⚡ electrical engineering

Trade-Offs in Decentralized Gigantic MIMO with Hard-Boundary Constraints

This paper proposes a novel adaptation of the WAX framework for decentralized gigantic MIMO systems operating in the FR3 band, demonstrating its ability to effectively manage trade-offs between processing complexity and performance under hard-boundary constraints using non-cooperating hardware modules.

Original authors: Juan Vidal Alegría, Joao Vieira, Ove Edfors

Published 2026-07-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Juan Vidal Alegría, Joao Vieira, Ove Edfors

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

The Big Picture: Building a "Gigantic" Brain for the Next Internet

Imagine the current mobile network (5G) as a highly efficient team of 100 chefs (antennas) working in a kitchen to serve a few hungry customers (users). This setup, called "Massive MIMO," is great at feeding everyone quickly and clearly.

But the next generation of internet (6G) wants to operate on higher frequencies. The problem? Signals at these high frequencies are like whispers in a windy storm; they get lost easily. To make the whisper heard, the kitchen needs to get much bigger. Instead of 100 chefs, we might need 400 to 800 chefs. This new, massive setup is called "Gigantic MIMO."

The Problem: The "Too Big to Cook" Dilemma

If you suddenly hire 800 chefs, you run into two major headaches:

Communication Chaos: If every chef tries to shout their recipe to the Head Chef (the central computer) at the same time, the kitchen becomes a noisy mess. The wires connecting them can't handle that much data.
Processing Overload: The Head Chef can't possibly listen to 800 voices and calculate the perfect meal plan instantly. It's too much work.

The paper asks: How do we organize this gigantic kitchen so it works efficiently without everyone talking to everyone else?

The Solution: The "WAX" Framework with Hard Walls

The authors propose a new way to organize the kitchen using a framework called WAX. Think of WAX as a set of rules for how to chop, mix, and plate the food before it reaches the Head Chef.

In previous versions of this idea, the chefs were allowed to talk to each other to figure out the best way to cook. But in the real world, you can't always have chefs from different stations chatting freely; they might be in different buildings or run by different companies.

This paper introduces a new rule: "Hard Boundaries."

The Analogy: Imagine the kitchen is divided into four separate rooms. Inside each room, the chefs can talk to each other freely. But there are soundproof walls between the rooms. Chefs in Room A cannot talk to chefs in Room B. They must solve their part of the puzzle completely on their own before sending their finished plates to the Head Chef.

How It Works: The Assembly Line

The authors figured out a clever way to arrange the math so that these isolated rooms can still work together perfectly without ever speaking to each other.

The Rooms (Decentralized Modules): The 800 antennas are split into groups (e.g., 4 rooms of 200 antennas each). Each room has its own mini-computer (a standard "Baseband Unit" or BBU) that processes the data from its own antennas.
The Hard Wall: These mini-computers do not share data with each other. They only process what they see locally.
The Head Chef (Central Unit): Once the rooms have done their local work, they send a simplified summary to the Head Chef. The Head Chef then does the final, easy calculation to serve the users.

The Results: Doing More with Less

The paper ran simulations to see if this "Hard Wall" idea actually works. Here is what they found:

It Works Perfectly: Even with the soundproof walls, the system can achieve the same high-quality results as if everyone were talking to everyone else. They call this "information-lossless."
Less Work for the Chefs: By using this method, the local computers (the mini-chefs) have to do significantly less math.
- The Analogy: In the old way, every chef might have to do 8 complex calculations. With this new method, they might only need to do 4 or even 2, while still getting the same delicious meal.
Using Existing Hardware: The best part is that these "rooms" can be built using the exact same equipment already used in 5G kitchens. You don't need to invent a new type of computer; you just need to arrange the existing ones differently.

The Trade-Off

The paper shows a "trade-off" (a give-and-take):

If you want perfect performance, the local computers still need to do a little bit of math (about 4 to 8 calculations per antenna).
If you want to save even more energy, you can reduce the math even further. You will lose a tiny bit of performance (maybe 10% less capacity), but you cut the work in half.

Summary

This paper proves that we can build the massive, 6G-ready networks of the future by simply taking existing 5G equipment, grouping them into isolated teams, and using a clever mathematical trick (the WAX framework) to make them work together without needing to chat. It solves the problem of "too many antennas" by turning one giant, overwhelmed brain into a team of independent, efficient specialists.

Technical Summary: Trade-Offs in Decentralized Gigantic MIMO with Hard-Boundary Constraints

Problem Statement
The transition from 5G to 6G involves moving from the sub-6GHz band (FR1) to the upper-mid band (FR3, 7–24 GHz). To maintain comparable antenna apertures and compensate for increased propagation loss at these higher frequencies, the number of antennas at the base station (BS) must increase by a factor of 4–8, leading to the paradigm of "gigantic MIMO." This scaling introduces significant challenges regarding processing complexity and interconnection bandwidth. While the WAX framework has previously been proposed to explore trade-offs between decentralization and complexity, existing literature has not established its applicability under "hard-boundary constraints." Specifically, previous models often assumed a level of cooperation among decentralized modules that is impractical when utilizing existing, non-cooperating hardware units like massive MIMO baseband units (BBUs). The core problem addressed is how to implement gigantic MIMO using decentralized processing where modules cannot share information or cooperate to compute filters, yet still achieve information-lossless performance.

Methodology
The authors propose a novel adaptation of the WAX decomposition framework tailored for systems with hard boundaries between decentralized modules. The system model considers an uplink multi-user MIMO scenario where $K$ users communicate with an $M$ -antenna BS ( $M \gg K$ ). The received signal is processed through a combination of decentralized linear filters ( $W$ ), a combining matrix ( $A$ ), and a central linear equalizer ( $X$ ).

Key methodological steps include:

Hard-Boundary Formulation: The authors impose a strict constraint where decentralized modules are non-cooperating. Each module $p$ must independently compute its assigned set of filters using only local channel state information (CSI) associated with its specific antennas.
Permutation and Re-indexing: To satisfy the hard-boundary constraint, the authors introduce a permutation function $\pi(\cdot)$ . This re-indexes the decentralized filters and the channel matrix blocks such that the mathematical conditions for information-lossless processing (derived from the WAX decomposition) can be satisfied within mutually exclusive sets of filters.
Structured Combining Module: A specific structure for the combining matrix $A$ is designed. By enforcing a block-diagonal structure on the sub-matrix $\tilde{A}_B$ (part of the combining module definition), the global optimization problem is decomposed into $P$ independent equations. Each equation corresponds to a single decentralized module and involves only the local channel blocks and local filters.
Complexity-Performance Trade-off: The framework allows for an arbitrary trade-off between the level of decentralization (number of inputs to the central processing unit, $T$ ) and the decentralized processing complexity (multiplications per antenna, $L$ ). The authors derive a new condition for information-lossless processing under these constraints: $L \geq K \frac{MP - TP}{MP - P}$ .

Key Contributions

Adaptation of WAX for Hard Boundaries: The paper extends the WAX framework to scenarios where decentralized modules cannot cooperate, a critical requirement for utilizing existing massive MIMO BBUs in gigantic MIMO deployments.
Decomposition Strategy: The authors demonstrate how to design a combining module and a filter permutation strategy that allows each decentralized module to solve its part of the WAX decomposition independently, without exchanging information with other modules.
Scalability and Backward Compatibility: The proposed architecture is explicitly designed to be scalable. It suggests that a gigantic MIMO system can be constructed by connecting multiple existing massive MIMO BBUs (each handling a subset of antennas) to a central unit, potentially implemented by another BBU, without requiring a complete redesign of the hardware.
Theoretical Trade-off Characterization: The work provides a theoretical bound (Equation 16) for the minimum processing complexity ( $L$ ) required per antenna to achieve information-lossless processing given a specific number of decentralized modules ( $P$ ) and central inputs ( $T$ ).

Results
Numerical simulations were conducted using a system with $M=256$ antennas divided into $P=4$ decentralized modules (each handling 64 antennas), with $T=64$ inputs to the CPU.

Performance: The proposed framework achieves information-lossless performance (matching the capacity of Maximum Ratio Combining, MRC) with significantly reduced complexity. For $K=5$ users, lossless performance was achieved with $L=8$ multiplications per antenna; for $K=9$ users, it was achieved with $L=4$ .
Complexity Reduction: Compared to a decentralized MRC baseline, the proposed method reduces the number of MAC operations per module. For instance, with $K=9$ , the baseline required 576 operations, whereas the proposed method achieved lossless performance with 256 operations ( $L=4$ ).
Flexibility: The results indicate that the framework allows for further reductions in processing complexity (e.g., $L=2$ or $L=1$ ) at the cost of minor performance degradation, achieving approximately 90% of the channel capacity with half the decentralized processing complexity.

Significance and Claims
The paper claims that this work facilitates the transition from massive MIMO to gigantic MIMO by enabling the use of state-of-the-art massive MIMO BBUs as the building blocks for larger systems. By proving that information-lossless processing is possible even with hard boundaries between non-cooperating modules, the authors argue that their approach offers a practical path for scalability and backward compatibility in 6G base station design. The framework is presented as a tool to exploit trade-offs between complexity and performance in practical implementations, rather than a theoretical ideal. The authors note that while the current work focuses on the baseband domain, the framework could potentially be adapted for analog domain processing in future work.