To meet the soaring demand for wireless data rates, next-generation networks are moving up the electromagnetic spectrum into millimeter-wave (mmWave) and Terahertz (THz) bands. These high frequencies offer massive chunks of untouched bandwidth. However, they come with a fatal flaw: they propagate almost strictly via line-of-sight (LoS). Physical obstacles—such as concrete walls, moving vehicles, or even a user's hand—can completely block the signal, creating severe shadowing and communication "dead zones."
From Passive Obstacles to Programmable Mirrors
Traditionally, wireless engineers treated the physical environment as an uncontrollable adversary. If a direct LoS path was blocked, the system had to rely on unpredictable, highly attenuated multi-path reflections, or deploy expensive, power-hungry active relays.
Reconfigurable Intelligent Surfaces (RIS)—also known as Intelligent Reflecting Surfaces (IRS)—fundamentally change this paradigm. An RIS is a planar metasurface composed of hundreds of low-cost, passive scattering elements. Instead of letting radio waves scatter randomly, an RIS programmatically alters the phase shifts of the reflected waves. As pioneered in early applications like the MoVR system by Abari et al., these surfaces act as intelligent, programmable mirrors that can detect an incoming signal and redirect it precisely toward a receiver, effectively bypassing the obstruction.
Why It Works: Overcoming the Double-Fading Effect
To understand why an RIS is so powerful, we must look at the physics of reflection. A standard passive wall or mirror scatters energy widely. The resulting signal experiences what is known as the double-fading effect: the signal power drops proportionally to the product of the transmitter-to-surface distance and the surface-to-receiver distance ($d_1^2 \cdot d_2^2$). This severe path loss usually renders passive reflections useless for high-speed data transfer.
An RIS overcomes this limitation through passive beamforming. By carefully orchestrating the phase shifts across its elements, the surface ensures that the reflected waves interfere constructively at the exact location of the receiver. This transforms a diffuse, scattered reflection into a highly focused, directional beam. As Gao et al. and Noor-A-Rahim et al. demonstrate, this active phase configuration mitigates the double-fading effect, drastically reducing path loss compared to standard non-line-of-sight propagation models. The result is a synthesized, high-quality channel known as a virtual line-of-sight (vLoS) path.
Joint Precoding and the Control Challenge
Creating a robust vLoS path requires a coordinated dance between the transmitter (the active base station) and the RIS (the passive reflector). The active beamforming at the base station must be precisely aligned with the passive phase shifts at the surface. Fang et al. tackled this by designing joint active and passive precoding algorithms that maximize the received signal power, proving that this synergy is essential for robust mmWave communication. Similarly, Cao et al. showed that optimizing this joint relationship allows networks to meet strict low-latency and low-power constraints.
However, calculating these optimal phase shifts in real-time is computationally expensive, especially when users are moving. To bypass heavy mathematical optimization, researchers have turned to machine learning:
- Geometric and Position Learning: Alexandropoulos et al. proposed using Multi-Layer Perceptrons trained on 3D positioning data to dynamically configure phases with minimal computational overhead.
- Reinforcement Learning: In highly dynamic or uncertain environments, Zhang et al. introduced Distributional Reinforcement Learning to maintain high data rates despite channel fluctuations. For complex, multi-node networks, Xu et al. utilized graph-embedded multi-agent deep reinforcement learning to coordinate multiple surfaces without overwhelming the system with data exchange.
- Multi-Task Learning: To handle rapid channel variations in aerial-to-terrestrial links, Cao et al. leveraged Multi-Task Learning to achieve a four-order-of-magnitude reduction in computation time, enabling near-real-time link enhancement.
Ubiquitous Deployment and Network Impact
Because RIS units are passive and do not require expensive radio frequency (RF) chains, they can be distributed widely throughout an environment. Using stochastic geometry, Xu et al. mathematically proved that deploying distributed RISs can boost coverage probability by over 45% and nearly double the network sum rate in high-blockage environments.
This capability makes vLoS technology a critical enabler across diverse, challenging domains:
- Urban and Non-Terrestrial Networks: In dense cities, high-altitude platforms (HAPs) or satellites can struggle to reach street-level users. Shang et al. demonstrated how intelligent surfaces can redirect Free Space Optical (FSO) and radio signals into urban dead zones.
- Advanced Multiple Access: By pairing RIS with Non-Orthogonal Multiple Access (NOMA), Zuo et al. showed that networks can serve multiple users simultaneously through the same synthesized vLoS path, maximizing spectral efficiency.
- Industrial and Transit Environments: In the metallic, highly cluttered environments of smart factories, Noor-A-Rahim et al. highlighted how vLoS beamforming prevents severe shadowing for human-robot collaboration. Meanwhile, Yaacoub et al. pointed to RIS as a future solution for maintaining unbroken connectivity in deep railway tunnels.
By turning the environment from a passive obstacle into an active participant, virtual line-of-sight propagation ensures that next-generation high-frequency networks remain robust, reliable, and truly ubiquitous.
Key Foundational Work
Cutting the Cord in Virtual Reality (2016) — contributed 2 concept elements to this direction.
Phase Configuration Learning in Wireless Networks with Multiple Reconfigurable Intelligent Surfaces (2020) — contributed 1 concept elements to this direction.
Intelligent Reflecting Surface Enhanced Millimeter-Wave NOMA Systems (2020) — contributed 1 concept elements to this direction.