On the Capacity and Diversity-Multiplexing Tradeoff of the Two-Way Relay Channel

Rahul Vaze, Robert W. Heath Jr.

TL;DR

This paper tackles the challenge of maximizing data exchange efficiency in two-way wireless communication systems that use relays. The authors propose novel strategies for "amplify-and-forward" relays to boost data rates and achieve optimal "diversity-multiplexing" trade-offs, especially in scenarios with multiple relays. They also establish how the system's capacity scales with the number of relays, showing significant improvements over traditional one-way communication.

What Problem Does This Paper Address?

In many wireless scenarios, like cellular networks or ad-hoc communication, two devices (let's call them T1 and T2) need to exchange data, but their direct link might be weak or non-existent. Relays are used to facilitate this communication. This paper focuses on a "two-way relay channel," where both T1 and T2 are simultaneously sending and receiving data, with the help of one or more relay nodes. The complexity arises from managing multiple antennas at the terminals and relays, optimizing relay behavior, and understanding the fundamental limits of this communication, known as the "capacity region." Specifically, the paper addresses two key problems:

  1. Maximizing Achievable Rate with Amplify-and-Forward (AF) Relays: AF relays amplify and retransmit the received signal. Finding the optimal way for these relays to amplify (beamforming) to maximize the overall data rate between T1 and T2 is challenging, especially with multiple relays and antennas.
  2. Achieving Optimal Diversity-Multiplexing (DM) Trade-off: This trade-off balances reliability (diversity) against data rate (multiplexing). The goal is to find strategies that achieve the best possible performance in this regard for two-way communication, particularly when direct links exist between T1 and T2.

Methods

The paper employs several sophisticated techniques:

  • Iterative Power Minimization for AF Beamforming: To find the optimal AF relay strategies, the authors formulate a power minimization problem subject to signal-to-interference-and-noise ratio (SINR) constraints. This problem is solved iteratively, with each step involving a non-convex optimization that is nevertheless guaranteed to find the optimal solution via Karush-Kuhn-Tucker (KKT) conditions.
  • Dual Channel Matching Strategy: To overcome the complexity of global channel state information (CSI) required by the optimal AF strategy, a simpler "dual channel matching" strategy is proposed. This method uses only local CSI at each relay and provides a good lower bound on the achievable rate region.
  • Cut-Set Bound for Capacity Upper Bound: To understand the fundamental limits, the authors derive an upper bound on the capacity region using the "cut-set bound" from information theory. This involves analyzing different "cuts" of the network to determine the maximum possible information flow.
  • Compress-and-Forward (CF) Strategy for DM-Tradeoff: For the diversity-multiplexing trade-off, a modified "compress-and-forward" strategy is proposed. CF relays compress the received signal before forwarding it. This strategy is shown to achieve the optimal DM-tradeoff for full-duplex systems and under certain conditions for half-duplex systems.
  • Asymptotic Analysis (K → ∞): The paper analyzes the system's behavior as the number of relays (K) becomes very large. This allows for deriving capacity scaling laws and understanding how the system performs in dense relay networks.

Datasets

Not discussed in the paper. The paper focuses on theoretical analysis and derivation of performance bounds and strategies, rather than empirical evaluation on specific datasets.

Key Insights and Ideas

  • Two-way communication offers significant advantages: The paper demonstrates that two-way relay channels can achieve higher capacities and better performance than unidirectional communication, effectively removing the rate loss typically associated with half-duplex operation.
  • Simpler strategies can be remarkably effective: The "dual channel matching" strategy, requiring only local CSI, provides a rate region very close to the optimal AF strategy, making it a practical choice for implementation.
  • Capacity scales logarithmically with the number of relays: In the asymptotic regime of many relays, the capacity of the two-way relay channel grows logarithmically with the number of relays (log K), indicating that adding more relays significantly boosts performance.
  • CF strategy is powerful for DM-tradeoff: The proposed CF strategy effectively decouples the two-way relay channel into two independent one-way channels, achieving optimal diversity-multiplexing performance in full-duplex scenarios.

Key Results

  • Optimal AF Beamforming Algorithm: An iterative algorithm is developed to find optimal relay beamformers for AF relays, maximizing the achievable rate region.
  • Capacity Scaling Law: The paper establishes that the capacity of a two-way relay channel with K relays scales as (M/2) log K + O(1), where M is the number of antennas at the terminals. This shows a significant improvement over unidirectional communication.
  • DM-Tradeoff Achieved by CF: For full-duplex systems, the CF strategy achieves the optimal DM-tradeoff. For half-duplex systems, it achieves optimality under specific conditions related to the communication protocol.
  • Close Gap Between Bounds: In asymptotic regimes, the gap between the derived lower and upper bounds on capacity is shown to be a constant O(1) term, indicating that the proposed strategies are near-optimal. Simulations also show a small gap for finite numbers of relays.

Limitations and Future Work

  • Optimal AF for Multi-Antenna Relays: The derived optimal AF strategy is primarily for single-antenna terminals and single-antenna relays. Extending this to multi-antenna relays (N > 1) is mentioned as a straightforward generalization, but the paper focuses on the single-antenna case for simplicity.
  • Global CSI Requirement for Optimal AF: The optimal AF strategy requires each relay to have global CSI (knowledge of all channels), which can be a significant practical limitation. The dual channel matching strategy addresses this by using only local CSI.
  • Closed-Form Expression for Optimal AF Rate Region: The optimal AF strategy does not yield a closed-form expression for the achievable rate region, making direct analytical comparison difficult. The dual channel matching strategy, however, does provide a closed-form expression.
  • Half-Duplex DM-Tradeoff Optimality: While the CF strategy achieves the optimal DM-tradeoff for full-duplex systems, its optimality for half-duplex systems is shown only under specific conditions. Proving general optimality for half-duplex cases remains an open challenge due to protocol dependency.
  • Future Work: The authors suggest further research into extending the optimal AF strategy to multi-antenna relays and proving the general optimality of the CF strategy for half-duplex two-way relay channels.

Summary updated

Abstract

In a two-way relay channel, two sources use one or more relay nodes to exchange data with each other. This paper considers a multiple input multiple output (MIMO) two-way relay channel, where each relay node has one or more antennas. Optimal relay transmission strategies for the two-way relay channel are derived to maximize the achievable rate with amplify and forward (AF) at each relay and to achieve the optimal diversity-multiplexing tradeoff (DM-tradeoff). To maximize the achievable rate with AF, an iterative algorithm is proposed which solves a power minimization problem subject to minimum signal-to-interference-and-noise ratio constraints at every step. The power minimization problem is nonconvex. The Karush Kuhn Tucker conditions, however, are shown to be sufficient for optimality. Capacity scaling law of the two-way relay channel with increasing number of relays is also established by deriving a lower and upper bound on the capacity region of the two-way relay channel. To achieve the optimal DM-tradeoff, a compress and forward strategy is proposed and its DM-tradeoff is derived. For the full-duplex two-way relay channel, the proposed strategy achieves the optimal DM-tradeoff, while for the half-duplex case the proposed strategy is shown to achieve the optimal DM-tradeoff under some conditions.
BibTeX
@article{vaze2011capacity,
  author = {Rahul Vaze and Robert W. Heath Jr.},
  title = {{On the Capacity and Diversity-Multiplexing Tradeoff of the Two-Way Relay Channel}},
  journal = {IEEE Trans. Information Theory},
  year = {2011},
  url = {https://openreview.net/forum?id=234WbDufiM},
}
Figure