Wave-Controlled Metasurface-Based Reconfigurable Intelligent Surfaces

TL;DR

This paper introduces a wave-controlled metasurface-based RIS that bypasses per-element wiring by biasing varactors with standing-wave modes, thereby reducing the degrees of freedom needed to tailor reflections. It formalizes a full-domain, basis-function approach to biasing, enabling smooth phase profiles across large RIS surfaces and addressing near-field coupling that limits arbitrary local phase control. The work outlines design choices, including biasing-line topology, standing-wave bias representations, and reduced-dimension parameterizations, while proposing geometry-based sparse channel models and channel-charting techniques to facilitate efficient CSI acquisition and beamforming in multi-cell deployments. The proposed framework aims to dramatically improve hardware efficiency, reduce control complexity, and enable scalable RIS-assisted communication, radar, and navigation with enhanced spectrum utilization and coexistence. Mathematical expressions such as $v(x,t)=V_0+ \sum_{p=1}^{P_x} V_p \sin(k_{b,p} x+\phi_{e,p}) \cos(\omega_p t+\phi_{v,p})$ illustrate the biasing scheme that underpins the wave-controlled RIS operation.

Abstract

Reconfigurable Intelligent Surfaces (RISs) are programmable metasurfaces that can adaptively steer received electromagnetic energy in desired directions by employing controllable phase shifting cells. Among other uses, an RIS can modify the propagation environment in order to provide wireless access to user locations that are not otherwise reachable by a base station. Alternatively, an RIS can steer the waves away from particular locations in space, to eliminate interference and allow for co-existence of the wireless network with other types of fixed wireless services (e.g., radars, unlicensed radio bands, etc.). The novel approach in this work is a wave-controlled architecture that properly accounts for the maximum possible change in the local reflection phase that can be achieved by adjacent RIS elements. It obviates the need for dense wiring and signal paths that would be required for individual control of every RIS element, and thus offers a substantial reduction in the required hardware. We specify this wave-controlled RIS architecture in detail and discuss signal processing and machine learning methods that exploit it in both point-to-point and multicell MIMO systems. Such implementations can lead to a dramatic improvement in next-generation wireless, radar, and navigation systems where RIS finds wide applications. They have the potential to improve the efficiency of spectrum utilization and coexistence by orders of magnitude.

Figures (5)

• Figure 1: RIS-aided wireless system.
• Figure 2: Use of ML for RIS. The presence of the building forces the BS to employ RIS1 to reflect the downlink signal to reach the user when its LoS path is blocked. This redirection operation can be initiated based on UE1's location and tracked over a three-dimensional map in real time. In addition, at points along UE1's trajectory, BS handovers will be required, which can be determined by means of ML algorithms based on a combination of location, signal strength, and predictions of UE1's mobility.
• Figure 3: A wave-controlled RIS provides simplified control of reflection properties. On the top layer the patches provide controlled wave reflection, at the bottom layers the biasing TLs provide the control. Instead of independently controlling each element of the metasurface, a reduced set of full-domain basis functions represented here as a set of voltage standing waves are used for biasing varactors. As an example, RIS elements made of a dual-pol patch with varactors is shown; alternatively the varactor control can be exerted on the patch feeding lines on the bottom level (not shown).
• Figure 4: Example of the optimal RIS phase at different cell positions.
• Figure 5: Opportunities for using ML in a network with RISs, control channels for BSs and RISs, and the use of channel charting DiRenzoetal19. Channel charting can be used for UE localization, handovers, and for determination of CSI. Alternatively, other ML algorithms can be employed for a number of these tasks as described in the text. ML algorithms can be employed in A: control channel operation of the RISs, B: channel estimation, C: configuring RISs to reflect the signal towards the UE, and D: predicting trajectory changes and for optimized beam alignment by the target BS. It is important to realize that there will be much fewer control channels and therefore the task of managing the control channels will be much simpler in the wave-controlled metasurface-based RIS as compared to conventional RISs.