{\arg\max}\ \rho(\mathbf{x}). \tag*{(8)}

Proof. All entries have constant modulus, so the norms are fixed and (7) follows from (4). Every point of R(x)\mathcal{R}(\mathbf{x}) is dominated by a point of (5), and for ψ′≤ψ\psi'\le\psi the point ϑ\vartheta of (5) is dominated by the point min⁡(ϑ,ψ′)\min(\vartheta,\psi') of the boundary for ψ′\psi'.

Dominance cannot be replaced by set inclusion: for ρ′=1\rho'=1, R(x′)\mathcal{R}(\mathbf{x}') is the segment {(tγˉc,tGˉt):0≤t≤1}\{(t\bar{\gamma}_{\mathrm{c}},t\bar{G}_{\mathrm{t}}):0\le t\le1\}, which does not contain (γˉc,ρ2Gˉt)(\bar{\gamma}_{\mathrm{c}},\rho^{2}\bar{G}_{\mathrm{t}}).

Theorem 3: Let Λ=λ/∣Δs∣\Lambda=\lambda/|\Delta_{\mathrm{s}}|. Then ρ(x)=1\rho(\mathbf{x})=1 iff xm∈x1+ΛZx_{m}\in x_{1}+\Lambda\mathbb{Z} for all mm, and

max⁡x∈XDρ(x)=1⟺D≥D⋆=(M−1)λ∣Δs∣,(9)\max_{\mathbf{x}\in\mathcal{X}_{D}}\rho(\mathbf{x})=1\Longleftrightarrow D\ge D^{\star}=\frac{(M-1)\lambda}{|\Delta_{\mathrm{s}}|}, \tag*{(9)}

attained by xm=(m−1)Λx_{m}=(m-1)\Lambda. At this placement, the Pareto boundary of the FC BD-RIS reduces to the single point (γˉc,Gˉt)(\bar{\gamma}_{\mathrm{c}},\bar{G}_{\mathrm{t}}): the user and the target are served as if each were alone.

Proof. Equality in the triangle inequality for Eq. (7) holds iff all phases 2πΔsxm/λ2\pi\Delta_{\mathrm{s}}x_{m}/\lambda coincide modulo 2π2\pi. Since ∣Δs∣≤2|\Delta_{\mathrm{s}}|\le2, Λ≥λ/2≥d0\Lambda\ge\lambda/2\ge d_{0}, so MM ordered lattice points are feasible iff their span (M−1)Λ(M-1)\Lambda fits in DD. With ψ=0\psi=0, Theorem 1 gives (γˉc,Gˉt)(\bar{\gamma}_{\mathrm{c}},\bar{G}_{\mathrm{t}}).

Proposition 2 (Clustered lattice): Let cc divide MM with cd0≤Λcd_{0}\le\Lambda. Placing M/cM/c clusters of cc elements, spaced d0d_{0} within each cluster, at the lattice points nΛn\Lambda achieves

ρc=∣sin⁡(cπΔsd0/λ)csin⁡(πΔsd0/λ)∣\rho_{c}=\left|\frac{\sin(c\pi\Delta_{\mathrm{s}}d_{0}/\lambda)}{c\sin(\pi\Delta_{\mathrm{s}}d_{0}/\lambda)}\right|

with Dc=(Mc−1)Λ+(c−1)d0D_{\mathrm{c}}=\left(\frac{M}{c}-1\right)\Lambda+(c-1)d_{0}.

Hence max⁡x∈XDρ(x)≥ρc\max_{\mathbf{x}\in\mathcal{X}_{D}}\rho(\mathbf{x})\ge\rho_{c} for all D≥DcD\ge D_{\mathrm{c}}, and c=1c=1 recovers Theorem 3.

Proof. The sum in (7) factorizes into the cluster array factor, which gives (10), and the lattice factor, which has unit modulus. Clusters do not overlap because cd0≤Λcd_{0}\le\Lambda.

Corollary 2 (Diagonal RIS): Under LoS, ∥gm∥2=Pr/M\lVert\mathbf{g}_{m}\rVert^{2}=P_{\mathrm{r}}/M, so a diagonal RIS realizes every constant-modulus u\mathbf{u}. (a) The endpoints of (5), u∝p\mathbf{u}\propto\mathbf{p} and u∝q~\mathbf{u}\propto\tilde{\mathbf{q}}, have constant modulus, so a diagonal RIS attains them for any x\mathbf{x}. (b) For D≥D⋆D\ge D^{\star}, a movable diagonal RIS on the lattice attains (γˉc,Gˉt)(\bar{\gamma}_{\mathrm{c}},\bar{G}_{\mathrm{t}}), and interconnections bring no further gain.

For D<D⋆D<D^{\star}, the diagonal region depends on all the phases 2πΔsxm/λ2\pi\Delta_{\mathrm{s}}x_{m}/\lambda, not only on ρ\rho: for M=4M=4, d0=λ/2d_{0}=\lambda/2 and Δs=1/2\Delta_{\mathrm{s}}=1/2, the placements (0,2,4,5)λ/10(0,2,4,5)\lambda/10 and (0,2,2.5,3.5)λ/10(0,2,2.5,3.5)\lambda/10 both give ρ=1/2\rho=1/2, but their best values of γc/γˉc+Gt/Gˉt\gamma_{\mathrm{c}}/\bar{\gamma}_{\mathrm{c}}+G_{\mathrm{t}}/\bar{G}_{\mathrm{t}} are 1.25 and 1.457. The placement (8) is therefore optimal for the FC BD-RIS only. For diagonal and group-connected RISs below D⋆D^{\star}, we use it as an achievable design.

Corollary 3 (Minimum aperture): Let ρmax⁡(D)=max⁡x∈XDρ(x)\rho_{\max}(D)=\max_{\mathbf{x}\in\mathcal{X}_{D}}\rho(\mathbf{x}). The FC BD-RIS meets γc≥ηγˉc\gamma_{\mathrm{c}}\ge\eta\bar{\gamma}_{\mathrm{c}} without any sensing loss iff D≥Dη=min⁡{D:ρmax⁡(D)≥η}D\ge D_{\eta}=\min\{D:\rho_{\max}(D)\ge\sqrt{\eta}\}. Moreover, Dη≤D⋆D_{\eta}\le D^{\star}, with equality for η=1\eta=1, and ρmax⁡(D)\rho_{\max}(D) is nondecreasing because XD⊂XD′\mathcal{X}_{D}\subset\mathcal{X}_{D'} for D≤D′D\le D'.

Remark 1 (Position errors): For i.i.d. errors em∼N(0,σ2)e_{m}\sim\mathcal{N}(0,\sigma^{2}) on xmx_{m}, Jensen’s inequality gives E[ρ]≥ρ(x)exp⁡(−2π2Δs2σ2/λ2)\mathbb{E}[\rho]\ge\rho(\mathbf{x})\exp(-2\pi^{2}\Delta_{\mathrm{s}}^{2}\sigma^{2}/\lambda^{2}). For σ=λ/20\sigma=\lambda/20 and Δs=0.52\Delta_{\mathrm{s}}=0.52, the factor is 0.9870.987, so the lattice is robust to small positioning errors.

Placement for D<D and Beyond LoS

For a smaller aperture, (8) is nonconvex. Using ∣z∣=max⁡ωℜ{e−jωz}|z|=\max_{\omega}\Re\{e^{-j\omega}z\},

Mmax⁡x∈XDρ(x)=max⁡ωmax⁡x∈XD∑m=1Mcos⁡(2πΔsxm/λ−ω).(10)M\max_{\mathbf{x}\in\mathcal{X}_{D}}\rho(\mathbf{x}) = \max_{\omega}\max_{\mathbf{x}\in\mathcal{X}_{D}}\sum_{m=1}^{M}\cos(2\pi\Delta_{\mathrm{s}}x_{m}/\lambda-\omega). \tag*{(10)}

For fixed ω\omega, the inner problem is separable apart from the ordering constraints in (1). On a grid of step δ\delta with d0=Kδd_{0}=K\delta and fi=cos⁡(2πΔsiδ/λ−ω)f_{i}=\cos(2\pi\Delta_{\mathrm{s}}i\delta/\lambda-\omega), the recursion

V1(i)=fi,Vm(i)=fi+max⁡j≤i−KVm−1(j),(11)V_{1}(i)=f_{i},\qquad V_{m}(i)=f_{i}+\max_{j\le i-K}V_{m-1}(j), \tag*{(11)}

evaluated with running maxima, solves the inner grid problem exactly in O(MD/δ)O(MD/\delta) operations.

Algorithm 1 Movable BD-RIS ISAC design

Input: Angles θu≠θt\theta_u \ne\theta_t, aperture DD, requirement η\eta

  1. Placement: If D≥D⋆D \ge D^\star, set xm=(m−1)λ/∣Δs∣x_m = (m-1)\lambda/|\Delta_s| (Theorem 3); otherwise solve (11) by the DP (12). For non-LoS channels, refine by coordinate ascent.

  2. Transmitter: MRT toward the RIS (Lemma 1).

  3. Field: FC: u=u⋆(min⁡(ψ,arccos⁡η))\mathbf{u}=\mathbf{u}^\star(\min(\psi,\arccos\sqrt{\eta})) (Theorem 1); group-connected or diagonal: search over z\mathbf{z} in (6).

  4. Scattering matrix: Synthesize Θ\mathbf{\Theta} and X\mathbf{X} blockwise by Lemma 2.

Proposition 3 (Accuracy): Let ρ^\hat{\rho} be the best ρ\rho returned by (12) over NωN_\omega equispaced values of ω\omega. Then ρ^≥cos⁡(π/Nω)[ρmax⁡(D)−2π∣Δs∣δ/λ]\hat{\rho} \ge\cos(\pi/N_\omega)[\rho_{\max}(D)-2\pi|\Delta_s|\delta/\lambda].

Proof. Rounding every position of a continuous optimizer down to the grid keeps it in XD\mathcal{X}_D, because d0=Kδd_0=K\delta, and changes each phase by at most 2π∣Δs∣δ/λ2\pi|\Delta_s|\delta/\lambda. The grid optimum therefore loses at most 2π∣Δs∣δ/λ2\pi|\Delta_s|\delta/\lambda in ρ\rho. For the grid optimum x∘\mathbf{x}^\circ, some ωk\omega_k lies within π/Nω\pi/N_\omega of ∠∑mej2πΔsxm∘/λ\angle\sum_m e^{j2\pi\Delta_s x_m^\circ/\lambda}, and the DP at ωk\omega_k returns x\mathbf{x} with Mρ(x)≥Mρ(x∘)cos⁡(π/Nω)M\rho(\mathbf{x}) \ge M\rho(\mathbf{x}^\circ)\cos(\pi/N_\omega).

We use Nω=96N_\omega=96 with local refinement around the best ω\omega, so the factor is 0.9995. For D≥D⋆D \ge D^\star, the analytical lattice of Theorem 3 is used directly.

For non-LoS channels, Theorem 1, Corollary 1 and Proposition 1 still hold, but γˉc\bar{\gamma}_c depends on x\mathbf{x}. For an absolute requirement γc≥Γ\gamma_c \ge\Gamma, Corollary 1(a) with η(x)=min⁡(1,Γ/γˉc(x))\eta(\mathbf{x})=\min(1,\Gamma/\bar{\gamma}_c(\mathbf{x})) gives a closed-form objective in ρ(x)\rho(\mathbf{x}) and ∥hu(x)∥\lVert\mathbf{h}_u(\mathbf{x})\rVert. If Γ>γˉc(x)\Gamma>\bar{\gamma}_c(\mathbf{x}), the user is served at γˉc(x)\bar{\gamma}_c(\mathbf{x}). We improve this objective by exact grid updates of each xmx_m between its neighbours, initialized at the lattice, accepting only strict improvements. The search is monotone and terminates, but it is local. Algorithm 1 summarizes the design.

Complexity: Step 1 costs O(NωMD/δ)O(N_\omega MD/\delta), step 3 costs O(M)O(M) for the FC BD-RIS and O(NsM)O(N_sM) for NsN_s search points otherwise, and step 4 costs O(MMg2)O(MM_g^2) over all groups. Unlike [4], [5], [7], [8], no alternating optimization is needed.

Simulation Results

In this section, we asses the perforamnce of the proposed design. Table I lists the simulation parameters. Under LoS, γˉc\bar{\gamma}_c and Gˉt\bar{G}_t do not depend on x\mathbf{x}, so we report metrics normalized by these single-function maxima; they hold for any transmit power PP, number of BS antennas NN, path loss and reflection coefficient ξ\xi. Since SNRs ∝Gt2\propto G_t^2, the sensing SNR loss is −20log⁡10(Gt/Gˉt)-20\log_{10}(G_t/\bar{G}_t) dB.

Pareto boundaries: Figure 2 compares D=10λD=10\lambda and D=20λD=20\lambda with the fixed λ/2\lambda/2 array. The FC curves are exact by Theorem 1. As a consistency check, the markers optimize X\mathbf{X} directly by L-BFGS from random starts, without using Lemma 2. The FC markers agree with (5) to within 10−910^{-9}. The diagonal markers (best of 150 starts) lie on the supported points from Proposition 1. The fixed array has nearly orthogonal channels (ρ=0.033\rho=0.033); movement raises ρ\rho to 0.39 and 0.81, pushing the boundary outward. For D≥D⋆D \ge D^\star, the Pareto boundary of both the FC and the diagonal RIS is the corner (1,1)(1,1) (green star), while the achievable region is the green dotted segment, as

ParameterSymbolValue
RIS elementsMM16 (32 in Figs. 4, 3b, 3c)
Minimum spacingd0d_0λ/2\lambda/2
Carrier frequencyfcf_c28 GHz (λ=1.07\lambda=1.07 cm)
BS–RIS angle of arrivalφ\varphi−35∘-35^\circ
User / target angleθu,θt\theta_u,\theta_t20∘/−10∘20^\circ/-10^\circ ([−60∘,15∘-60^\circ,15^\circ] in Fig. 3b)
Spatial-frequency offsetΔs\Delta_s0.516
Lattice spacingΛ\Lambda1.94λ1.94\lambda
Trade-off-free apertureD⋆D^\star29.1λ29.1\lambda (60.1λ\lambda for M=32M=32)
Fixed baseline—λ/2\lambda/2-spaced array
DP grid / phase searchδ,Nω\delta,N_\omegaλ/100\lambda/100 / 96
User SNR loss toleranceη\eta1 dB, 3 dB
Group sizeMgM_g4
Rician RIS–user link (Fig. 3c)
Rician factorKK−5-5 to 25 dB
Scattered pathsLL5, angles uniform in [−60∘,60∘][-60^\circ,60^\circ]
User requirementΓ\Gamma3 dB below LoS maximum
Position step / realizations—λ/50\lambda/50 / 100

Table 1. Simulation parameters.

Theorem 3 predicts. The diagonal boundary touches the FC boundary at its endpoints (Corollary 2); for the fixed array it is nonconvex. The figure thus separates the two degrees of freedom: interconnection matters only at fixed positions, where the largest gap between the FC and diagonal boundaries is 0.16, whereas movement shrinks this gap to 0.04 and 0.003 at D=10λD=10\lambda and 20λ20\lambda and removes it at D≥D⋆D \ge D^\star, while enlarging the region of both architectures far more.

Moving vs. interconnecting: Fig. 3a shows the sensing SNR loss when the user may lose 1 dB or 3 dB of SNR. The fixed FC BD-RIS loses 12.7 and 5.5 dB; the fixed diagonal RIS (not shown) loses 17.1 and 8.5 dB. With movement, the FC loss vanishes once ρ≥η\rho\ge\sqrt{\eta} (Corollary 3), at D=25λD=25\lambda and 15.5λ15.5\lambda. With the same placement, the diagonal RIS matches the FC BD-RIS from these apertures on and trails by at most 0.61 and 0.29 dB for D≥10λD \ge10\lambda. A group-connected RIS with Mg=4M_g=4 stays within 0.14 dB of the FC BD-RIS.

Geometry mismatch: In Fig. 3b, the aperture is fixed at D=[D⋆]D=[D^\star] for the nominal angle, and the user must keep its full SNR (η=1\eta=1, so Gt/Gˉt=ρ2G_t/\bar{G}_t=\rho^2). The comparison isolates the value of movability at equal aperture: a fixed sparse array on the nominal lattice is optimal at θt=−10∘\theta_t=-10^\circ, but away from it its illumination stays below 0.53 (M=16M=16) and 0.03 (M=M= Fig. 3. Sensing performance of the proposed architecture: (a) impact of aperture size and user SNR requirements; (b) target-angle mismatch; (c) robustness under Rician fading.

Figure 3a. Sensing SNR loss versus aperture for user SNR losses of 1 dB (solid) and 3 dB (dashed), M=16M = 16.

Figure 3b. Normalized illumination at full user SNR versus target angle, at equal aperture D=⌈D⋆⌉D = \lceil D^\star\rceil designed for θt=−10∘\theta_t = -10^\circ.

Figure 3c. Average sensing SNR loss versus the Rician factor of the RIS–user link (absolute user requirement 3 dB below the LoS maximum, 100 realizations, 95% confidence intervals).

Figure 4. Maximum correlation versus aperture for M∈{16,32}M \in\{16, 32\}; dotted lines mark D⋆D^\star (Theorem 3), green circles Proposition 2.

32), close to the λ/2\lambda/2 array. The movable surface, re-placed by the DP for each θt\theta_t, keeps Gt=GˉtG_t = \bar{G}_t for θt≤−10∘\theta_t \leq-10^\circ, where D≥D⋆D \geq D^\star, and above 0.80Gˉt0.80\bar{G}_t when the target moves toward the user and D<D⋆D < D^\star.

Rician fading: In Fig. 3c, the RIS–user link has L=5L = 5 scattered paths at angles uniform in [−60∘,60∘][-60^\circ, 60^\circ], and the user requires an absolute SNR 3 dB below its LoS maximum. Results are averaged over 100 realizations with 95% confidence intervals. The LoS lattice of Theorem 3 loses 1.1 dB at a Rician factor of −5-5 dB and less than 0.2 dB from 0 dB on. Coordinatewise re-optimization reduces the loss to at most 0.05 dB (FC) and 0.14 dB (diagonal) for both MM, while the fixed λ/2\lambda/2 array loses 4.7–5.5 dB. The requirement cannot be met in at most 3% of the realizations with the re-optimized placement and 7% with the fixed array; in those cases the user is served at its maximum.

Phase transition: Fig. 4 plots ρmax⁡(D)\rho_{\max}(D) for M∈{16,32}M \in\{16, 32\}. The DP curves approach 1 as D→D⋆D \to D^\star, where the analytical lattice gives exactly ρ=1\rho= 1. At D=D⋆/2D = D^\star/2 the DP attains ρ=0.69\rho= 0.69 for both MM, whereas the best uniform spacing stays at 0.08 (M=16M = 16) and 0.04 (M=32M = 32). The plateau is the clustered lattice of Proposition 2 with c=2c = 2 (circles): the DP returns pairs of elements on the lattice and attains ρ2=0.6895\rho_2 = 0.6895 at D2=14.1λD_2 = 14.1\lambda (29.6λ\lambda for M=32M = 32).

Conclusion

For a movable BD-RIS serving one user and illuminating one target, the communication–sensing Pareto boundary is available in closed form and is governed by a single quantity, the correlation between the user and target channels at the surface. Under LoS, the placement that maximizes this correlation is optimal for the FC BD-RIS. A lattice of spacing λ/∣Δs∣\lambda/|\Delta_s| steers a grating lobe at the target and reduces the Pareto boundary to the single-function optimum once D≥(M−1)λ/∣Δs∣D \geq(M-1)\lambda/|\Delta_s|; there, a movable diagonal RIS is also optimal. Beyond this aperture, a DP with certified accuracy and a clustered lattice give the placement, and the diagonal RIS remains within a fraction of a decibel of the FC BD-RIS at moderate apertures. The gains hold at equal aperture under geometry mismatch and, after local re-optimization, under Rician fading.

References

  1. [1]F. Liu, Y. Cui, C. Masouros, J. Xu, T. X. Han, Y. C. Eldar, and S. Buzzi, “Integrated sensing and communications: Toward dual-functional wireless networks for 6G and beyond,” IEEE J. Sel. Areas Commun., vol. 40, no. 6, pp. 1728–1767, Jun. 2022.DOI
  2. [2]S. Chen, B. Clerckx, and R. Murch, “Modeling and architecture design of reconfigurable intelligent surfaces using scattering parameter network analysis,” IEEE Trans. Wireless Commun., vol. 21, no. 2, pp. 1229–1243, Feb. 2022.
  3. [3]M. Nerini, S. Shen, and B. Clerckx, “Closed-form global optimization of beyond diagonal reconfigurable intelligent surfaces,” IEEE Trans. Wireless Commun., vol. 23, no. 2, pp. 1037–1051, Feb. 2024.arxiv.org/abs/2211.06117
  4. [4]T. Esmaeilbeig, K. V. Mishra, and M. Soltanalian, “Beyond diagonal RIS: Key to next-generation integrated sensing and communications?” IEEE Signal Process. Lett., vol. 32, pp. 216–220, 2025.arxiv.org/abs/2402.14157
  5. [5]Z. Liu, Y. Liu, S. Shen, Q. Wu, and Q. Shi, “Enhancing ISAC network throughput using beyond diagonal RIS,” IEEE Wireless Commun. Lett., vol. 13, no. 6, pp. 1670–1674, Jun. 2024.
  6. [6]L. Zhu, W. Ma, and R. Zhang, “Movable antennas for wireless communication: Opportunities and challenges,” IEEE Commun. Mag., vol. 62, no. 6, pp. 114–120, Jun. 2024.arxiv.org/abs/2306.02331
  7. [7]J. Ye, P. Zhang, X.-P. Li, L. Huang, and Y. Liu, “Joint beamforming and position optimization for fluid RIS-aided ISAC systems,” IEEE Trans. Commun., vol. 74, pp. 114 422–114 438, 2026.arxiv.org/abs/2501.13339
  8. [8]X. Peng, Q. Tao, Y. L. Guan, and X. Chen, “Exploiting movable elements of intelligent reflecting surface for enhancement of integrated sensing and communication,” IEEE Trans. Wireless Commun., vol. 25, pp. 1717–1733, 2026.arxiv.org/abs/2507.23296
  9. [9]S. Xu, M. Nerini, and B. Clerckx, “Movable beyond-diagonal reconfigurable intelligent surfaces: MOK, interconnecting, or both?” IEEE Wireless Commun. Lett., vol. 15, pp. 2144–2148, 2026.
  10. [10]E. A. Jorswieck, E. G. Larsson, and D. Danev, “Complete characterization of the Pareto boundary for the MISO interference channel,” IEEE Trans. Signal Process., vol. 56, no. 10, pp. 5292–5296, Oct. 2008.DOI
  11. [11]R. A. Horn and C. R. Johnson, Topics in Matrix Analysis. Cambridge, U.K.: Cambridge Univ. Press, 1991.

Paper details

vlnc:W_pbn6xqgh
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Engineering

Not recorded for this paper

View statement
Looking for related papers…
Contents