In the push toward 6G wireless networks, Reconfigurable Intelligent Surfaces (RIS) have emerged as a powerful tool to control the wireless environment. By dynamically shifting the phase of incoming electromagnetic waves, these surfaces can bypass obstacles and steer signals directly toward a receiver.
However, traditional RIS architectures are entirely passive. This passivity introduces a severe "blindness" problem: the surface cannot process or perceive the signals it reflects. This creates two massive bottlenecks:
- The Channel Estimation Headache: To route signals accurately, the network needs to know the exact channel state between the base station, the surface, and the user. For a passive surface with hundreds of elements, estimating this channel requires an overwhelming amount of training overhead.
- The Four-Hop Path Loss Penalty: In Integrated Sensing and Communication (ISAC) systems—where wireless signals double as radar to detect physical objects—a signal must travel from the base station to the surface, out to the target, back to the surface, and all the way back to the base station. This multi-hop journey causes catastrophic signal attenuation.
To solve these challenges, a new paradigm shifts the architecture from purely passive arrays to semi-passive, surface-integrated sensing systems. By embedding a small fraction of active sensors or power detectors directly into the surface substrate, researchers have found a way to short-circuit path loss and achieve low-overhead environmental awareness.
Short-Circuiting Path Loss: From Four Hops to Two
The primary reason radar sensing struggles with passive surfaces is the physics of geometric path loss. In a conventional setup, the radar echo must make a long round-trip back to a central base station. Because signal power attenuates rapidly over distance, forcing the echo to travel those final hops back to the base station severely degrades the Signal-to-Interference-Plus-Noise Ratio (SINR).
Traditional Passive Sensing (4-Hop Path):
Base Station ──> Surface ──> Target ──> Surface ──> Base Station (Severe Attenuation)
Surface-Integrated Sensing (2-Hop Path):
Base Station ──> Target ──> Surface-Integrated Sensor (Immediate Capture)
Integrating sensing hardware directly onto the surface alters this geometry. As demonstrated by Zhaolin Wang et al. and Dongsheng Han et al., placing sensors on the surface collapses the traditional four-hop radar link into a two-hop link. The surface intercepts and processes the target's echo immediately at the point of reflection, bypassing the return trip to the base station entirely.
This proximal echo reception drastically reduces signal attenuation. Work by Na Xue et al. and Yue Xiu et al. confirms that this architectural shift boosts the signal strength of detected echoes, allowing the system to pinpoint distant targets with high accuracy while maintaining the energy efficiency of a mostly passive relay.
Dimension-Independent Channel Estimation
For a passive surface to optimize its reflection phase shifts, it normally relies on the base station to calculate the Channel State Information (CSI). As surfaces scale up to include thousands of elements, the time and bandwidth required to estimate every single element's channel explode exponentially.
Hybrid sensing architectures solve this by giving the surface "eyes" of its own. Xiaojun Yuan et al. and Shicong Liu et al. showed that embedding a limited number of active elements equipped with radio frequency (RF) chains allows the surface to directly sample the electromagnetic environment. This local sampling provides the necessary observability to run compressive sensing algorithms, reconstructing the full channel matrix with minimal training overhead.
Taking this intuition a step further, Jieao Zhu et al. proved that the surface does not even need expensive, power-hungry RF chains to estimate the channel. By integrating low-cost power detectors into individual reflective elements, the surface can measure local signal intensities directly. This physics-inspired approach allows the network to acquire the necessary beamforming information with a constant overhead, completely independent of how large the surface becomes.
Distributed Arrays and 360-Degree Awareness
By distributing these semi-passive elements across the surface, the hardware effectively transforms into a distributed sensor array. This spatial diversity unlocks high-precision localization and full-space coverage.
- Millimeter-Level Localization: Xiaoling Hu et al. and Zhouyuan Yu et al. designed frameworks where a small fraction of sensing elements capture uplink communication signals. Because these elements are spread out, they capture spatial information from multiple distinct viewpoints simultaneously, enabling millimeter-level positioning accuracy without requiring dedicated positioning reference signals.
- 360-Degree Coverage: Standard surfaces only handle reflections on one side. To achieve full-space awareness, Xidong Mu et al. introduced Simultaneously Transmitting and Reflecting Surfaces (STARS). By mounting active sensing components directly onto a substrate that both transmits and reflects, the system prevents echoes from different spatial regions from blending together, enabling true 360-degree environmental mapping.
Balancing the Active-Passive Trade-off
The core engineering challenge in surface-integrated sensing is a classic resource trade-off: How many sensors do we actually need? Active sensors require power and hardware complexity, while passive elements are cheap but blind.
Yuanwei Liu et al. formalized this fundamental design trade-off, mapping out the mathematical balance between deploying more passive reflecting elements (for communication gain) versus active micro-sensors (for sensing resolution).
The consensus across the literature is that you only need a tiny fraction of active elements to reap massive rewards. For instance, Bo Yang et al. demonstrated that integrating just a small subset of semi-active elements allows a surface to perform high-accuracy spectrum sensing and user identification via analog-to-digital conversion. This minimal hardware footprint is enough to power decentralized edge intelligence, such as Federated Spectrum Learning, without sacrificing the low-cost, low-power appeal of metasurfaces.
Furthermore, when multiple semi-passive surfaces are deployed cooperatively, their collective power scales dramatically. Wen Chen et al. showed that a collaborative network of semi-passive surfaces can jointly optimize active and passive beamforming to bypass severe line-of-sight blockages, outperforming standard benchmarks by roughly 15 dB in localization accuracy.
By moving the sensing receiver directly to the point of reflection, surface-integrated sensing fundamentally rewrites the link budget of next-generation wireless networks, turning passive walls into intelligent, self-sensing nodes.
Key Foundational Work
IRS-Based Integrated Location Sensing and Communication for mmWave SIMO Systems (2022) — contributed 2 concept elements to this direction.
STARS Enabled Integrated Sensing and Communications (2022) — contributed 2 concept elements to this direction.
Federated Spectrum Learning for Reconfigurable Intelligent Surfaces-Aided Wireless Edge Networks (2022) — contributed 2 concept elements to this direction.