Abstract
This paper addresses the challenge of achieving network-layer stability in cooperative relay networks by bridging information-theoretic physical layer coding with network-layer queue management. We introduce a scalable "encoding-based" queue architecture and a Max-Weight style algorithm. This approach achieves throughput optimality without requiring prior knowledge of channel statistics or arrival rates, overcoming the "queue explosion" problem inherent in previous state-based architectures.
Introduction: The "Unconsummated Union"
In traditional networking, the physical layer is often treated as a simple "bit pipe"—a basic abstraction where links have fixed capacities. However, in cooperative relay networks, this abstraction fails. Relays in these systems do not merely pass bits; they employ strategies like Amplify-and-Forward (AF) or Compress-and-Forward (CF), transmitting complex functions of the signals they receive.
The primary challenge lies in the "unconsummated union" between information theory (coding strategies) and network theory (queue stability). Previous attempts to bridge these fields resulted in "state-based" queue architectures that suffered from "queue explosion"—where the number of virtual queues grew exponentially with the number of destinations. This paper seeks a low-complexity, scalable solution for managing these queues while maintaining theoretical optimality.
Proposed Architecture and Algorithm
The authors introduce two major innovations to solve the scalability problem:
- Encoding-based Queue Architecture: Inspired by modern adaptive modulation and coding (AMC), this architecture uses a finite set of "encoding schemes" (specific combinations of encoding functions and rate vectors). Relays maintain virtual queues based on these schemes and the first-hop fading state, reducing the queue count from exponential to a manageable linear scale.
- Back-pressure-based Algorithm: A centralized controller makes transmission decisions in every time block by calculating the "pressure"—a function of queue lengths and potential rates. The algorithm selects actions that maximize this pressure, effectively balancing the load across the source-to-relay and relay-to-destination hops.
Key Insights
- The Coding-Scheduling Link: In cooperative networks, the choice of a physical-layer codebook imposes inherent scheduling restrictions. Coding and transmission timing cannot be separated.
- Scalability via Practicality: By limiting the system to a finite set of encoding schemes—mirroring real-world hardware—the authors resolve the state explosion problem.
- Causal Knowledge Sufficiency: The algorithm is "blind" to underlying probability distributions of fading and arrival rates, requiring only the current channel state for optimal decision-making.
- Handling Real-Valued "Packets": Unlike traditional bit-based packets, cooperative schemes like AF use vectors of real-valued signal samples. This architecture is the first to handle such non-traditional packets in a throughput-optimal manner.
Results and Stability Analysis
The core contribution is Theorem 3, which proves that the proposed algorithm stochastically stabilizes all queues for any arrival rate vector within the network's capacity region. Using Foster-Lyapunov stability theory, the authors demonstrate a "strict negative drift," ensuring queues tend toward a stable state rather than growing indefinitely. Furthermore, the number of virtual queues at each relay is significantly reduced to $|M||F|^N$, where $M$ is the set of encoding schemes and $F$ is the fading state space.
Conclusion and Future Work
While the use of a finite set of encoding schemes introduces a small quantization loss compared to the absolute theoretical maximum, this loss can be minimized with a sufficiently fine-grained set of schemes. The current analysis focuses on a two-hop network (as seen in Figure 1) under a half-duplex constraint. Future research may extend these principles to full-duplex cooperative relaying and more complex topologies involving direct source-to-destination links.
Summary updated