Evaluating Task Execution Performance Under Energy Measurement Overhead DOI: XXXXXXX.XXXXXXX Conference: 2nd Workshop on Energy Neutral and Sustainable IoT Devices and Infrastructure; Sep. 22, 2025; KUL, Leuven ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computing methodologies Simulation evaluation CCS: H
Mateen Ashraf, Shahab Jahanbazi, Onel L. A. López · 2025
Read Evaluating Task Execution Performance Under Energy Measurement Overhead DOI: XXXXXXX.XXXXXXX Conference: 2nd Workshop on Energy Neutral and Sustainable IoT Devices and Infrastructure; Sep. 22, 2025; KUL, Leuven ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computing methodologies Simulation evaluation CCS: H by Mateen Ashraf, Shahab Jahanbazi, Onel L. A. López (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-08757.txt.
Introduction
Mateen Ashraf, Shahab Jahanbazi, and Onel Lopez weigh the hidden cost of energy checks on harvest fed web things, in a paper from 2025. Such things live on light or shake power and can bend their work pace to the charge on hand. That bend needs charge reads, and reads cost charge too, a cost most past work waves off.
The team sets blind and aware work plans side by side and counts tasks done. For set gear traits, a best pair of read pace and work pace tops the count for both plans. Pick those paces poorly and the aware plan can lose to the blind one, since its reads eat its edge. The paper gives a tune guide for field builds.
Catalog record
| CREATOR | Mateen Ashraf, Shahab Jahanbazi, Onel L. A. López |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-08757
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Evaluating Task Execution Performance Under Energy Measurement Overhead DOI: XXXXXXX.XXXXXXX Conference: 2nd Workshop on Energy Neutral and Sustainable IoT Devices and Infrastructure; Sep. 22, 2025; KUL, Leuven ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computing methodologies Simulation evaluation CCS: Hardware Reusable energy storage Mateen Ashraf, Shahab Jahanbazi, Onel L. A. López Abstract . Energy-awareness for adapting task execution behavior can bring several benefits in terms of performance improvement in energy harvesting (EH) Internet of Things (IoT) devices. However, the energy measurement cost of acquiring energy information, which is traditionally ignored, can potentially neutralize or even reverse the potential benefits. This paper highlights operational parameters, such as energy measurement frequency and task execution frequency, which can be tuned to improve the task execution performance of an EH-IoT device. To this end, we consider energy-blind (EB) and energy-aware (EA) task decision approaches and compare their task completion rate performance. We show that, for specific hardware design parameters of an EH-IoT device, there exists an optimal energy measurement/task execution frequency that can maximize the task completion rate in both approaches. Moreover, if these parameters are not chosen appropriately, then energy measurement costs can cause EA scheduling to underperform compared to EB scheduling. 1. Introduction (a) EB scheduling. (b) EA scheduling. Figure 1. EH-IoT device with EB and EA scheduling. (a) EB scheduling: Task execution is attempted every F time slots, with F=6 used as example, regardless of the amount of available energy in the energy storage. (b) EA scheduling: Task execution decisions are guided by the periodic energy measurements every Q time slots, with Q=3 as an example. In energy harvesting (EH) Internet of Things (IoT) devices, efficient energy management is critical for reliable operation under highly random and often scarce energy conditions ( DMa ; UMESH ; onel2 ) . One widely considered approach is to measure or estimate the available energy prior to executing tasks, allowing the device to adapt its task execution decisions to the available energy budget, and thereby reducing the probability of task failure. These schemes, which base task execution decisions on available energy, EH statistics, energy consumption patterns and/or their forecasts, are called energy-aware (EA) schemes. However, it is noted that the measurement process in EA schemes incurs a non-negligible energy overhead ( GLius ; Ylee ) , which can even negate its intended benefits, especially in ultra-low-power devices such as EH-IoT devices. Energy measurement in EH-IoT devices can generally be classified into two main approaches ( 9406578 ) : hardware-based methods and software-based (or model-based) estimation. The former methods involve the direct monitoring of energy levels through sensors and specialized circuitry ( BANOTRA ) , such as energy monitoring integrated circuits designed especially for low-power applications. These methods typically offer high accuracy but come at the cost of increased system complexity and additional energy consumption. In contrast, the latter approaches rely on mathematical models and observed system behavior to estimate energy usage. Techniques like energy profiling ( SABOVIC ) , which leverage pre-measured energy consumption patterns of tasks to estimate runtime usage, have less overhead but generally provide lower accuracy. Moreover, the energy profiling methods require higher computational capabilities at the device side to extract information needed for building energy profiles. This can be a serious hurdle in implementation of these methods in EH-IoT devices. In addition to energy consumption estimation, profiling energy arrival is a crucial factor in enabling an energy-efficient task decision approach for EH-IoT devices. A significant challenge in this context is that the EH profile is typically unknown and highly variable due to environmental fluctuations. To address this, two primary approaches have been developed: prediction-based models ( prediction ; onel ) and stochastic or modeling-based approaches ( stochastic_approach ) . In prediction-based methods, historical EH data are used to train models that can forecast future energy availability. These techniques often employ time series analysis or machine learning algorithms to anticipate short- or long-term trends. In contrast, modeling-based approaches treat the energy arrival process as a stochastic phenomenon, using probabilistic models (such as Markov processes) to describe and reason about the uncertainty in energy supply. These models are particularly useful for theoretical analysis and for designing robust EA algorithms under uncertainty. Across both approaches, several common challenges persist, including ensuring measurement accuracy, managing the energy overhead introduced by the measurement process itself, and addressing the latency. Despite its potential benefits in modern EH-IoT devices, the effects of energy measurement on the overall device performance remains underexplored. To consider the trade-offs between energy-awareness and system efficiency, various studies proposed adaptive measurement techniques. For instance, the authors in ( dutycycle ) used an approach named duty-cycled sensing to reduce measurement overhead while maintaining acceptable accuracy. Therefore, understanding and quantifying the trade-offs introduced by energy measurement is essential for designing robust, self-sustaining IoT systems that maximize task completion rate while minimizing unnecessary energy expenditure. We take a step in this direction in the current work. This paper highlights the critical system parameters which influence the task completion performance of the EH-IoT device. Specifically, we consider EB and EA approaches for making decisions about task execution. Herein, the energy-blind (EB) approach adopts a periodic task execution without any energy measurement mechanism, and therefore it does not entail any energy measurement overheads. On the other hand, the EA approach uses available energy estimates for making decision regarding task execution. The energy estimates are based on periodically carried out energy measurements, entailing an energy measurement overhead in the form of measurement energy consumption, it is shown that for a given set of unalterable hardware parameters such as energy storage capacity, task buffer size, and energy measurement circuitry, there exists an optimal task execution frequency for the EB approach, and an optimal value for the energy measurement frequency in the EA approach. The rest of the paper is organized as follows. The considered system model is discussed in Section 2. Section 3 discusses EB and EA scheduling. The simulation results are presented in Section 4. Finally, the conclusions are provided in Section 5. 2. System Model We consider an EH-IoT device that needs to execute a certain task repeatedly for a long period of time. This can represent an IoT sensor that needs to repeatedly pre-process the reading of an environmental parameter over an indefinite period of time. 1 1 1 It can be any task such as sensing, computing communicating, and actuating (or a combination thereof). We assume that the total time is divided into equal time slots, where the duration \Delta t of the individual time slots is at least equal to the time needed for executing the task successfully conditioned on the availability of sufficient energy. This is actually dictated by the operational speed of the hardware equipment, e.g., central processing unit speed when executing computational tasks. The current work does not take task deadline aspects into consideration, while addressing this aspect could serve as a promising direction for future research. A pictorial representation of the system model is shown in Fig. 1 . The total number of time slots is assumed to be T_{max} . For ease of readability, specific details related to the modeling of tasks, energy, and the performance metric are provided in the following subsections. 2.1. Task