Enhancing User Fairness in Two-Layer RSMA: A Movable Antenna Approach

This paper proposes a novel movable antenna approach for two-layer rate-splitting multiple access (RSMA) systems that employs a two-loop iterative algorithm to jointly optimize beamforming, user clustering, common rate allocation, and antenna positions, thereby significantly enhancing user fairness compared to existing benchmark schemes.

The Gist

Imagine a busy coffee shop (the Base Station) trying to serve a large group of customers (the Users) who all want different drinks (Data Streams) at the same time.

In the old days, the barista had to stand in one spot. If a customer was far away or blocked by a pillar, they got a bad signal, and the whole group's experience suffered because the "weakest link" determined how fast everyone could be served. This is the problem of User Fairness.

This paper proposes a clever new solution using Movable Antennas and a smart serving strategy called Two-Layer RSMA. Here is the breakdown in simple terms:

1. The Problem: The "Weakest Link" Bottleneck

In traditional systems (like standard Wi-Fi or older cellular tech), the barista (Base Station) sends a "Common Message" to everyone. But this message can only be as fast as the customer with the worst connection. If one person is far away, everyone slows down to wait for them. This is unfair to the people sitting right next to the barista.

2. The New Strategy: Two-Layer Serving (RSMA)

The authors propose a "Two-Layer" approach, like a tiered menu:

• Layer 1 (The Group Drink): A message sent to a whole table (a cluster of users).
• Layer 2 (The Personal Drink): A private message sent just to one specific person.

By splitting the message this way, the system doesn't have to wait for the worst connection to serve the whole group. It can serve the "Group Drink" to a small, well-connected group first, then serve the "Personal Drinks" to individuals. This is much more efficient.

3. The Secret Weapon: Movable Antennas (The "Dancing Barista")

Here is the game-changer. Instead of the barista standing still, imagine the barista has robotic arms that can physically slide left and right along the counter.

• Fixed Antennas: The barista is glued to one spot. If a customer is in a blind spot, they get a bad signal.
• Movable Antennas (MA): The barista can slide their position to find the perfect angle to serve everyone. They can move closer to a distant customer or shift slightly to avoid a pillar. This creates "better paths" for the signals.

4. The Challenge: Too Many Choices

The problem is that the barista has to decide:

1. Where to stand? (Antenna Position)
2. Who sits at which table? (User Clustering)
3. How much coffee to pour for the group vs. the individual? (Rate Allocation)
4. Which direction to aim the coffee pot? (Beamforming)

There are millions of combinations. Trying every single one would take forever.

5. The Solution: The "Smart Search" Algorithm

The authors created a two-step "Smart Search" algorithm to solve this puzzle:

• The Outer Loop (The Scout): This part uses a method called DNPPSO (Dynamic Neighborhood Pruning Particle Swarm Optimization). Imagine a swarm of birds searching for the best spot to land. They fly around, testing different positions for the barista.

  • The "Pruning" Trick: If a group of birds is all flying in the same direction and getting the same result, the algorithm tells them, "Stop wasting energy; we know this area isn't the best." It cuts out the redundant searchers to save time.

• The Inner Loop (The Manager): Once the scout finds a good spot for the barista, the manager steps in.

  • Grouping: It looks at who has similar tastes (channel similarity) and puts them at the same table.
  • Serving: It calculates exactly how much "Group Drink" and "Personal Drink" to give to maximize the speed for the slowest person in the room.

6. The Result: Everyone Gets a Fair Cup

The simulations show that this new system is a huge winner:

• Fairness: The person at the back of the room gets a much faster connection than before.
• Efficiency: The system handles more users without slowing down.
• Cost: You need fewer antennas to get the same performance, saving money on hardware.

Summary Analogy

Think of it like a concert.

• Old Way: The singer stands in one spot. The people in the back can't hear well, so the singer has to shout very slowly so everyone can understand.
• New Way: The singer has a sliding stage (Movable Antenna). They can slide closer to the back row when needed. They also use a two-tier system: they sing a chorus to the whole crowd (Layer 1), but then whisper a special verse to the front row while the back row listens to the chorus (Layer 2).
• The Algorithm: It's the stage manager who figures out exactly where to slide the stage and how to group the audience so that everyone hears the music perfectly, even the person in the very back.

In short: By letting the antennas move and using a smarter way to group users, this paper ensures that no user is left behind, making the whole network fairer and faster.

Technical Summary

Here is a detailed technical summary of the paper "Enhancing User Fairness in Two-Layer RSMA: A Movable Antenna Approach."

1. Problem Statement

The paper addresses the critical challenge of user fairness in advanced multi-user communication systems, specifically within the context of Two-Layer Rate-Splitting Multiple Access (2L-RSMA) for 6G networks.

• The Bottleneck: While 2L-RSMA improves upon conventional one-layer RSMA by introducing a hierarchical structure (inter-cluster and intra-cluster common streams) to manage interference, its performance is still fundamentally limited by static wireless channels. In traditional systems, the common rate is constrained by the user with the weakest channel, creating a fairness bottleneck.
• The Gap: Existing research on Movable Antennas (MAs) has primarily focused on sum-rate maximization or one-layer RSMA. The potential of MAs to enhance user fairness (specifically the minimum user rate) in a two-layer RSMA framework remains unexplored.
• Objective: The authors aim to maximize the minimum user rate (max-min fairness) by jointly optimizing:
  1. Beamforming matrices.
  2. Common rate allocation.
  3. User clustering.
  4. Antenna Position Vector (APV) (leveraging MAs).

2. Methodology

The authors propose a novel system model and a two-loop iterative algorithm to solve the resulting non-convex, non-smooth optimization problem.

A. System Model

• Architecture: A downlink system where a Base Station (BS) equipped with $N_T$ MAs serves $K$ single-antenna users.
• Signal Structure: The transmitted signal includes:
  • One inter-cluster common stream ($s_{c2}$) decoded by all users.
  • $Q$ intra-cluster common streams ($s_{c1,q}$), each decoded only by users in cluster $q$.
  • $K$ private streams ($s_{p,k}$) decoded only by their respective users.
• Channel Model: A multi-path channel model where the channel vector depends on the MA positions ($t$). The Field Response Vector (FRV) changes dynamically as antennas move, providing additional Degrees of Freedom (DoFs).
• Constraints:
  • MA Constraints: Antennas must stay within a linear region $[T_{min}, T_{max}]$ and maintain a minimum spacing $D_0$ to avoid coupling.
  • Power Constraint: Total transmit power is limited.
  • SIC Order: Users decode streams in a specific order (Inter-cluster common $\to$ Intra-cluster common $\to$ Private) using Successive Interference Cancellation (SIC).

B. Optimization Algorithm

The problem is decomposed into a Two-Loop Iterative Algorithm:

1. Outer-Loop (Antenna Position Optimization):

   • Goal: Optimize the Antenna Position Vector (APV) $t$.
   • Method: Uses Dynamic Neighborhood Pruning Particle Swarm Optimization (DNPPSO).
   • Mechanism:
     • Particles represent candidate APVs.
     • A fitness function evaluates the minimum user rate obtained from the inner loop, penalizing constraint violations.
     • Pruning: To reduce computational complexity, particles converging near the global best position are identified via a dynamic neighborhood radius and "pruned" (skipped) in subsequent iterations, as they offer diminishing returns.

2. Inner-Loop (Resource Allocation & Clustering):

   • Goal: For a fixed APV, optimize user clustering, beamforming, and rate allocation.
   • Step 1: User Clustering: A greedy algorithm based on channel similarity (cosine similarity of channel vectors) is used. Users with highly correlated channels are grouped together to maximize intra-cluster common rates.
   • Step 2: Beamforming & Rate Allocation:
     • Formulated as a non-convex problem.
     • Semidefinite Relaxation (SDR): Converts beamforming vectors into matrix variables.
     • Successive Convex Approximation (SCA): Handles the non-convex rate constraints (Difference of Convex structure) by replacing convex terms with their first-order Taylor lower bounds.
     • The rank-one constraint is relaxed, with theoretical justification that the optimal solution naturally satisfies it.

3. Key Contributions

1. Novel Integration: First to integrate Movable Antennas into a Two-Layer RSMA framework specifically targeting user fairness (max-min rate), addressing the common rate bottleneck of conventional systems.
2. Joint Optimization Formulation: Formulated a comprehensive optimization problem jointly designing beamforming, rate allocation, user clustering, and antenna positions.
3. Efficient Algorithm: Developed a two-loop iterative algorithm combining DNPPSO for spatial optimization and SCA/SDR for resource allocation. The inclusion of a dynamic pruning mechanism significantly reduces computational complexity without sacrificing performance.
4. Channel-Aware Clustering: Proposed a specific clustering strategy based on channel similarity to maximize the efficiency of intra-cluster common streams.

4. Simulation Results

The proposed scheme was evaluated against eight benchmark schemes (including FPA-based RSMA, NOMA, SDMA, and various MA-assisted baselines) across different metrics:

• Number of Users: The proposed method maintains competitive max-min rates as user count increases, outperforming benchmarks.
• Transmit Power: The proposed method significantly outperforms FPA-based and one-layer RSMA schemes. While "Classic PSO" (without pruning) shows slightly higher rates, the proposed DNPPSO achieves near-optimal performance with significantly lower complexity.
• Number of Antennas: The scheme achieves target rates with fewer antennas compared to benchmarks, highlighting potential hardware cost reductions.
• User Distance: The performance gap between the proposed Two-Layer RSMA and One-Layer RSMA widens as user distance increases, proving the effectiveness of the hierarchical interference management in handling diverse channel conditions.

5. Significance

• Fairness Enhancement: The work demonstrates that dynamically adjusting antenna positions can effectively reshape the wireless channel to improve the worst-case user rate, a critical metric for equitable 6G service.
• Hardware Efficiency: By achieving high fairness with fewer antennas or lower power, the approach offers a pathway to more cost-effective and energy-efficient 6G base station designs.
• Algorithmic Innovation: The combination of DNPPSO with SCA/SDR provides a robust template for solving complex, non-convex joint optimization problems in future reconfigurable intelligent surface (RIS) and MA-assisted networks.
• Theoretical Benchmark: The study establishes a theoretical performance benchmark for MA-assisted 2L-RSMA, assuming perfect CSI, while acknowledging practical constraints (movement time, energy) as future work.