From Passive Walls to Programmable Radio Environments

Traditional wireless networks treat the propagation environment — walls, buildings, furniture — as a fixed, uncontrollable obstacle. The base station and UE adapt to whatever channel the environment provides. Reconfigurable Intelligent Surfaces (RIS) invert this paradigm: they make the environment itself a controllable element of the network.

A RIS is a planar surface composed of hundreds to thousands of sub-wavelength elements (meta-atoms) whose electromagnetic response can be individually programmed. By adjusting the reflection phase of each element, the surface steers, focuses, or reshapes reflected signals without any RF chain, power amplifier, or baseband processing. 3GPP initiated a formal study on RIS in TR 38.857 ("Study on Reconfigurable Intelligent Surfaces"), examining deployment scenarios, channel models, and potential specification impact.

RIS Architecture and Hardware

Meta-Atom Design

Each meta-atom is a printed metallic patch (typically λ/5 to λ/3 in size) on a dielectric substrate, loaded with one or more tunable components. The two dominant tuning technologies are:

PIN diode arrays dominate current prototypes due to manufacturing simplicity and fast switching. A 1-bit RIS quantizes the phase to either 0° or 180°, incurring a ~3.9 dB gain penalty compared to ideal continuous phase control (per the sinc-squared quantization loss formula). 2-bit designs (0°/90°/180°/270°) reduce this penalty to ~0.9 dB.

Control Architecture

The RIS control plane connects each element (or a group of elements) to a microcontroller via a shift register or SPI bus. A separate low-rate control channel links the RIS controller to the base station (gNB). The controller receives beam configuration commands and programs the element phases accordingly. No RF chain exists on the RIS — the reflected signal is entirely passive.

Passive vs Active vs Hybrid RIS

The RIS ecosystem has branched into three architectures with distinct SNR scaling laws:

SNR Gain for Passive RIS

For a passive RIS with N elements and ideal phase alignment, the received power scales as:

`

P_rx ∝ N² · |h_i|² · |g_i|² · P_tx

`

where h_i is the BS-to-RIS channel for element i and g_i is the RIS-to-UE channel. The N² scaling arises because each element adds coherently in both amplitude and phase, yielding a power gain of N² relative to a single element. This is the passive beamforming gain, analogous to coherent combining in an antenna array but applied to the reflected path.

For active RIS, each element introduces an amplification factor α but also amplifier noise, resulting in SNR scaling proportional to N rather than N². The crossover point — where active RIS outperforms passive — depends on the per-element SNR: when individual element SNR is very low (long-range links), active RIS is preferred.

Worked Example 1: Passive RIS Gain Calculation

A passive RIS with N = 400 elements assists a 28 GHz link. Without RIS, the NLoS path loss is 120 dB. Each RIS element has an effective area of (λ/2)² = (5.35 mm)² ≈ 28.6 mm².

`

Passive RIS gain = 10·log10(N²) = 10·log10(400²) = 10·log10(160,000) = 52.0 dB

But this is the ideal coherent combining gain relative to a single element.

The effective RIS-assisted path gain (per TR 38.857 Annex A model):

= Gt + Gr + 10·log10(N²) + 10·log10(dx·dy/λ²) - FSPL_1 - FSPL_2

where FSPL_1 = BS-to-RIS loss, FSPL_2 = RIS-to-UE loss

Assuming BS-to-RIS = 50 m (LoS): FSPL_1 = 92.4 + 20·log10(28) + 20·log10(0.05)

= 92.4 + 28.9 - 26.0 = 95.3 dB

RIS-to-UE = 20 m (NLoS, +10 dB excess): FSPL_2 = 92.4 + 28.9 + 20·log10(0.02) + 10

= 92.4 + 28.9 - 34.0 + 10 = 97.3 dB

Total reflected path loss = 95.3 + 97.3 - 52.0 = 140.6 dB

`

The RIS-assisted path at 140.6 dB is 20.6 dB worse than the direct NLoS path (120 dB) in this geometry. This illustrates a key reality: RIS is most valuable when the direct path is fully blocked (e.g., deep shadow with >30 dB additional penetration loss), not when a weak direct path exists. In a fully blocked scenario where direct loss exceeds 170 dB, the RIS path at 140.6 dB provides viable coverage.

3GPP TR 38.857: RIS Study Item

3GPP RAN initiated the study item on RIS in Release 18 with the technical report TR 38.857. The study covers:

• Deployment scenarios: outdoor BS + outdoor RIS, outdoor BS + indoor RIS (O2I), indoor BS + indoor RIS
• Channel model extensions: cascaded BS → RIS → UE channel, spatial correlation across RIS elements
• Performance evaluation: system-level simulation results showing 5–15 dB SINR improvement in NLoS coverage holes
• Signaling impact: RIS configuration via gNB control, potential new RRC signaling for RIS beam management
• Comparison with repeaters: RIS vs NCR (network-controlled repeater, per TS 38.867)

The study concludes that RIS provides the most benefit in scenarios with severe NLoS conditions and moderate BS-to-RIS distances (<100 m). For longer distances, the double path loss (BS → RIS + RIS → UE) erodes the gain advantage, and active solutions (NCR or small cells) become more effective.

Worked Example 2: RIS vs Network-Controlled Repeater

Compare a 400-element passive RIS with a network-controlled repeater (NCR) providing 30 dB amplification, both assisting a 28 GHz link:

`

Scenario: BS → relay/RIS = 50 m (LoS), relay/RIS → UE = 30 m (NLoS)

NCR path loss:

FSPL (BS→NCR, 50 m, 28 GHz) = 95.3 dB

NCR gain: +30 dB

FSPL (NCR→UE, 30 m, 28 GHz) = 92.4 + 28.9 + 20·log10(0.03) = 90.9 dB

NLoS excess: +10 dB

Total: 95.3 - 30 + 90.9 + 10 = 166.2 dB effective loss

RIS path loss (N = 400):

Total (from previous calculation, adjusted for 30 m):

FSPL_2 at 30 m NLoS: 92.4 + 28.9 + 20·log10(0.03) + 10 = 100.9 dB

Total: 95.3 + 100.9 - 52.0 = 144.2 dB

RIS advantage: 166.2 - 144.2 = 22.0 dB

`

The passive RIS outperforms the 30 dB NCR by 22 dB in this scenario, while consuming under 1 W vs the NCR's ~10 W. However, the NCR maintains constant gain regardless of distance geometry, whereas RIS gain degrades rapidly as BS-to-RIS distance increases.

Channel Estimation: The Cascaded Challenge

The fundamental signal processing challenge with RIS is cascaded channel estimation. The BS cannot directly estimate the individual BS → RIS and RIS → UE channels because the RIS has no receive chain. Instead, the system must estimate the composite channel H = diag(g) · Φ · h, where Φ is the diagonal RIS phase matrix.

Standard approaches require N pilot transmissions (one per RIS configuration) to estimate all N element channels, creating prohibitive overhead for large surfaces. Research solutions include:

• Codebook-based beam sweeping with hierarchical search (reduces overhead to O(log N))
• Compressive sensing exploiting channel sparsity at mmWave/sub-THz frequencies
• Sparse active elements embedded in the RIS for partial channel sensing (hybrid RIS with 1–5% active elements)
• AI-based estimation using neural networks trained on environment geometry (per TR 38.843 AI/ML for NR air interface)

Real-World Trials and Demonstrations

NTT DOCOMO — 28 GHz RIS Field Trial

NTT DOCOMO conducted outdoor RIS trials in downtown Tokyo using a 28 GHz (n257) base station and a 400-element passive RIS panel mounted on a building facade. The trial, conducted in collaboration with AGC Inc. (glass manufacturer), demonstrated:

• 14 dB SNR improvement in a deep NLoS shadow zone behind a building corner
• Coverage extension to an area previously below -110 dBm RSRP (now -96 dBm)
• RIS panel dimensions: 60 × 60 cm, thickness <5 mm (integrated into building glass)
• Beam reconfiguration time: <50 µs via PIN diode switching

DOCOMO envisions RIS as a cost-effective alternative to mmWave small cells for NLoS coverage, targeting commercial deployment by 2026–2027.

ETRI Korea — RIS-Aided mmWave

The Electronics and Telecommunications Research Institute (ETRI) of South Korea developed a 1,024-element RIS operating at 28 GHz with 2-bit phase quantization. Field measurements in an indoor office environment showed:

• 18 dB received power improvement at locations 15 m from the RIS in NLoS
• Achievable throughput increase from 50 Mbps (without RIS) to 1.2 Gbps (with RIS)
• The system operated entirely passively, with total RIS power consumption of 0.8 W

Huawei MetaAAU Field Trial

Huawei's MetaAAU concept integrates a reconfigurable metasurface lens in front of a standard massive MIMO antenna panel. Deployed in field trials with China Mobile, the system demonstrated:

• 6 dB cell-edge throughput improvement in a 2.6 GHz (n41) macro deployment
• ~30% coverage area extension without additional power consumption
• The metasurface acts as a programmable lens, focusing energy toward cell-edge users

While not a pure "passive RIS" (it augments an active antenna), the MetaAAU represents the nearest-term commercial RIS-adjacent technology.

Deployment Scenarios and Use Cases

The O2I scenario is particularly compelling. Building penetration loss at 28 GHz can exceed 30 dB through modern low-E glass (per TR 38.901 Table 7.4.3-1). A window-mounted RIS operating on the exterior side avoids this loss entirely by reflecting the outdoor signal to an indoor receiver through an alternative path.

Standardization Roadmap

Key Takeaway: Reconfigurable Intelligent Surfaces transform passive building surfaces into programmable reflectors that steer signals toward users in NLoS zones. Passive RIS achieves N² power gain scaling without RF chains, but the cascaded channel estimation challenge and double path loss limit practical gains to scenarios with fully blocked direct paths. Field trials from NTT DOCOMO (14 dB gain at 28 GHz) and ETRI (18 dB gain, 1,024 elements) validate the concept. With 3GPP studying RIS in TR 38.857 and commercial prototypes emerging, expect initial standardization in Release 20 and deployment alongside 6G networks post-2030.