Deep-Learning Estimation of Complex Reverberant Wave Fields with a Programmable Metasurface

Frazier, Benjamin W.; Antonsen, Thomas M.; Anlage, Steven M.; Ott, Edward

doi:10.1103/physrevapplied.17.024027

Cited by 13 publications

(8 citation statements)

References 112 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the other hand, our specific experimental implementation in the microwave domain leveraging programmable metasurfaces relates to a large body of literature: these arrays of meta-atoms with individually reconfigurable scattering properties (usually reflection coefficient) are primarily used for free-space applications such as adaptive beamforming 3 , holography 70 , diffuse scattering 71 , wireless communication 72 – 74 , (intelligent) imaging 66 , 75 – 80 and spatio-temporal wave control 81 . However, they also find increasingly use inside rich-scattering environments 82 as evidenced by various experiments on focusing 83 – 86 , (sub-wavelength) sensing 87 – 89 and transmission matrix engineering 9 , 90 , 91 . Nonetheless, the generality of the discussed wave concepts implies that our idea can also be implemented in tunable acoustic or optical scattering systems 92 – 95 .…”

Section: Introductionmentioning

confidence: 99%

Meta-programmable analog differentiator

2022

View full text Add to dashboard Cite

We present wave-based signal differentiation with unprecedented fidelity and flexibility by purposefully perturbing overmoded random scattering systems such that zeros of their scattering matrices lie exactly at the desired locations on the real frequency axis. Our technique overcomes limitations of hitherto existing approaches based on few-mode systems, both regarding their extreme vulnerability to fabrication inaccuracies or environmental perturbations and their inability to maintain high fidelity under in-situ adaptability. We demonstrate our technique experimentally by placing a programmable metasurface with hundreds of degrees of freedom inside a 3D disordered metallic box. Regarding the integrability of wave processors, such repurposing of existing enclosures is an enticing alternative to fabricating miniaturized devices. Our over-the-air differentiator can process in parallel multiple signals on distinct carriers and maintains high fidelity when reprogrammed to different carriers. We also perform programmable higher-order differentiation. Conceivable applications include segmentation or compression of communication or radar signals and machine vision.

show abstract

Section: Introductionmentioning

confidence: 99%

Meta-programmable analog differentiator

2022

View full text Add to dashboard Cite

show abstract

“…Various implementations of S-ANN (for different sensing tasks and measurement protocols) were reported in Refs. [15]- [20].…”

Section: Operation Principlementioning

confidence: 99%

“…This wave-fingerprinting-based localization technique can be implemented robustly in dynamic environments [18] and it may be possible to learn a shorter task-specific fixed series of RIS configurations to improve latency through "learned sensing" [19]. The related sensing task of object recognition (instead of localization) inside a rich scattering environment has also been implemented with compressed sensing using spectral diversity (broadband measurements) [20].…”

Section: Introductionmentioning

confidence: 99%

A self-adaptive RIS that estimates and shapes fading rich-scattering wireless channels

Saigre-Tardif¹,

Hougne²

2022

Preprint

View full text Add to dashboard Cite

We present a framework for operating a selfadaptive RIS inside a fading rich-scattering wireless environment. We model the rich-scattering wireless channel as being doubleparametrized by (i) the RIS, and (ii) dynamic perturbers (moving objects, etc.). Within each coherence time, first, the self-adaptive RIS estimates the status of the dynamic perturbers (e.g., the perturbers' orientations and locations) based on measurements with an auxiliary wireless channel. Then, second, using a learned surrogate forward model of the mapping from RIS configuration and perturber status to wireless channel, an optimized RIS configuration to achieve a desired functionality is obtained. We demonstrate our technique using a physics-based end-toend model of RIS-parametrized communication with adjustable fading (PhysFad) for the example objective of maximizing the received signal strength indicator. Our results present a route toward convergence of RIS-empowered localization and sensing with RIS-empowered channel shaping beyond the simple case of operation in free space without fading.

show abstract

“…Leaving aside the constraints from Eq. (29b) for a moment, one may be tempted to suspect that by inversing the forward mapping F : c ↦ → H(c), i.e., by formulating the inverse model I : H(c) ↦ → c, the inverse design problem is solved [75]. Unfortunately, inverse problems are generally ill-posed.…”

Section: Taxonomy Of Algorithmic Strategies For Optimizationmentioning

confidence: 99%

RIS-Based Radio Localization in Rich Scattering Environments: Harnessing Multi-Path with ANN Decoders

Hougne

2021

2021 IEEE 22nd International Workshop on Signal Processing Advances in Wireless Communications (SPAWC)

View full text Add to dashboard Cite

Radio localization is a key enabling technology for situational awareness but conventional techniques based on elaborate ray-tracing approaches naturally struggle in rich scattering environments (inside rooms, metro stations, planes, vessels, . . . ). Here, we discuss a completely different approach to radio localization: instead of attempting to understand rich scattering wave propagation in terms of rays, we harness the overwhelming complexity because it assigns unique wave fingerprints to each object position. We interpret wave propagation as a physical encoder of the sought-after localization information in multiplexed measurements and detail artificial neural network (ANN) architectures suitable to decode these measurements for a single or multiple, discrete or continuous, sought-after location variable(s). Capitalizing on recent physics-driven experiments, we clarify that the proposed technique is very robust to measurement noise and capable of achieving deeply sub-wavelength localization precision. The discussed technique can be implemented with multiplexing across spatial, spectral or configurational degrees of freedom, corresponding to sensor networks, broadband measurements and RIS-programmable environments, respectively. Specifically, multiplexing across a fixed random sequence of RIS configurations enables single-frequency localization with a single node. Finally, we propose an end-to-end vision of the technique in which programmable RIS elements take the role of physical weights in a hybrid analog-digital ANN. Thereby, relevant information for the localization task can be discriminated from irrelevant information already in the measurement process, enabling substantial latency improvements.

show abstract

Deep-Learning Estimation of Complex Reverberant Wave Fields with a Programmable Metasurface

Cited by 13 publications

References 112 publications

Meta-programmable analog differentiator

Meta-programmable analog differentiator

A self-adaptive RIS that estimates and shapes fading rich-scattering wireless channels

RIS-Based Radio Localization in Rich Scattering Environments: Harnessing Multi-Path with ANN Decoders

Contact Info

Product

Resources

About