Repeater-Assisted Massive MIMO Full-Duplex Communications

Technical Summary: Repeater-Assisted Massive MIMO Full-Duplex Communications

Problem Statement
This paper addresses the performance limitations of massive multiple-input multiple-output (mMIMO) systems, which are often constrained by signal attenuation, line-of-sight (LoS) propagation issues, and shadowing. While distributed MIMO architectures (e.g., cell-free mMIMO) offer solutions, they typically require complex synchronization and heavy fronthaul infrastructure. To mitigate these issues, the authors investigate a system employing network-controlled, single-antenna repeaters that amplify and re-transmit signals. The specific challenge tackled is the optimization of repeater weights in a full-duplex (FD) setting where a base station (BS) simultaneously serves multiple uplink (UL) and downlink (DL) single-antenna user equipment (UE) over the same frequency band. The core difficulty lies in managing the additional interference and noise injection introduced by the repeaters while maximizing the weighted sum of the minimum spectral efficiencies (SEs) for both UL and DL users.

Methodology
The authors propose a system model where $L$ single-antenna repeaters assist an $M$-antenna FD mMIMO BS serving $K_{dl}$ DL and $K_{ul}$ UL users. The channel model incorporates direct links and compound indirect links via the repeaters, accounting for self-interference (SI) at the BS and inter-user interference. The BS employs zero-forcing (ZF) precoding for DL and combining for UL.

To optimize system performance, the authors formulate a non-convex optimization problem (P1) aimed at maximizing the weighted sum of the minimum UL and DL SEs, subject to Quality of Service (QoS) constraints and repeater power limits. The problem is recast (P2) using auxiliary variables to handle the min-max structure. Due to the non-convexity of the signal-to-interference-plus-noise ratio (SINR) constraints, the authors employ a Successive Convex Approximation (SCA) technique based on the Convex-Concave Procedure (CCP).

Key methodological steps include:

1. Feasibility Pursuit: To ensure the algorithm converges to a feasible solution from a random initialization, non-negative feasibility parameters ($\phi$) are introduced. These are iteratively driven to zero by adding regularization terms to the objective function.
2. Convexification: The non-convex SINR constraints are approximated using convex lower bounds derived from the SCA framework, transforming the problem into a convex optimization problem (P3).
3. Iterative Solution: The resulting convex problem is solved efficiently using CVX in an iterative manner.

Key Contributions

• Framework Development: The paper establishes a framework to quantify performance gains in multiuser FD mMIMO systems aided by a swarm of repeaters, a scenario not extensively covered in existing literature which often assumes half-duplex (HD) operation.
• Optimization Algorithm: An efficient iterative algorithm is proposed to optimize repeater amplification weights. This algorithm specifically targets the maximization of the weighted minimum SE while mitigating the detrimental effects of repeater-induced noise and interference.
• Comprehensive Analysis: The study analyzes both UL and DL performance under ZF processing, explicitly modeling the compound channels and the residual self-interference at the FD base station.

Numerical Results
The proposed system (RA-FD Optimized) was evaluated against three benchmarks:

1. RA-FD (Random): Same setup but with random repeater weights.
2. RA-HD (Optimized): A half-duplex counterpart with optimized repeater weights.
3. FD-mMIMO: A centralized FD system without repeater assistance.

Key findings include:

• Superiority of Optimization: Optimizing repeater weights is critical. The optimized RA-FD system significantly outperforms the RA-FD system with random weights, demonstrating that effective interference management is possible.
• FD Gains: FD operation yields substantial improvements over HD counterparts, with average per-UE SE improvements of 98% in DL and 252% in UL.
• Repeater Benefits: The inclusion of distributed repeaters in the FD mMIMO system provides further gains. Compared to the centralized FD-mMIMO baseline, the proposed system shows consistent improvements in both UL and DL scenarios.
• Magnitude of Improvement: The optimized FD design achieves spectral efficiency improvements of up to 4-fold and 2.5-fold compared to its half-duplex counterpart. In terms of fairness (minimum SE), the proposed design achieves more than a 10-fold improvement over conventional non-repeater-assisted mMIMO systems.
• Design Sensitivity: The results highlight that if the RA-FD system is not carefully designed (i.e., without weight optimization), performance may degrade below that of RA-HD systems due to amplified noise and interference.

Significance and Claims
The paper claims that the detrimental effects of injected noise and interference from repeater retransmissions can be effectively mitigated through the proposed weight optimization. By guaranteeing QoS levels for both UL and DL users, the study demonstrates that repeater-assisted FD mMIMO networks can achieve promising performance gains. Specifically, the authors conclude that in such systems, FD operation at the BS yields up to an 82% improvement in the system's minimum weighted sum SE compared to its HD counterpart. The work underscores the potential of combining swarm repeaters with FD mMIMO to enhance coverage and spectral efficiency in beyond 5G networks, provided that the amplification gains are rigorously optimized.