🔬 optics

OAM Light Demultiplexing from an Intensity Profile using Orthogonality Renormalization of Pair Modes

This paper proposes a novel demultiplexing method for Orbital Angular Momentum (OAM) light that utilizes orthogonality renormalization of pair modes to extract channel information from a single intensity profile, thereby enabling efficient OAM communication without the need for additional receiver-side optical elements.

Original authors: Junsu Kim, Hyunchae Chun, SeungRyong Park

Published 2026-05-22

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Junsu Kim, Hyunchae Chun, SeungRyong Park

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to send a massive amount of data using a single beam of light. In the world of advanced physics, this light can be "twisted" like a corkscrew. This twisting is called Orbital Angular Momentum (OAM). Think of each twist as a different "lane" on a highway. If you can twist the light in many different ways (different lane numbers), you can send many messages at once, massively speeding up your internet connection.

However, there is a big problem: How do you read the message?

The Problem: The "Blind" Camera

In the past, to separate these twisted lanes and read the messages, scientists needed complex, expensive machinery. They used special mirrors, gratings, and filters to physically untangle the light before a camera could see it. It was like trying to sort a pile of mixed-up colored marbles by running them through a complicated machine with spinning gears. If the machine wasn't perfect, or if the air was shaky (like wind blowing the light), the sorting would fail, and you'd lose your data.

Furthermore, cameras can only see brightness (intensity), not the invisible "twist" (phase) of the light. When you mix two different twists together, the camera just sees a blurry, overlapping pattern of light and dark spots. The special "lane" information seems to disappear because the camera can't distinguish the individual lanes in that single, messy picture.

The Solution: The "Subtraction" Trick

The authors of this paper, Junsu Kim, Hyunchae Chun, and Seungryong Park, came up with a clever, software-based trick to solve this without needing any extra hardware.

Here is their analogy:
Imagine you have a group of friends standing in a circle, each holding a flashlight.

Friend A flashes their light in a pattern that goes "Bright-Dark-Bright-Dark" twice around the circle.
Friend B flashes their light in a pattern that goes "Bright-Dark-Bright-Dark" four times around the circle.

If they flash at the same time, the camera sees a messy mix. But, the authors realized something special: if you pair up friends with opposite twists (one twisting left, one twisting right), their combined light creates a specific, predictable pattern of bright and dark spots.

The Magic Step:
The researchers proposed a simple math trick called "Orthogonality Renormalization."

Take a Photo: The camera takes a picture of the messy, combined light.
Find the Average Brightness: They calculate the "average" brightness of the whole picture.
Subtract the Average: They subtract this average brightness from the picture.

Why this works:
Think of the light pattern as a wave going up and down. The "average" is the flat line in the middle. By subtracting that middle line, the waves that were previously "all positive" (just bright spots) now swing both up (positive) and down (negative).

Once you do this subtraction, the different "lanes" (the different twists) become mathematically invisible to each other. They stop interfering. It's like tuning a radio: once you subtract the static, the different stations become perfectly clear and distinct.

The Result: A Single Snapshot

Now, instead of needing a complex machine to untangle the light, the computer just takes that single photo, does the subtraction, and runs a standard mathematical tool called a Fourier Transform (which is like a super-fast pattern recognizer).

The Output: The computer instantly sees distinct peaks. Each peak corresponds to a specific "lane" (a specific message).
The Benefit: You can read 128 different messages simultaneously from just one single image, without any extra lenses or mirrors.

Real-World Performance (The Simulation)

The team tested this idea using computer simulations:

Perfect Conditions: Even with 128 channels running at once, the system worked perfectly.
Noisy Conditions: They added "static" (simulating noise or atmospheric interference) to the image. Even with 2% to 5% noise, the system could still read the messages correctly with almost zero errors.
Radial Distribution: They also tested what happens when the light spreads out in a circle (like a donut). By looking at the entire donut shape rather than just a thin slice, the system became even better at ignoring noise.

The Bottom Line

This paper proposes a way to untangle complex, twisted light signals using only a camera and a simple math trick.

No extra hardware: No need for expensive gratings or special mirrors.
Fast: It processes the data in a single step (subtraction + pattern recognition).
Robust: It handles noise well, especially if you look at the whole light beam rather than just a slice of it.

It's like turning a chaotic, noisy crowd of people shouting into a clear conversation just by asking everyone to lower their voices by the same amount and then listening for the specific rhythm of their speech. The paper claims this makes high-speed optical communication much simpler, cheaper, and more reliable.

Technical Summary: OAM Light Demultiplexing from an Intensity Profile using Orthogonality Renormalization of Pair Modes

Problem Statement
Orbital Angular Momentum (OAM) of light offers a promising degree of freedom for next-generation optical communication, enabling multi-channel division through the orthogonality of distinct OAM modes. However, a fundamental challenge arises in the demultiplexing (demux) of superposition states. While OAM modes are orthogonal in the electric field domain, this orthogonality is lost when measuring light intensity ( $|E|^2$ ), as the intensity profiles of different modes overlap and are not mutually orthogonal. Consequently, existing demultiplexing methods rely on spatial-domain processing using additional optical elements such as diffraction gratings, meta-surfaces, refractive modulators, or mode sorters. These approaches introduce structural complexity, potential signal loss, and alignment constraints. Furthermore, computational methods utilizing neural networks or template correlation require reference libraries or trained models, limiting their analytical transparency and scalability.

Methodology
The authors propose a demultiplexing method that operates entirely within a single intensity profile, eliminating the need for additional receiver-side optical structures. The core of the methodology is Orthogonality Renormalization applied to OAM pair modes.

Pair Mode Definition: The system utilizes communication channels based on OAM pair states $|p\rangle$ , defined as a superposition of topological charges with equal magnitude but opposite sign: $|p\rangle \equiv \frac{1}{\sqrt{2}}(|\ell\rangle + |-\ell\rangle)$ , where $p = |\ell|$ . The electric field of such a pair is proportional to $\cos(p\phi)$ , resulting in an intensity profile proportional to $\cos^2(p\phi)$ .
Renormalization: Since $\cos^2(p\phi)$ $cos^{2} (pϕ)$ modes are not orthogonal, the authors introduce a renormalization step. By subtracting half of the maximum (or mean) intensity from the detected superposed profile, the basis function is transformed from $\cos^2(p\phi)$ $cos^{2} (pϕ)$ to $\cos(2p\phi)$ $cos (2 pϕ)$ .
- Mathematically: $I^{(R)}_p \propto \cos^2(p\phi) - 0.5 \propto \cos(2p\phi)$ .
- This transformation restores mutual orthogonality among different pair channels ( $p \neq p'$ ) in the intensity domain, as $\int \cos(2p\phi)\cos(2p'\phi)d\phi = 0$ .
Demultiplexing via Fourier Transform: The renormalized intensity profile, representing a sum of orthogonal oscillation functions, is processed using a single Fourier Transform (FT). This allows for the parallel recovery of the continuous amplitudes ( $A_p$ ) of all multiplexed channels.
Radial Integration (Optional): To address the radial distribution of Laguerre-Gaussian (LG) beams, the method can be extended by integrating the intensity over the radial coordinate. This averages noise over a larger area and extracts state information in a single operation, improving the Signal-to-Noise Ratio (SNR) compared to single-radius scanning.

Key Contributions

Hardware Minimization: The method enables OAM demultiplexing using only a standard intensity detector (e.g., CCD or CMOS) without spatial light modulators (SLMs), gratings, or mode sorters.
Analytical Transparency: Unlike machine learning approaches, the demultiplexing process is a deterministic analytical procedure (subtraction followed by FT), removing the need for trained models or reference templates.
Orthogonality Restoration: The paper demonstrates that orthogonality can be mathematically restored in the intensity domain through renormalization, allowing for the simultaneous separation of multiple amplitude-modulated channels from a single intensity frame.
Robustness to Phase Errors: The FT magnitude used for amplitude recovery is shown to be independent of arbitrary phase offsets between conjugate OAM components, provided the phase is stable within a single frame acquisition.

Results
Numerical simulations were conducted to validate the method under various conditions, including Perfect Vortex (PV) scenarios (no radial dependence) and scenarios considering radial distribution.

Demultiplexing Performance: In a system with 128 multiplexed pair channels using 2-Pulse Amplitude Modulation (2-PAM), the method achieved error-free decoding (Bit Error Rate, BER = 0) under ideal conditions and with 2% white noise.
Radial Integration Benefits: When considering radial distribution, the integration method demonstrated superior noise tolerance. At 128 modes with 5% noise, the radial integration method achieved a BER of approximately 8.9%, whereas the single-radius PV method yielded a BER of approximately 18%.
Scalability and Limits: The study established quantitative operating boundaries. The maximum number of modes is limited by the azimuthal sampling resolution (Nyquist criterion), allowing for up to ~785 modes in the simulated geometry. However, increasing the modulation order (e.g., from 2-PAM to 8-PAM) or the number of modes significantly increases the BER, particularly in high-noise environments.
Trade-offs: While the method scales linearly with channel count, the effective channel count is reduced by a factor of two compared to methods addressing individual $\ell$ states, as each channel combines two opposite-sign states.

Significance and Claims
The paper claims to present a "distinct paradigm of multi-channel intensity domain communication." Its primary significance lies in simplifying the receiver architecture for OAM-based systems. By shifting the complexity from optical hardware to a straightforward analytical post-processing step, the method offers a practical solution for scalable optical communication. The authors emphasize that this approach allows the full incident photon flux to reach the detector without diffractive or absorptive losses associated with additional optical elements.

The work is modest in its scope, noting that the analysis is conducted entirely through numerical simulation with idealized beam models and additive white noise. The authors explicitly state that effects such as atmospheric turbulence, inter-modal crosstalk, and imperfect beam generation are beyond the current study's scope and remain subjects for future experimental investigation. The proposed scheme is positioned as a hardware-minimized and analytically transparent framework that facilitates the practical implementation of scalable OAM communication systems.

The Problem: The "Blind" Camera

The Solution: The "Subtraction" Trick

The Result: A Single Snapshot

Real-World Performance (The Simulation)

The Bottom Line

More like this