MIMO Systems Aided by Microwave Linear Analog Computers: Capacity-Achieving Architectures with Reduced Circuit Complexity

1. Problem Statement

The evolution of wireless networks toward 6G requires Gigantic MIMO systems, involving antenna arrays with thousands of elements. While fully digital MIMO offers maximum performance, it is impractical at this scale due to prohibitive hardware costs, power consumption, and computational complexity (requiring one RF chain per antenna).

Microwave Linear Analog Computers (MiLACs) have emerged as a promising solution, enabling fully analog beamforming via reconfigurable microwave networks. However, existing MiLAC designs are fully-connected, meaning every port is interconnected with every other port via tunable impedance components.

• The Bottleneck: The circuit complexity (number of required tunable components) of fully-connected MiLACs scales quadratically ($O(N^2)$) with the number of antennas.
• The Goal: The authors aim to design MiLAC architectures that maintain capacity-achieving performance (matching fully digital systems) while drastically reducing circuit complexity to a linear scaling ($O(N)$), making Gigantic MIMO practically feasible.

2. Methodology

The paper employs a graph-theoretical approach to model and optimize MiLAC architectures.

• Graph Theoretical Modeling:

  • The MiLAC is modeled as a graph $G = (V, E)$, where vertices ($V$) represent ports and edges ($E$) represent tunable admittance components.
  • A fully-connected MiLAC corresponds to a complete graph.
  • The authors propose restricting the graph topology to reduce edges (components) while preserving the mathematical ability to achieve channel capacity.

• Proposed Architecture: Stem-Connected MiLAC:

  • The authors introduce a specific family of architectures called stem-connected MiLACs.
  • Topology: Based on a "center graph" with a center size $Q = 2N_S - 1$ (where $N_S$ is the number of data streams).
  • Structure: A subset of ports (the "central vertices") are connected to all other ports. The remaining ports are connected only to these central vertices, not to each other.
  • Constraint: For the transmitter, the central vertices must include all input ports. For the receiver, they must include all output ports.

• Optimization Strategy:

  • The authors formulate a feasibility check problem to determine if a specific graph topology can satisfy the necessary conditions for capacity achievement (matching the singular vectors of the channel matrix).
  • They derive closed-form solutions for the susceptance matrices (the tunable components) of both transmitter and receiver stem-connected MiLACs.
  • The optimization relies on Singular Value Decomposition (SVD) of the channel matrix to calculate the required admittance values.

3. Key Contributions

1. Graph-Theoretical Model: Established a rigorous model linking MiLAC circuit topology to graph theory, enabling the systematic exploration of sparse interconnection architectures.
2. Stem-Connected Architecture: Proposed a novel family of MiLACs that reduces circuit complexity from quadratic to linear while theoretically guaranteeing capacity achievement.
3. Closed-Form Optimization: Derived explicit, closed-form mathematical solutions for configuring the tunable components of stem-connected MiLACs at both the transmitter and receiver to achieve channel capacity for any channel realization.
4. Complexity Analysis: Proved that the number of tunable components scales as $O(N_S N_T)$ (linear with antennas) for stem-connected MiLACs, compared to $O((N_S + N_T)^2)$ for fully-connected MiLACs.

4. Results

Theoretical analysis and Monte Carlo simulations (using Rayleigh fading channels) confirmed the following:

• Capacity Achievement: Stem-connected MiLACs achieve the exact same channel capacity as fully-connected MiLACs and fully digital MIMO systems. The achievable rate curves overlap perfectly with the theoretical capacity limit across various Signal-to-Noise Ratios (SNR) and numbers of streams ($N_S$).
• Complexity Reduction:
  • Fully-Connected: Complexity grows quadratically with the number of antennas.
  • Stem-Connected: Complexity grows linearly with the number of antennas.
  • Impact: For large antenna arrays (e.g., 256 antennas), the stem-connected architecture requires significantly fewer tunable components, leading to reduced hardware cost and lower power consumption (as power is tied to the number of control voltages for tunable components).

5. Significance

• Enabling Gigantic MIMO: This work bridges the gap between theoretical capacity and practical implementation for 6G. It demonstrates that one does not need a fully connected, prohibitively expensive analog network to achieve optimal performance.
• Scalability: By shifting the complexity scaling from quadratic to linear, the proposed architecture makes it feasible to deploy antenna arrays with thousands of elements without overwhelming hardware constraints.
• Hardware Efficiency: The reduction in tunable components directly translates to lower power consumption and simpler control circuitry, addressing two major hurdles in analog beamforming.
• Theoretical Foundation: The paper provides a mathematical proof that sparse connectivity in analog networks is sufficient for capacity achievement, challenging the assumption that full connectivity is required for optimal performance.

In conclusion, the paper presents a transformative approach to Gigantic MIMO by introducing stem-connected MiLACs, a low-complexity, capacity-achieving architecture that leverages graph theory to optimize analog signal processing for future 6G networks.