Exploration of Intercell Wireless Millimeter-Wave Communication in the\n Landscape of Intelligent Metasurfaces

Anna C. Tasolamprou, Alexandros Pitilakis, Sergi Abadal, Odysseas Tsilipakos, Xavier Timoneda, Hamidreza Taghvaee, M. S. Mirmoosa, Fu Liu, Christos Liaskos, Ageliki Tsioliaridou, Sotiris Ioannidis, Nikolaos V. Kantartzis, Dionysios Manessis, Julius Georgiou, Albert Cabellos‐Aparicio, Eduard Alarcón, Andreas Pitsillides, Ian F. Akyildiz, Sergei Tretyakov, E. N. Economou, Maria Kafesaki, Costas M. Soukoulis

TL;DR

This paper explores how to turn "smart" metasurfaces into a truly intelligent "AI fabric" by enabling wireless millimeter-wave (mm-wave) communication between their internal controllers. By evaluating three different physical layers within the metasurface stack—the chip layer, a dedicated waveguide, and the metasurface itself—the researchers identify the trade-offs between integration complexity, signal loss, and communication speed.

What Problem Does This Paper Address?

Metasurfaces are thin, artificial structures that can manipulate electromagnetic waves in extraordinary ways, such as creating perfect absorbers or steering beams. To make them "software-defined" or "intelligent," researchers embed thousands of tiny electronic controllers (chips) into the surface to tune its properties in real-time.

However, connecting these thousands of chips presents a massive "routing nightmare." Traditional wired networks-on-chip (NoC) suffer from pin scarcity, high power consumption over distance, and a lack of modularity. Wireless intercell communication is a promising alternative, but the internal environment of a metasurface is electromagnetically complex. This paper addresses the fundamental question: Where and how should these wireless signals travel within the metasurface structure to ensure reliable communication without interfering with the metasurface’s primary function?

Methods

The authors performed a rigorous comparative study using full-wave electromagnetic simulations (via CST Microwave Studio) to model a 5×5 grid of controller nodes. They investigated three distinct "landscapes" for the communication channel:

  1. The Chip Layer: Signals travel through the silicon substrate and the space where the control electronics reside.
  2. A Dedicated Layer: An added parallel-plate waveguide specifically designed for communication, isolating it from the rest of the system.
  3. The Metasurface Layer: Signals travel through the dielectric space between the metasurface patches and the metallic ground plane.

For each channel, they analyzed performance metrics in both the frequency domain (path loss and S-parameters) and the time domain (mean delay, delay spread, and coherence bandwidth) across a frequency range of 25 to 200 GHz.

Datasets

Traditional training datasets were not used in this study. Instead, the "data" consists of high-fidelity electromagnetic simulation results generated by the authors to characterize the physical properties of the proposed communication channels. The benchmarks established are based on standard communication metrics like path loss exponent ($n$) and coherence bandwidth ($B_c$).

Key Insights and Ideas

  • The Silicon Tax: In the chip layer, the lossy nature of silicon leads to high attenuation (40–70 dB). However, the authors discovered that adding a layer of Aluminium Nitride (AlN) acts as a "surface wave enabler," binding the signal to the interface and reducing path loss by approximately 12 dB.
  • Security through Isolation: The dedicated parallel-plate waveguide offers the most stable and secure environment. Because it is enclosed by metallic plates, it prevents signal leakage and protects the internal communication from external interference, though it adds volume to the device.
  • Opportunistic Integration: The metasurface layer itself can be used as a waveguide. While this is the most volume-efficient approach, it requires a delicate "co-design" balance to ensure the communication signals don't leak out or disrupt the metasurface's intended interaction with external microwave signals.
  • The Delay Trade-off: There is a clear inverse relationship between signal strength and clarity. Techniques that reduce path loss (like adding dielectric layers) often increase "multipath richness," leading to higher delay spreads and lower coherence bandwidths.

Key Results

  • Path Loss: The dedicated layer performed best, with path losses ranging from 5–25 dB. The metasurface layer followed with a promising 20–40 dB at 60 GHz. The chip layer was the most challenging, often exceeding 50 dB of loss.
  • Speed and Bandwidth: The chip layer (without enhancements) showed the lowest delay spread (<20 ps), allowing for a coherence bandwidth of over 50 GHz. In contrast, the dedicated layer had a much higher delay spread (~100 ps), limiting its coherence bandwidth to roughly 10 GHz.
  • Feasibility: The study confirms that multi-Gbps wireless intercell communication is physically feasible within the footprint of a standard metasurface unit cell (approx. 12mm pitch).

Limitations and Future Work

The authors acknowledge that implementing these transceivers requires custom, non-standard ASICs capable of operating at mm-wave frequencies while fitting into a tiny area.

Future work will likely focus on the "co-design" of the metasurface layer, where the geometry must be optimized simultaneously for its primary electromagnetic function (e.g., beamsteering) and its secondary role as a communication channel. Additionally, moving from 5×5 grids to much larger, heterogeneous "AI fabrics" will require new protocols to handle the unique interference patterns found in these enclosed environments.

Summary updated

Abstract

Software-defined metasurfaces are electromagnetically ultra-thin, artificial\ncomponents that can provide engineered and externally controllable\nfunctionalities. The control over these functionalities is enabled by the\nmetasurface tunability, which is implemented by embedded electronic circuits\nthat modify locally the surface resistance and reactance. Integrating\ncontrollers within the metasurface cells, able to intercommunicate and\nadaptively reconfigure it, thus imparting a desired electromagnetic operation,\nopens the path towards the creation of an artificially intelligent (AI) fabric\nwhere each unit cell can have its own sensing, programmable computing, and\nactuation facilities. In this work we take a crucial step towards bringing the\nAI metasurface technology to emerging applications, in particular exploring the\nwireless mm-wave intercell communication capabilities in a software-defined\nHyperSurface designed for operation is the microwave regime. We examine three\ndifferent wireless communication channels within the landscape of the\nreflective metasurface: Firstly, in the layer where the control electronics of\nthe HyperSurface lie, secondly inside a dedicated layer enclosed between two\nmetallic plates, and, thirdly, inside the metasurface itself. For each case we\nexamine the physical implementation of the mm-wave transponder nodes, we\nquantify communication channel metrics, and we identify complexity vs.\nperformance trade-offs.\n
BibTeX
@inproceedings{tasolamprou2019exploration,
  author = {Anna C. Tasolamprou and Alexandros Pitilakis and Sergi Abadal and Odysseas Tsilipakos and Xavier Timoneda and Hamidreza Taghvaee and M. S. Mirmoosa and Fu Liu and Christos Liaskos and Ageliki Tsioliaridou and Sotiris Ioannidis and Nikolaos V. Kantartzis and Dionysios Manessis and Julius Georgiou and Albert Cabellos‐Aparicio and Eduard Alarcón and Andreas Pitsillides and Ian F. Akyildiz and Sergei Tretyakov and E. N. Economou and Maria Kafesaki and Costas M. Soukoulis},
  title = {{Exploration of Intercell Wireless Millimeter-Wave Communication in the\textbackslash\{\}n Landscape of Intelligent Metasurfaces}},
  booktitle = {IEEE Access},
  year = {2019},
  url = {https://openreview.net/forum?id=Jip00kxizu},
}

Loading related content...

Figure