Age-Minimal Online Policies for Energy Harvesting Sensors with Incremental Battery Recharges

Ahmed Arafa, Jing Yang, Sennur Ulukus, H. Vincent Poor

Introduction

In the burgeoning Internet of Things (IoT) landscape, sensors are increasingly deployed to monitor dynamic physical processes such as traffic flow or environmental temperature. For these applications, the value of data is inextricably linked to its "freshness." This research, conducted by a team from Princeton, Penn State, and the University of Maryland, addresses the challenge of maintaining data freshness in sensors powered by unpredictable, harvested energy.

The Age of Information Challenge

The primary metric used to quantify data freshness is the Age of Information (AoI). AoI is defined as the time elapsed since the most recently received update was generated at the sensor. As shown in Figure 1, the age increases linearly over time and drops abruptly whenever a new update is received.

Evolution of the Age of Information (AoI) over time.

Managing AoI in energy-harvesting sensors is difficult due to two constraints:

  1. Energy Causality: An update can only be transmitted if the sensor has sufficient stored energy.
  2. Finite Storage: Batteries have limited capacity; if a battery is full, any newly harvested energy is lost to overflow.

While previous models often assumed "full recharges" (instantaneous jumps from empty to full), this paper focuses on the more realistic incremental recharge model, where energy arrives in discrete units over time.

Methods: Renewal Policies and Incremental Recharges

The researchers employed renewal theory and Lagrangian optimization to identify the optimal online policy. They proved that the optimal strategy for minimizing average AoI is a "renewal policy," where the decision to update depends solely on the current battery level and the current age, rather than the entire history of the system.

Incremental energy arrival patterns within a renewal interval.

For a battery capacity of $B=2$ units, the authors derived exact energy-dependent thresholds. By analyzing the expected area under the AoI curve relative to the expected time between renewal events (as illustrated in Figure 4), they formulated an optimization problem to find the precise moment to "spend" energy on an update.

Key Insights and Results

A central finding of the study is that a sensor's "eagerness" to update should vary with its battery state. When the battery is nearly full, the sensor should be more aggressive (updating at a lower age threshold) to prevent energy overflow. Conversely, when energy is scarce, it should wait longer to ensure it has enough power for future critical updates.

Key results include:

  • Optimal Thresholds: For a 2-unit battery, the optimal age threshold was found to be $\lambda^* \approx 0.72$, compared to $\approx 0.90$ for a 1-unit battery. This indicates that larger batteries allow for more frequent, "fresher" updates.
  • Incremental Advantage: The study revealed that frequent, small energy arrivals (incremental) are more beneficial for AoI than infrequent, large arrivals (bulk), even if the average power is identical. This is because incremental arrivals provide more opportunities to reset the age clock.
  • Performance: The proposed threshold-based policy significantly outperforms simple "best-effort" strategies, approaching the theoretical lower bound of $0.5$ as battery size increases.

Conclusion and Future Work

This work provides a rigorous mathematical framework for optimizing data freshness in energy-constrained IoT devices. While the $B=2$ case was solved explicitly, the authors note that larger battery sizes introduce significant combinatorial complexity. Future research may focus on developing near-optimal, computationally efficient heuristics for larger systems and extending these principles to multi-hop sensor networks.

Summary updated

BibTeX
@inproceedings{arafa2018age,
  author = {Ahmed Arafa and Jing Yang and Sennur Ulukus and H. Vincent Poor},
  title = {{Age-Minimal Online Policies for Energy Harvesting Sensors with Incremental Battery Recharges}},
  booktitle = {ITA},
  year = {2018},
  doi = {10.1109/ita.2018.8503180},
}

Loading related content...

Figure