⚡ electrical engineering

Z-Opt: A Near-Optimal Reduced-Complexity Two-Dimensional Grassmannian Constellation

This paper introduces two efficient construction methods (S-Opt and Z-Opt) and corresponding low-complexity detectors for two-dimensional Grassmannian constellations that operate on the Bloch sphere, achieving near-optimal performance with linear detection complexity while approaching theoretical packing bounds.

Original authors: Kotaro Shigenaga, Hiroki Iimori, Yuto Hama, Chandan Pradhan, Szabolcs Malomsoky, Naoki Ishikawa

Published 2026-05-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Kotaro Shigenaga, Hiroki Iimori, Yuto Hama, Chandan Pradhan, Szabolcs Malomsoky, Naoki Ishikawa

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 secret message using a flashlight in a thick fog. The fog represents a wireless channel that changes rapidly and unpredictably. In the old days, you would have to send a "test signal" (like a pilot) to tell the receiver what the fog looks like right now, so they could decode your message. But this takes up a lot of time and space, making your communication slower and less efficient.

This paper proposes a smarter way: Noncoherent Communication. Instead of asking the receiver to measure the fog, you send your message in a way that the receiver can understand without knowing the fog's current state.

Here is how the authors solved the problem of designing these "fog-proof" messages, explained simply:

The Core Idea: The Bloch Sphere

To make these messages, the authors use a mathematical shape called the Grassmann Manifold. That sounds scary, but the paper reveals a secret: for their specific setup, this complex shape is exactly the same as a globe (a sphere). In the world of quantum computing, this globe is called the Bloch Sphere.

Think of your message not as a word, but as a dot on the surface of a globe.

To send a message, you pick a specific dot.
To receive the message, the receiver looks at where the signal landed on the globe and guesses which dot you picked.
The Rule: To make the message hard to confuse, the dots need to be as far apart from each other as possible. If two dots are too close, the "fog" might make the receiver think you sent Dot A when you actually sent Dot B.

The Problem

Mathematicians have known for a long time how to pack dots on a sphere so they are far apart (this is called the Tammes problem). However, finding the perfect arrangement for a huge number of dots is incredibly hard for computers, and checking if a receiver picked the right dot usually requires comparing the signal against every single possible dot, which is slow and energy-intensive.

The Solution: Two New Methods

The authors propose two new ways to arrange these dots and two new ways to find them.

1. S-Opt: The "Perfect Packing" Method

The Analogy: Imagine you have a library of pre-solved puzzles. Mathematicians have already figured out the best ways to pack dots on a sphere for small numbers. S-Opt simply uses these pre-solved, perfect arrangements.
How it works: They take these known, perfect dot patterns and map them to their communication system.
The Benefit: Because the dots are perfectly spaced, the message is as clear as it can possibly be.
The Detector (The Finder): Usually, finding the right dot requires checking every single one. But because the dots are arranged in a specific mathematical way, the authors created a "smart map" (using something called a KD-tree). Instead of checking every dot, the receiver can zoom in on the right area quickly. It's like using a GPS to find a house instead of knocking on every door in the city.
- Speed: Very fast. It scales well even if you add more antennas.

2. Z-Opt: The "Stacked Polygons" Method

The Analogy: Imagine building a tower of rings. You start with a small ring at the top, a bigger ring in the middle, and another small ring at the bottom. You rotate each ring slightly so the dots don't line up perfectly. This creates a beautiful, structured pattern.
How it works: Instead of trying to solve the whole puzzle at once, they build the constellation layer by layer. They only need to tweak the "height" of these rings to make the dots as far apart as possible.
The Benefit: This method is much easier to design than the perfect packing method. It gets almost as close to the "perfect" distance as possible, but with much less math required to build it.
The Detector (The Finder): This is the real magic. Because the dots are arranged in neat, stacked rings, the receiver doesn't need to search the whole globe.
- They just look at the "latitude" (how high up the ring is) and the "longitude" (where on the ring).
- This narrows the search down to just four possible dots.
- Speed: Extremely fast. It doesn't matter how many dots you have; the search time stays the same.
- Memory: The receiver doesn't need to store the location of every single dot. It only needs to remember the "heights" of the rings. This saves a massive amount of memory.

The Results

The authors tested these methods against existing techniques:

S-Opt achieves the theoretical maximum clarity (the dots are as far apart as math allows).
Z-Opt gets very close to that maximum clarity but is much cheaper to build.
Both detectors are just as accurate as the "brute force" method (checking every single dot), but they are much faster and use less memory.

Summary

In short, this paper is about designing a better alphabet for wireless communication in fast-changing environments.

They realized the alphabet lives on a globe.
S-Opt uses the best-known globe patterns for maximum clarity.
Z-Opt builds the alphabet using stacked rings for a balance of clarity and simplicity.
Both come with super-fast search tools that let the receiver find the message instantly without needing a massive computer or a huge memory bank.

This allows for faster, more efficient wireless communication, especially for things like high-speed trains or drones where the connection changes constantly.

Technical Summary: Z-Opt: A Near-Optimal Reduced-Complexity Two-Dimensional Grassmannian Constellation

Problem Statement
High-mobility wireless communications present significant challenges for conventional coherent multiple-input multiple-output (MIMO) systems, as rapid channel variations necessitate a high proportion of pilot symbols for channel estimation, thereby reducing spectral efficiency. Noncoherent communication, which eliminates the need for channel estimation, offers a promising alternative. In this regime, Grassmannian signaling is particularly effective, where codewords span distinct subspaces invariant under channel multiplication. However, the efficient construction of Grassmannian constellations remains difficult. While numerical optimization methods (Man-Opt) can achieve optimal minimum chordal distances, they lack structural regularity, requiring exhaustive search detection with high complexity. Conversely, structured construction methods (e.g., Exp-Map, Cube-Split, Grass-Lattice) offer lower complexity but often fail to match the performance of optimized constellations. Furthermore, existing detectors for Grassmannian constellations often scale linearly with the constellation size $C$ , creating a bottleneck for large constellations.

Methodology
The paper focuses on the case of $M=1$ transmit antenna and $T=2$ orthogonal time-frequency resources, corresponding to one-dimensional subspaces in a two-dimensional space, denoted as the Grassmann manifold $G(2, 1)$ . The authors leverage the geometric isomorphism between $G(2, 1)$ and the surface of a unit sphere, known as the Bloch sphere in quantum computing.

Geometric Insight: The authors establish that the chordal distance $d_c$ on the Grassmann manifold is directly proportional to the Euclidean distance $d_E$ on the Bloch sphere, specifically $d_E = 2d_c$ . This relationship allows the problem of maximizing the minimum chordal distance to be mapped to the classical Tammes problem (sphere packing) on the Bloch sphere.
Theoretical Bound: By utilizing the Fejes–Tóth bound on the minimum Euclidean distance for sphere packings, the authors derive a corresponding theoretical upper bound for the minimum chordal distance on $G(2, 1)$ .
S-Opt Constellation and Detector:
- Construction: S-Opt is constructed by mapping optimally or near-optimally arranged points from the Bloch sphere (solutions to the Tammes problem) directly to Grassmannian codewords. This method attains the derived theoretical upper bound for optimal configurations.
- Detection: A low-complexity detector is proposed that maps the received signal to the Bloch sphere and performs nearest-neighbor detection using a KD-tree algorithm. This approach is applicable to any Grassmannian constellation on $G(2, 1)$ .
Z-Opt Constellation and Detector:
- Construction: Inspired by the structure of S-Opt, Z-Opt is constructed by stacking regular polygons at different heights (latitudes) on the Bloch sphere. The optimization variables are reduced to the vertical positions ( $\theta$ ) of these layers, requiring only $O(\sqrt{C})$ variables. The structure exploits symmetry to minimize the number of distance metric evaluations during optimization.
- Detection: A specialized low-complexity detector utilizes the structured "structure diagram" of the constellation. By determining the grid region of the received signal on the Bloch sphere, the detector narrows the candidate codewords to at most four, eliminating the need to store the entire constellation.

Key Contributions

S-Opt Construction and Universal Detector: The paper proposes S-Opt, which achieves the theoretical upper bound for minimum chordal distance when optimal sphere packings are used. It introduces a universal low-complexity detector for $G(2, 1)$ with time complexity $O(N + \log_2 C)$ (where $N$ is the number of receive antennas) that yields the same performance as the Generalized Likelihood Ratio Test (GLRT) detector.
Z-Opt Construction and Specialized Detector: The authors propose Z-Opt, a structured constellation that approaches the theoretical upper bound using only $O(\sqrt{C})$ optimization variables. A corresponding detector is developed with time complexity $O(N)$ and space complexity $O(\sqrt{C})$ , independent of the constellation size $C$ .
Theoretical Derivation: The paper derives a tight theoretical upper bound on the minimum chordal distance for $G(2, 1)$ based on the Fejes–Tóth bound, grounded in the proportionality between chordal and Euclidean distances on the Bloch sphere.

Results

Performance: Simulation results indicate that S-Opt matches the derived Fejes–Tóth upper bound for evaluated constellation sizes. Z-Opt approaches this bound closely. In terms of Symbol Error Rate (SER), S-Opt and the conventional Man-Opt method (which has the largest minimum chordal distance) exhibit the lowest SERs, followed by Z-Opt.
Complexity:
- GLRT: Time complexity $O(NC)$, Space complexity $O(C)$ .
- S-Opt Detector: Time complexity $O(N + \log_2 C)$ , Space complexity $O(C)$ .
- Z-Opt Detector: Time complexity $O(N)$ , Space complexity $O(\sqrt{C})$ .
Construction Cost: The number of distance metric evaluations for Z-Opt is $O(\sqrt{C})$ , significantly lower than the $O(C^2)$ required for Man-Opt and Exp-Map. Similarly, the number of optimization variables for Z-Opt is $O(\sqrt{C})$ , compared to $O(C)$ for other methods.
Detection Accuracy: The proposed low-complexity detectors (S-Opt and Z-Opt) achieve the same detection accuracy as the exhaustive GLRT detector for their respective constellations.

Significance
The paper claims that by exploiting the Bloch sphere representation, it is possible to bridge the gap between the performance of numerically optimized constellations and the complexity of structured constellations. The S-Opt approach provides a benchmark for optimal performance on $G(2, 1)$ , while the Z-Opt approach offers a practical, near-optimal solution with drastically reduced construction and detection complexity. The proposed detectors enable efficient noncoherent communication for high-mobility scenarios without the spectral efficiency penalty of pilot symbols or the computational burden of exhaustive search, particularly for large constellation sizes. The authors note that future work will focus on extending these construction methods and detectors to general values of $M$ and $T$ .