Closed-Form Delay-Optimal Power Control for Energy Harvesting Wireless System With Finite Energy Storage

Zhang Fan (Michael), Fan Zhang, Vincent K. N. LauIEEE TSP 2014 · DOI

TL;DR

This paper tackles the challenge of optimizing power control in wireless systems that harvest energy from renewable sources, aiming to minimize data transmission delays while managing finite battery storage. They introduce a novel closed-form, adaptive power control strategy that balances data urgency, energy availability, and channel conditions, outperforming existing methods.

What Problem Does This Paper Address?

Modern wireless systems are increasingly powered by renewable energy sources like solar and wind. However, these sources are intermittent, and the energy storage (batteries) is finite. This creates a complex problem: how to efficiently use harvested energy to transmit data in real-time applications, especially when data arrives in bursts and the system needs to minimize delays. Existing approaches often rely on assumptions like infinite battery capacity, known future energy arrivals (non-causal), or simplified models that don't fully capture the interplay between data and energy queues. This paper addresses the need for an online, delay-optimal power control strategy that works with finite energy storage and bursty arrivals, adapting dynamically to the system's state.

Methods

The authors formulate the problem as a Markov Decision Process (MDP) to find a delay-optimal power control policy. To overcome the "curse of dimensionality" and derive a practical solution, they introduce a Virtual Continuous Time System (VCTS). This allows them to leverage calculus and differential equations to derive closed-form approximate priority functions. These functions essentially tell the system how to prioritize data transmission based on the current data queue length (DQSI), energy queue length (EQSI), and channel state information (CSI). The core of their approach involves solving a Hamilton-Jacobi-Bellman (HJB) equation for the VCTS, which then provides a closed-form expression for the priority function. This priority function dictates a multi-level water-filling structure for the power control action, where the "water level" is adaptively determined by the DQSI and EQSI.

Datasets

Not discussed in the paper. The evaluation is based on simulations with configured system parameters.

Key Insights and Ideas

  • The Interplay is Crucial: The paper highlights that effectively managing energy harvesting systems requires a deep understanding of how data queue dynamics and energy queue dynamics are coupled. The service rate depends on available energy, and energy availability dictates how much data can be sent.
  • Virtual Continuous Time System (VCTS) as a Powerful Abstraction: By transforming the discrete-time MDP into a continuous-time system, the authors unlock powerful analytical tools (calculus, differential equations) to derive closed-form solutions, which are often intractable in discrete settings.
  • Closed-Form Priority Functions for Insight: The derived priority functions offer valuable insights into how system parameters (data arrival rate, energy arrival rate, battery capacity) influence the optimal power control strategy across different operating regimes.
  • Multi-Level Water-Filling Structure: The resulting power control policy isn't a simple threshold; it's a sophisticated multi-level water-filling mechanism that dynamically adjusts based on the system's state, offering a more nuanced approach than simpler greedy strategies.
  • Asymptotic Regimes Reveal Design Principles: Analyzing the system in different asymptotic regimes (e.g., large data arrival, energy-sufficient) reveals distinct optimal control behaviors, providing guidelines for system design and parameter tuning.

Key Results

The paper proposes a closed-form delay-optimal power control policy that is adaptive to CSI, DQSI, and EQSI. Simulations demonstrate that this proposed scheme achieves significant performance gains in terms of average delay compared to various baselines, including greedy strategies and CSI-only or queue-weighted water-filling methods. The performance of the proposed scheme is shown to be very close to that of an optimal value iteration algorithm (VIA), which is computationally intensive. The computational complexity of the proposed scheme is comparable to simpler baselines, making it practical for real-world deployment.

Limitations and Future Work

  • Computational Complexity: While the proposed solution is closed-form and computationally efficient compared to optimal algorithms like VIA, solving the HJB equation itself can be complex. The paper addresses this by deriving closed-form solutions for specific asymptotic regimes.
  • Assumptions: The model relies on certain assumptions, such as i.i.d. data arrivals and block i.i.d. energy arrivals. Real-world scenarios might involve more complex arrival processes.
  • Future Work: The authors suggest exploring more complex scenarios, such as multi-user systems, interference channels, and more sophisticated energy arrival models. They also highlight the importance of further investigating the impact of finite energy storage capacity on system performance.

Summary updated

BibTeX
@article{michael2014closed,
  author = {ZHANG Fan (Michael) and Fan Zhang and Vincent K. N. Lau},
  title = {{Closed-Form Delay-Optimal Power Control for Energy Harvesting Wireless System With Finite Energy Storage}},
  journal = {IEEE TSP},
  year = {2014},
  doi = {10.1109/tsp.2014.2355777},
}
Figure