Throughput Maximization for the Gaussian Relay Channel with Energy Harvesting Constraints

Chuan Huang, Rui Zhang, Shuguang Cui

Introduction

In modern wireless communications, the shift from fixed-battery nodes to energy harvesting (EH) nodes presents a paradigm shift in resource management. Unlike traditional systems where power is a static constraint, EH nodes must operate under time-varying and intermittent energy arrivals. This paper investigates the maximization of data throughput in a three-node Gaussian relay channel where both the source and the relay are powered by EH. The core contribution is the identification of "energy diversity"—the ability to exploit independent energy arrival patterns at different nodes to enhance network performance.

System Model and Problem Formulation

The study considers a half-duplex, orthogonal Gaussian relay channel employing a Decode-and-Forward (DF) protocol. The network consists of a source, a relay, and a destination. Both the source and the relay are equipped with energy harvesting capabilities, modeled as buffers that store harvested energy for future use.

The three-node Gaussian relay channel model with energy harvesting source and relay nodes.

The objective is to maximize the total throughput over $N$ transmission blocks. The authors analyze two distinct scenarios based on decoding requirements:

  1. Delay-Constrained (DC): The destination decodes messages with a one-block delay, requiring immediate forwarding.
  2. No-Delay-Constrained (NDC): The destination can wait until the end of all $N$ blocks to decode, allowing the relay to buffer data and align transmissions with energy availability.

Optimization Methodology

The authors assume a deterministic EH model where energy arrival times and amounts are known a priori. For the DC case, the problem is a convex optimization solved via Karush-Kuhn-Tucker (KKT) conditions and a specialized "forward search" algorithm.

For the NDC case, the problem is initially non-convex. However, the authors prove a Separation Principle, demonstrating that source power allocation can be optimized independently of the relay. This allows for a two-stage convex optimization approach, significantly simplifying the search for an optimal solution.

Key Insights: Energy Diversity and Power Monotonicity

A central finding is the concept of energy diversity. In a relay channel, the source and relay typically experience independent energy harvesting processes. Even in a time-invariant wireless channel, these independent "energy paths" create opportunities for optimization. If the source has an energy surplus while the relay is constrained, the system can adapt its scheduling to maintain high throughput.

Optimal power allocation profiles compared against energy harvesting arrival constraints.

As shown in Figure 2, the optimal power allocation follows a non-decreasing staircase profile. Because the capacity function is concave, the most efficient strategy is to keep power levels as constant as possible. When energy arrivals force a change, the power should only increase to avoid depleting energy needed for future blocks, ensuring that no energy is "wasted" by being held too long if it could have been used earlier to smooth the transmission rate.

Results and Conclusion

Numerical simulations using periodic energy arrival models validate that the proposed optimal algorithms significantly outperform "greedy" strategies. The NDC scheme consistently achieves higher throughput than the DC scheme by better exploiting energy diversity through data buffering.

While this work assumes deterministic arrivals and infinite battery capacity, it establishes a critical theoretical upper bound. Future research will need to address stochastic energy arrivals and finite battery constraints to bring these insights into practical, real-world EH relay networks.

Summary updated

BibTeX
@article{huang2013throughput,
  author = {Chuan Huang and Rui Zhang and Shuguang Cui},
  title = {{Throughput Maximization for the Gaussian Relay Channel with Energy Harvesting Constraints}},
  journal = {IEEE J. Sel. Areas Commun.},
  year = {2013},
  url = {https://openreview.net/forum?id=mJWvVuSukDHV},
}
Figure