FIRES: Fluid Integrated Reflecting and Emitting Surfaces

This letter introduces the concept of fluid integrated reflecting and emitting surface (FIRES), which constitutes a new paradigm seamlessly integrating the flexibility of fluid-antenna systems (FASs) with the dual functionality of simultaneous transmitting and reflecting reconfigurable intelligent surfaces (STAR-RISs). The potential of the proposed metasurface structure is studied though an FIRES-enabled multicast system based on the energy splitting protocol. In this model, the FIRES is divided into non-overlapping subareas, each functioning as a ‘fluid’ element capable of concurrent reflection and transmission and changing its position of radiation within the subarea. In particular, we formulate an optimization problem for the design of the triple tunable features of the surface unit elements, which is solved via a tailored particle swarm optimization approach. Our results showcase that the proposed FIRES architecture significantly outperforms its conventional STAR-RIS counterpart.

💡 Research Summary

The paper introduces FIRES (Fluid Integrated Reflecting and Emitting Surfaces), a novel metasurface architecture that merges the reconfigurability of fluid‑antenna systems (FAS) with the dual‑mode capability of simultaneous transmitting and reflecting RIS (STAR‑RIS). FIRES partitions a physical aperture of area A into M non‑overlapping sub‑areas; each sub‑area hosts a “fluid” element that can select one of Nₘ preset positions within its region, thereby allowing dynamic relocation of the radiating point. In addition to position, each element independently controls a power‑splitting ratio βᵤ (βᵣ+βₜ=1) and two phase shifts ϕᵣₘ, ϕₜₘ for reflected and transmitted signals. This three‑fold tunability (position, phase, power) provides far more degrees of freedom than conventional RIS or even STAR‑RIS.

The authors consider a downlink multicast scenario where a fixed‑position base station (BS) communicates with two users—one located in the reflection region (user r) and the other in the transmission region (user t)—through FIRES, while the direct BS‑user links are blocked. Channels BS‑FIRES and FIRES‑user are modeled as Rician fading with both LOS and NLOS components. The LOS component is expressed via steering vectors that depend on the instantaneous element coordinates (xₘ,yₘ) and the azimuth/elevation angles of arrival/departure. NLOS components are spatially correlated using Jakes’ model, yielding a correlation matrix R_q that captures the effect of closely spaced preset positions.

The received signal at user u follows the energy‑splitting (ES) protocol: yᵤ = √P hᵤᴴ Φᵤ h_f x + zᵤ, where Φᵤ = diag(√βᵤ e^{jϕᵤ₁}, …, √βᵤ e^{jϕᵤ_M}) and βᵤ is common across elements for simplicity. The resulting SNR γᵤ = P|h_fᴴ Φᵤ hᵤ|²/σ² leads to an achievable rate Rᵤ = log₂(1+γᵤ). The system objective is to maximize the minimum of the two users’ rates (max‑min fairness), i.e., maximize R_eff = min{R_r,R_t}.

The optimization problem (P1) jointly selects the positions rₘ ∈ Sₘ, the power‑splitting coefficients βᵣ,βₜ, and the phase shifts ϕᵤₘ, subject to (i) each element staying inside its designated sub‑area, (ii) a minimum inter‑element spacing D (to suppress mutual coupling), (iii) βᵣ+βₜ=1, and (iv) a total transmit‑power budget P. Because (P1) is highly non‑convex, the authors adopt a tailored particle‑swarm optimization (PSO) algorithm. Each particle encodes a candidate solution (positions, phases, β values). The velocity update follows the standard PSO rule with inertia weight w, cognitive coefficient c₁, and social coefficient c₂. Constraint violations are penalized through large penalty terms B_power and B_spacing, weighted by a factor τ, and incorporated into the fitness function O = min_u R_u – τ·(B_power + B_spacing). The algorithm iteratively updates particles, enforces the spacing constraint by re‑optimizing any violating element, and stops when convergence criteria are met.

Simulation parameters include M = {4, 9, 16}, N = 100 preset positions per element, aperture A = 4 m², carrier f_c = 3.5 GHz (λ ≈ 8.6 cm), D = λ/2, BS‑FIRES distance d_f = 100 m, FIRES‑user distance d_u = 200 m, path‑loss exponent α = 2.5, noise σ² = –90 dBm, and Rician K‑factors K_f = K_u = 5. The FIRES performance is benchmarked against a conventional STAR‑RIS of equal size and element count. Results show that FIRES consistently outperforms STAR‑RIS, achieving 20–35 % higher effective rates across all M values. The gain is especially pronounced when the users’ angular separation is large, highlighting the benefit of spatial repositioning of elements. The PSO converges within a modest number of iterations (≈100) and scales linearly with M, indicating practical computational complexity.

In summary, the paper makes four key contributions: (1) it formalizes the FIRES concept, integrating fluid‑type position reconfigurability with dual‑mode RIS functionality; (2) it develops a comprehensive system model that captures LOS steering, spatially correlated NLOS fading, and power‑splitting; (3) it proposes a PSO‑based algorithm capable of handling the non‑convex, multi‑variable optimization; and (4) it validates through extensive simulations that FIRES delivers substantial rate improvements over existing STAR‑RIS solutions. The work opens avenues for hardware prototyping, multi‑user extensions, and real‑time adaptive control in dynamic 6G environments.